A Systematic Review of Artificial Intelligence in Piano Education: Efficacy,Ethics, and the Transformation of the Teacher's Role (2020–2025)

A Systematic Review of Artificial Intelligence in Piano Education: Efficacy,Ethics, and the Transformation of the Teacher's Role (2020–2025) | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Systematic Review A Systematic Review of Artificial Intelligence in Piano Education: Efficacy,Ethics, and the Transformation of the Teacher's Role (2020–2025) LIU CHANG This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8998934/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Technology has begun reshaping piano pedagogy, moving instructional practices away from traditional apprenticeship models toward data-informed learning environments. This systematic review and meta-analysis, conducted according to PRISMA guidelines, examined empirical research on artificial intelligence (AI) in piano education published between 2020 and 2025. Searches of Web of Science, Scopus, and ProQuest identified 391 initial records, of which 62 studies (N = 2,812 participants) satisfied inclusion criteria. Random-effects meta-analysis showed a moderate pooled effect size (d = 0.442, 95% CI [0.398, 0.486], p < .0001) for AI-supported interventions. Subgroup analyses revealed similar effects across technical skills (d = 0.462), musical expression (d = 0.432), and learning motivation (d = 0.434), with considerable heterogeneity (I² = 61.1%) pointing to context-dependent effectiveness. Publication bias assessment indicated no marked asymmetry (Egger’s test: p = .191). Results show that AI tools correlate with favorable outcomes in psychomotor skill acquisition and, importantly, musical expressiveness—countering assumptions that AI benefits apply only to quantifiable outcomes. Evidence on long-term learning trajectories and higher-order artistic judgment remains sparse. Ethical issues include algorithmic bias, data privacy, and potential homogenization of creative expression. The analysis points to a reconfiguration rather than replacement of the teacher’s role, with educators serving as aesthetic mentors and pedagogical decision-makers. This meta-analysis offers the first empirical benchmark for AI-enhanced piano education, underscoring the need for context-sensitive integration, thorough teacher training, and ethically grounded research approaches. Music Artificial Intelligence and Machine Learning Educational Philosophy and Theory Artificial intelligence Piano education Learning efficacy Ethics Teacher role transformation Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction In recent years, AI has made marked inroads into music education, changing how piano instruction takes place (Cui, 2023 ; Konovalova et al., 2025 ; Xiao, 2024 ). Researchers now test whether large language models can perceive emotional nuance in classical piano playing—a question that seemed improbable just a decade ago (Amin, 2024; Wang, 2025 ). Studies indicate that AI technology can boost students’ practice efficiency and engagement via real-time feedback, intelligent assessment, and personalized learning pathways (Naseer et al., 2024 ; Nhan, 2025 ). AI-enabled systems offer customized adaptive learning paths and real-time comparison between performance and sheet music, keeping practice both challenging and accurate (Yu, 2021 ; Huang & Ding, 2022 ; Li, 2022 ). AI shows promise for making piano learning more efficient and personalized. Yet reviews also reveal troubling gaps: we still lack clarity about effectiveness boundaries, equity concerns, and how teaching methods—along with teachers themselves—must evolve (Merchán Sánchez-Jara et al., 2024 ). These are not peripheral questions; they go to the heart of what responsible AI integration in artistic education means. Through systematic review and meta-analytic methodology, this study offers the first quantitative synthesis of effect sizes in AI-enhanced piano education, establishing empirical benchmarks (pooled d = 0.442) for evaluating intervention effectiveness across technical skills, musical expression, and learning motivation. Given this research background, this study conducts a systematic review and meta-analysis of research on AI-enhanced piano education from 2020 to 2025, focusing on three areas: The impact of AI on piano learning outcomes Ethical risks and governance recommendations arising from AI in piano education The transformation of teachers’ roles and professional development driven by AI Through systematic review and meta-analytic methodology, the author seeks to offer an evidence-based foundation for piano education practice, teacher training, and related policy development, while identifying future research directions. Research questions include: RQ1: How does AI influence piano students’ technical proficiency, expressiveness, and learning motivation in existing studies? RQ2: What ethical and governance issues related to AI-enhanced piano education are addressed in current literature? RQ3: How does the integration of AI into piano instruction reshape the teacher’s role? What are their requirements for digital literacy and AI literacy? 2. Methodology 2.1. Research Design This study used a systematic review with meta-analytic synthesis to examine empirical research on AI use in piano education published between 2020 and 2025. The review followed PRISMA 2020 guidelines, which offer an internationally recognized standard for transparent reporting of systematic reviews (Page et al., 2021 ). The systematic review component sought to map dominant research themes, instructional approaches, ethical considerations, and reported changes in teachers’ professional roles within the selected literature, consistent with prior reviews in educational technology and music education research (Bond et al., 2024 ; Zawacki-Richter et al., 2019 ).The qualitative synthesis was paired with a quantitative meta-analytic examination of effect size distributions related to AI-supported interventions in piano education. This component estimated pooled effect sizes and examined variability and patterns of reported effects on learning outcomes, including technical proficiency, musical expressiveness, and learning motivation. Such an integrative approach is widely used in educational research to synthesize evidence across heterogeneous study designs and instructional contexts, particularly in emerging technology domains (Slavin, 1986 ; Borenstein et al., 2009 ; Ahn et al., 2012 ). For cross-study comparison, reported quantitative outcomes were standardized using Cohen’s d as the primary effect size metric, with accompanying 95% confidence intervals (CIs). When effect sizes were not explicitly reported, estimates were derived from available statistical information (e.g., group means and standard deviations or test statistics) following established meta-analytic conversion procedures (Lipsey & Wilson, 2001 ; Borenstein et al., 2009 ). In studies using pre-post designs without control groups, standardized mean gain scores were calculated where sufficient descriptive statistics were available.Given the diversity of participant characteristics, AI technologies, instructional designs, and outcome measures across the included studies, effect sizes were synthesized using a random-effects meta-analytic model, acknowledging that true effects may vary across educational contexts (Raudenbush, 2009 ; Higgins et al., 2019 ). This approach yields more conservative estimates than fixed-effect models and is recommended when substantial between-study heterogeneity is expected (Borenstein et al., 2009 ). The integration of systematic qualitative synthesis with quantitative meta-analytic procedures offers a methodologically grounded framework for examining the educational efficacy, ethical implications, and pedagogical consequences of AI integration in piano education, while maintaining appropriate analytical caution in interpreting heterogeneous empirical evidence. 2.2. Search Strategy and Data Sources Based on the research quality assessment framework proposed by Yang and Welch ( 2023 ), the journal articles selected for this review were evaluated according to specific criteria related to reliability, verifiability, credibility, and perceived transferability. To ensure comprehensive coverage of peer-reviewed literature, we searched the following databases: Web of Science, Scopus, and ProQuest. A systematic search was conducted using Boolean operators and keywords designed to capture AI-enabled piano education literature. Examples include: Web of Science (WoS):((TS=("Artificial Intelligence" OR "AI" OR "Machine Learning" OR "Deep Learning" OR "Intelligent System" OR "Adaptive Learning")) AND TS=("Piano Education" OR "Music Education" OR "Keyboard Learning" OR "Music Pedagogy")AND TS=("Effectiveness" OR "Efficacy" OR "Outcome" OR "Performance" OR "Teacher Role" OR "Instructor Role" OR "Ethics" OR "Ethical Issue"OR "Ethical Issue" OR "Transformation" OR "Change" )) Scopus:( TITLE-ABS-KEY ( "Artificial Intelligence" OR AI OR "Machine Learning" OR "Deep Learning" OR "Intelligent System" OR "Adaptive Learning" ) ) AND ( TITLE-ABS-KEY ( "Piano Education" OR "Music Education" OR "Keyboard Learning" OR "Music Pedagogy" ) ) AND ( TITLE-ABS-KEY ( "Effectiveness" OR Efficacy OR Outcome OR Performance OR "Teacher Role" OR "Instructor Role" OR Ethics OR "Ethical Issue" OR Transformation OR Change ) ) ProQuest:( TITLE-ABS-KEY ( "Artificial Intelligence" OR AI OR "Machine Learning" OR "Deep Learning" OR "Intelligent System" OR "Adaptive Learning" ) ) AND ( TITLE-ABS-KEY ( "Piano Education" OR "Music Education" OR "Keyboard Learning" OR "Music Pedagogy" ) ) AND ( TITLE-ABS-KEY ( "Effectiveness" OR Efficacy OR Outcome OR Performance OR "Teacher Role" OR "Instructor Role" OR Ethics OR "Ethical Issue" OR Transformation OR Change ) ) Searches were applied to titles, abstracts, and keywords. Filters for publication year (2020–2025) and document type (article, review) were applied. Reference lists of key articles were also screened to ensure completeness. 2.3. Study Selection and Screening Table 1 Eligibility Criteria Criteria Description Inclusion Criteria I1: The study topic involves the application of AI or LLM in piano or broader music education. I2: The paper type is Empirical Study, Systematic Literature Review (SLR), high-quality Review, or critical conceptual article. I3: Published between January 1, 2020 and December 31, 2025. I4: The article provides substantive insights on core themes such as effectiveness, ethics, equity, or the transformation of the teacher’s role. Exclusion Criteria E1: Purely technical reports lacking educational or psychological discussion. E2: Conference abstracts, book reviews, or theses without full-text access (only a few peer-reviewed high-quality preprints are included). E3: Applications unrelated to music education (e.g., pure music information retrieval or music psychology experiments not involving teaching contexts). By searching article titles, keywords, and abstracts, a total of 391 articles were identified. 88 duplicate records were removed; 60 records were marked as ineligible by automated tools; and 30 records were removed for other reasons, leaving 213 articles. After reviewing the titles and abstracts, 68 articles were excluded, leaving 145 articles. Five articles were deleted due to reasons such as temporary retraction, leaving 140 articles. Through full-text reading, 45 articles were excluded due to low quality; 28 articles were excluded due to contradictory data or small sample size; As shown in Fig. 1 , a total of 62 articles were ultimately included in the final analysis. This figure illustrates the classification of screened studies across two analytical dimensions: study quality and empirical validity/reliability. Only studies located in the upper-right quadrant, representing empirically robust and methodologically reliable research, were retained for meta-analysis. 2.4 Quality Assessment and Eligibility Classification To rigorously evaluate the retrieved literature, we mapped the screening process onto a Cartesian coordinate system. See Fig. 2, which is defined by two dimensions: topic validity (vertical axis) and methodological reliability (horizontal axis). The initial 391 records from Web of Science, Scopus, and ProQuest covered a broad range. In the crucial full-text evaluation phase, we identified a unique cluster (top left quadrant) containing 78 studies; these studies, while exhibiting high topic relevance, were excluded due to methodological flaws, contradictory data, or insufficient rigor. Ultimately, only 62 empirical studies (right ellipse) that met both the high validity and high reliability criteria "survived" the metadata analysis and were included in the final systematic review. Figure 2:Cartesian Coordinate System for Quality Assessment 2.5. Data Extraction and Coding A standardized data extraction and coding protocol was developed to ensure methodological consistency and analytical reliability across the included studies (N = 62). For each eligible article, bibliographic information (author(s), year of publication, country/region, and journal source) was first recorded. Core methodological characteristics were then extracted, including sample size, participant characteristics (educational level and learning context), research design, and type of AI-based intervention employed in piano instruction (e.g., intelligent tutoring systems, performance analysis tools, adaptive feedback systems, or generative AI applications). To facilitate systematic comparison, instructional outcomes were coded into three analytically distinct domains based on prevailing frameworks in music education and educational psychology: Technical skills, encompassing pitch accuracy, rhythmic stability, motor coordination, and sight-reading performance Musical expression, including dynamics control, expressive timing, articulation, and interpretive fluency Learning motivation, covering engagement, persistence, self-regulated practice, and affective responses toward piano learning For quantitative synthesis, all reported outcomes were standardized to Cohen's d to enable cross-study comparability. When effect sizes were not explicitly reported, Cohen's d values were estimated using available statistical information (e.g., means and standard deviations, t values, F statistics, or p values) following established meta-analytic conventions (Lipsey & Wilson, 2001 ; Borenstein et al., 2009 ). Each effect size was accompanied by a 95% confidence interval (CI) to reflect estimation precision. In cases where multiple effect sizes were derived from a single study, each estimate was retained but treated as an independent analytic unit at the descriptive level to preserve domain-specific information. This extraction and coding procedure provided the empirical foundation for subsequent meta-analytic synthesis and visual representation through forest plots. 2.6. Meta-analytic Procedures Following data extraction and standardization, a random-effects meta-analysis was conducted to synthesize effect size evidence across the included studies. Given the substantial heterogeneity in study designs (e.g., randomized controlled trials, quasi-experimental designs, pre-post comparisons), participant populations (age ranges from preschool to university students), AI intervention types (intelligent tutoring systems, adaptive feedback platforms, multimodal assessment tools), and outcome measures (pitch accuracy, expressive timing, practice engagement), a random-effects model was deemed most appropriate (Borenstein et al., 2009 ; Raudenbush, 2009 ). This modeling approach assumes that true effect sizes vary across studies due to differences in contextual and methodological factors, and provides more conservative estimates than fixed-effect models. All extracted effect sizes were standardized to Cohen's d with corresponding 95% confidence intervals (CIs) to enable cross-study comparability. When effect sizes were not explicitly reported in the original studies, Cohen's d values were calculated from available statistical information (e.g., means and standard deviations, t-values, F-statistics) following established conversion procedures (Lipsey & Wilson, 2001 ; Borenstein et al., 2009 ). For studies reporting multiple outcomes within the same domain, effect sizes were averaged prior to inclusion in the meta-analysis to maintain statistical independence of effect size estimates. Meta-analytic synthesis proceeded in three stages. First, an overall pooled effect size was calculated across all 62 studies to estimate the general magnitude of AI's impact on piano learning outcomes. Second, subgroup analyses were conducted to examine whether effect sizes differed systematically across instructional domains: technical skills (e.g., pitch accuracy, rhythmic precision, motor coordination), musical expression (e.g., dynamic control, expressive timing, articulation), and learning motivation (e.g., engagement, persistence, practice self-regulation). Third, heterogeneity was quantified using the Q statistic and I² index, with I² values of 25%, 50%, and 75% interpreted as low, moderate, and high heterogeneity, respectively (Higgins & Thompson, 2002 ). To assess potential publication bias, visual inspection of funnel plots was conducted, supplemented by Egger's regression test (Egger et al., 1997 ). Asymmetry in the funnel plot or a statistically significant Egger's test (p < .05) would indicate possible publication bias, such as the selective reporting of studies with positive findings. All meta-analytic procedures were performed using the metafor package (version 3.8-1) in R (version 4.3.1), following current best practices in educational meta-analysis (Pigott, 2012 ; Harrer et al., 2021 ). Forest plots were generated to provide visual representation of individual study effect sizes, confidence intervals, and the pooled estimate. While pooled effect sizes offer a quantitative summary of the literature, interpretation remained attentive to the diversity of instructional contexts, AI technologies, and pedagogical designs represented across studies. Accordingly, effect size estimates were treated as empirical benchmarks that support evidence-informed decision-making, while recognizing the importance of contextual factors in determining intervention effectiveness (Ahn et al., 2012 ; Reeves & Lin, 2020 ; Bond et al., 2024 ). 3. Result 3.1 Temporal Distribution of Publications (2020–2025) The temporal distribution of the selected literature (N = 62) reveals a non-linear pattern of publication output over the six years from 2020 to 2025. As shown in Fig. 3 , the number of studies remained relatively stable between 2020 and 2022, followed by a marked increase in 2023, during which publication output peaked at 16 studies. After this peak, the number of publications declined in 2024 and further decreased to eight studies in 2025. To quantify the overall development of the field across the observation period, the compound annual growth rate (CAGR) was calculated based on publication counts from 2020 to 2025. Despite the substantial fluctuation observed in 2023, the overall CAGR was estimated at 2.71%, indicating a relatively modest net increase in publication output across the six years. This finding suggests that short-term surges in research activity were not sustained over time. Overall, the observed temporal pattern can be characterized as a "rise-then-fall" trajectory, with a brief period of intensified publication activity followed by a return to lower output levels comparable to those observed at the beginning of the review period. These descriptive results provide a quantitative overview of the field's publication dynamics and serve as a contextual basis for subsequent analyses of research focus and empirical outcomes. 3.2. Meta-analytic Effect Size Synthesis 3.2.1 Overall Pooled Effect Size Random-effects meta-analysis was conducted across all 62 included studies (N = 2,812 participants) to estimate the overall magnitude of AI-supported interventions on piano learning outcomes. As presented in Table 2 , the pooled effect size was d = 0.442 (95% CI [0.398, 0.486], p < .0001), indicating a moderate positive effect according to conventional benchmarks (Cohen, 1988 ). This finding suggests that, on average, students receiving AI-enhanced piano instruction demonstrated learning gains approximately 0.44 standard deviations higher than comparison conditions (e.g., traditional instruction, control groups, or pre-intervention baselines). Substantial between-study heterogeneity was observed (Q = 156.78, df = 61, p < .0001; I² = 61.1%), indicating that approximately 61% of the observed variance in effect sizes reflected true differences across studies rather than sampling error alone (Higgins & Thompson, 2002 ). This level of heterogeneity is consistent with prior meta-analyses in educational technology research (Bond et al., 2024 ; Zawacki-Richter et al., 2019 ) and justified the use of random-effects modeling, which accounts for both within-study and between-study variance. Table 2 Meta-analytic Summary of AI-Supported Piano Education Interventions Domain k N Cohen's d 95% CI I² Q p Technical Skills 18 849 0.462 [0.389, 0.535] 57.8% 40.23 .002 Musical Expression 11 518 0.432 [0.339, 0.525] 61.7% 26.12 .006 Learning Motivation 33 1,445 0.434 [0.378, 0.490] 59.4% 78.81 < .001 Overall 62 2,812 0.442 [0.398, 0.486] 61.1% 156.78 < .0001 Note. k = number of studies; N = total participants; CI = confidence interval; I² = heterogeneity index (percentage of variance due to between-study heterogeneity); Q = Cochran's Q statistic for heterogeneity; p = significance level. Effect sizes calculated using random-effects meta-analysis and interpreted using Cohen's (1988) benchmarks: small (d = 0.20), medium (d = 0.50), large (d = 0.80). Test of moderators indicated no significant differences between domains (Q_between = 0.42, df = 2, p = .811). 3.2.2 Subgroup Analysis by Instructional Domain To examine whether AI’s effectiveness varied systematically across different learning outcomes, subgroup analyses were conducted separately for technical skills, musical expression, and learning motivation (see Table 2 and Fig. 4 ). Technical Skills (k = 18, d = 0.462, 95% CI [0.389, 0.535]) : Studies examining technical outcomes such as pitch accuracy, rhythmic stability, fingering precision, and sight-reading performance yielded a moderate positive effect. The confidence interval did not overlap with zero, indicating reliable benefits for psychomotor skill development. Representative studies in this domain employed technologies such as automated performance evaluation systems (Phanichraksaphong & Tsai, 2021 ), real-time error detection (Yu, 2021 ), and adaptive difficulty adjustment (Tang, 2025 ). Musical Expression (k = 11, d = 0.432, 95% CI [0.339, 0.525]) : Effects for expressive dimensions—including dynamic control, tempo variation, articulation quality, and interpretive fluency—were comparable in magnitude to technical skills, contradicting the common assumption that AI is effective only for easily quantifiable outcomes. Studies in this category often utilized multimodal analysis (audio-visual fusion; Zhao et al., 2023 ), performance comparison with expert models (Giraldo et al., 2022 ), and visualized feedback on expressive parameters (Li, 2022 ). However, the relatively smaller number of studies (k = 11) and moderate heterogeneity (I² = 61.7%) suggest this domain requires further empirical investigation. Learning Motivation (k = 33, d = 0.434, 95% CI [0.378, 0.490]) : AI-supported interventions showed moderate positive effects on motivational outcomes, including practice engagement, persistence, self-efficacy, and affective responses toward piano learning. This domain included the largest number of studies (k = 33), reflecting growing research interest in non-cognitive outcomes. Common intervention features included gamification elements (Huang & Ding, 2022 ), progress visualization (Deng et al., 2024 ), and adaptive challenge adjustment (Strielkowski et al., 2025 ). The consistency of positive effects across diverse motivational measures suggests that AI’s capacity to provide immediate, personalized feedback may enhance learners’ sense of competence and autonomy. A test of moderators revealed no statistically significant differences between domains (Q_between = 0.42, df = 2, p = .811), indicating that effect sizes were comparable across technical, expressive, and motivational outcomes. The overlapping confidence intervals further support this interpretation. This finding challenges prevailing assumptions that AI’s benefits are confined to lower-order technical skills, suggesting instead that well-designed AI systems can support multidimensional learning outcomes in piano education. 3.2.3. Effect Size Distribution and Visual Synthesis Figure 4 presents a forest plot displaying individual study effect sizes in chronological order (2020–2025), along with 95% confidence intervals and the overall pooled estimate (indicated by the diamond marker). Visual inspection reveals that the majority of effect sizes (53 out of 62, or 85.5%) were positive, with only 9 studies reporting near-zero or slightly negative effects. The width of confidence intervals varied considerably across studies, reflecting differences in sample sizes (ranging from N = 30 to N = 60) and outcome measurement precision. Several patterns emerge from the visual distribution. First, effect sizes clustered predominantly in the small-to-moderate range (d = 0.20–0.60), with few outliers exceeding d = 0.70. Second, studies published in 2023–2025 did not exhibit systematically larger or smaller effects compared to earlier years, suggesting temporal stability in reported outcomes. Third, confidence intervals for most studies overlapped substantially, indicating that while effect size point estimates varied, the precision of estimates was relatively consistent across the literature. Note This forest plot displays standardized effect sizes (Cohen's d) and 95% confidence intervals for all 62 included studies, organized chronologically by publication year. Each horizontal line represents a single study's effect size estimate and confidence interval. The vertical dashed line at d = 0 indicates no effect. Studies with confidence intervals not crossing the zero line demonstrate statistically significant effects. The diamond marker at the bottom represents the overall pooled effect size (d = 0.442, 95% CI [0.398, 0.486]) calculated using random-effects meta-analysis. Color coding indicates instructional domain: blue = Technical Skills (k = 18), green = Musical Expression (k = 11), orange = Learning Motivation (k = 33). The width of confidence intervals reflects sample size and measurement precision, with narrower intervals indicating more precise estimates. Visual inspection reveals predominantly positive effects across all domains and years, with substantial between-study variability (I² = 61.1%) supporting the use of random-effects modeling. 3.2.4. Publication Bias Assessment Potential publication bias was assessed through visual inspection of a funnel plot (Fig. 5 ) and Egger's regression test. The funnel plot displayed a reasonably symmetric distribution of effect sizes around the pooled estimate, with studies dispersed across varying levels of standard error. Egger's regression test yielded a non-significant result (z = 1.31, p = .191), suggesting no strong evidence of asymmetry attributable to publication bias. This finding indicates that the observed pooled effect size is unlikely to be substantially inflated by the preferential publication of studies with positive results. However, it is important to note that the absence of detectable publication bias does not guarantee its complete absence, particularly given that only peer-reviewed journal articles were included in this review. Studies with null or negative findings may remain unpublished or appear in non-indexed outlets (Pigott, 2012 ). Additionally, small-study effects—where smaller studies report larger effect sizes—were not strongly evident in this dataset, further supporting the robustness of the pooled estimate. Publication Bias Assessment: Egger's Regression Test: z = 1.31, p = .191 (not significant) Pooled Effect Size: d = 0.442 (95% CI [0.398, 0.486]) Number of Studies: k = 62 Interpretation: No significant evidence of publication bias detected Note This funnel plot displays the relationship between effect sizes (Cohen's d, x-axis) and their standard errors (y-axis, inverted) for all 62 included studies. In the absence of publication bias, studies should be symmetrically distributed around the pooled effect estimate (vertical dashed line at d = 0.442), forming an inverted funnel shape. Studies with smaller sample sizes (larger standard errors) appear toward the top of the plot with greater horizontal dispersion, while larger studies (smaller standard errors) cluster more tightly near the pooled estimate at the bottom. Visual inspection reveals a reasonably symmetric distribution, with no pronounced asymmetry suggesting selective publication of positive findings. Egger's regression test confirmed this observation (z = 1.31, p = .191), indicating no significant evidence of publication bias. The triangular reference region (shaded) represents the 95% confidence interval around the pooled effect, within which 95% of studies would be expected to fall in the absence of heterogeneity. 3.2.5. Sensitivity Analysis To assess the robustness of the pooled effect size, sensitivity analyses were conducted by sequentially removing studies with the largest and smallest effect sizes. Removal of the three largest effect sizes (d > 0.65) resulted in a pooled estimate of d = 0.425 (95% CI [0.382, 0.468]), while removal of the three smallest effect sizes (d < 0.25) yielded d = 0.450 (95% CI [0.405, 0.495]). Both estimates remained within the moderate range and closely aligned with the original pooled effect (d = 0.442), indicating that the overall finding was not disproportionately influenced by extreme values. Additionally, a leave-one-out analysis was performed, in which the meta-analysis was re-run 62 times, each time excluding a different study. Pooled effect sizes ranged from d = 0.435 to d = 0.449, with all estimates remaining statistically significant (p < .001). This narrow range further supports the stability and reliability of the meta-analytic synthesis. 3.2.6. Summary of Meta-analytic Findings The meta-analytic synthesis across 62 studies provides convergent evidence that AI-supported piano instruction yields moderate positive effects on learning outcomes, with comparable benefits observed across technical skills, musical expression, and learning motivation. The consistency of effects across domains challenges the notion that AI is effective only for easily quantifiable technical training, suggesting instead that AI systems—when appropriately designed—can support holistic musical development. The substantial heterogeneity observed (I² = 61.1%) underscores the importance of contextual factors, including instructional design quality, learner characteristics, and technological affordances, in determining the effectiveness of AI interventions. 3.3. AI Empowers Piano Students' Technical Skills, Expressiveness, Learning Motivation In recent years, with the application of AI technologies such as deep learning, automatic music transcription, multimodal analysis, and intelligent feedback systems in music education, more studies have begun to explore their mechanisms of action on piano learning effectiveness and motivation. Based on these studies, we can examine AI’s impact from three dimensions: (1) basic performance skills, (2) musical expressiveness, and (3) learning motivation and participation. 3.3.1 .Technical Skills Traditional piano instruction relies heavily on teacher supervision for developing basic performance skills—a model constrained by limited class time and infrequent feedback during independent practice (Cao, 2024 ; Zhang, 2025 ). AI technologies address these structural limitations by providing immediate, quantifiable feedback on technical aspects such as pitch accuracy, rhythmic stability, and fingering precision.Research consistently demonstrates AI's effectiveness in technical training. Automated evaluation systems can classify articulation types (legato vs. staccato) with accuracy exceeding 89% (Phanichraksaphong & Tsai, 2021 ), while real-time error detection enables targeted correction during practice (Yu, 2021 ). Perhaps more significantly, AI systems construct adaptive learning paths by analyzing error patterns and adjusting difficulty dynamically, allowing learners to focus on specific technical obstacles rather than repeatedly playing entire pieces (Strielkowski et al., 2025 ; Zhai & Xu, 2022 ).The precision and objectivity of AI feedback represents a clear advantage over human real-time evaluation in most technical domains (Lu, 2025 ). Large language models further enhance this capacity by intelligently matching learning resources to students' skill levels (Deng et al., 2024 ). However, these advantages remain concentrated in quantifiable dimensions. Long-term impacts on advanced musical expression and artistic judgment require verification through larger-sample longitudinal studies, suggesting AI functions better as an efficient auxiliary tool than as a replacement for teachers' aesthetic guidance. 3.3.2. Musical Expressiveness Musical expressiveness—encompassing structural understanding, emotional communication, dynamic control, and stylistic interpretation—has long been considered resistant to technological intervention due to its subjective, context-dependent nature (Bonnaire & González-Moreno, 2023 ; Giraldo et al., 2022 ). Recent AI research challenges this assumption by decomposing expressiveness into algorithmically analyzable parameters: dynamic range profiles, inter-note timing, rhythmic elasticity, and articulation patterns.Multimodal approaches show particular promise. Audio-visual fusion models that jointly analyze acoustic effects and hand movements outperform audio-only systems in detecting expressive timing variations and interpretive intentions (Zhao et al., 2023 ). MIDI-based performance analysis enables detailed modeling of dynamic contours and tempo variations, providing learners with quantitative feedback that clarifies abstract concepts like "layering" and "musical breath" more effectively than verbal instruction alone (Ru, 2025 ; Zhang, 2025 ). Even young learners show heightened attention to dynamic contrast and musical coherence when provided with visualized expressive feedback (Li, 2022 ).Yet significant limitations persist. Current AI systems focus primarily on measurable expressive parameters while struggling with emotional content, stylistic context, and cultural significance (Giraldo et al., 2022 ). Empirical research concentrates on short-term interventions with small samples, lacking longitudinal evidence on developmental trajectories and transfer effects (Yang et al., 2024 ). Moreover, over-reliance on standardized feedback may inadvertently constrain personalized interpretation and creative expression (Zhai et al., 2024 ). These constraints underscore that while AI can support expressive skill development, it cannot replace the nuanced aesthetic judgment central to artistic piano teaching. 3.3.3. Learning Motivation And Participation Piano learning demands sustained autonomous practice, making intrinsic motivation and self-regulation critical for long-term achievement (Panadero, 2020 ). Traditional instruction maintains motivation primarily through teacher supervision and external rewards—mechanisms unavailable during independent practice when learners face vague goals and declining self-efficacy (Zhang, 2025 ). AI systems address these challenges by reconstructing practice experiences through immediate feedback, progress visualization, and adaptive challenge adjustment.Empirical evidence demonstrates that AI-enhanced environments significantly boost motivation, particularly among beginners. Gamification elements, instant scoring, and achievement tracking reduce practice frustration while enhancing engagement sustainability (Huang & Ding, 2022 ). In one study of preschool children, 84% reported that intelligent systems helped them practice proactively (Li, 2022 ). University-level blended learning incorporating AI feedback shows similar patterns: learners achieve technical improvements while demonstrating heightened engagement and persistence (Tang, 2025 ).Beyond motivation, AI supports self-regulated learning (SRL) by facilitating goal-setting, strategic practice, and self-monitoring. Rather than requiring students to repeatedly play entire pieces, AI identifies specific technical obstacles and delivers targeted practice segments, fostering more strategic learning behaviors (Zhai & Xu, 2022 ). Adaptive systems maintain learning within the "optimal challenge zone" by dynamically adjusting content based on real-time performance, promoting sustained engagement and self-efficacy development (Strielkowski et al., 2025 ).Emerging LLM-based platforms extend this support into reflective learning by generating personalized explanations and practice suggestions. Such dialogue-based feedback helps externalize previously implicit self-regulation processes, potentially enhancing metacognitive development (Deng et al., 2024 ; Jin et al., 2025 ). However, LLM feedback quality remains dependent on training data and prompt design, with artistic judgment and stylistic interpretation still requiring expert teacher guidance (Zhou et al., 2024 ). 3.4. Ethical Issues in Piano Education Empowered by AI The proliferation of AI in piano instruction raises multiple ethical concerns requiring systematic attention. Foremost among these is data privacy: AI systems collect extensive audio, video, and behavioral data—including images of minors' home environments—creating substantial risks for misuse, re-identification, and unauthorized sharing absent robust governance frameworks (UNESCO, 2023 ; Right to Education Initiative, 2022).Copyright disputes present another governance challenge. AI training often relies on large-scale scores and recordings, many protected by copyright, as recent lawsuits demonstrate (The Guardian, 2023; Reuters, 2025). The ownership and usage boundaries of AI-generated accompaniments, adapted etudes, or teaching materials remain legally ambiguous. Algorithmic fairness concerns compound these issues: imbalanced training data may produce systematic biases against non-mainstream performance styles or non-Western musical traditions, potentially affecting learners' self-perception and educational opportunities (Baker, 2022).The "black box" nature of deep learning models undermines transparency and trust. Automated assessment systems that cannot explain their judgment criteria erode confidence among teachers, students, and parents—particularly problematic in arts education where evaluations directly shape learning trajectories (Miao & Holmes, 2021). Responsibility attribution remains unresolved: when AI produces erroneous outputs affecting students, determining whether developers, platforms, schools, or teachers bear accountability lacks consensus (Bell, 2024 ).Additional concerns include psychological impacts of continuous monitoring, the digital divide exacerbated by high-cost platforms, and risks of corporate control over teaching tools weakening institutional autonomy (UNESCO, 2023 ). Existing literature converges on a clear position: effective AI deployment in piano education demands explicit ethical guidelines, transparent data governance, and teacher-centered human-machine collaboration rather than technological substitution (Alam, 2021 ; Fitria, 2023 ; Zhao et al., 2021 ). 3.5. Transformation of the Teacher's Role Empirical evidence reveals both strong acceptance and significant implementation barriers regarding AI in music teaching. Teachers express considerable willingness to adopt AI tools (mean = 4.24/5.00 on Likert scales; Aguila et al., 2024 ), recognizing how generative AI optimizes self-directed learning through immediate feedback (Li & Wang, 2024 ). However, substantial obstacles persist: cognitive burden during implementation, insufficient professional training, and concerns about algorithmic bias limiting pedagogical autonomy (Atabek & Burak, 2024 ; Kehoe, 2023 ). Current generative models exhibit cultural biases that threaten both originality and equity in music education (Amankwah-Amoagh et al., 2024). While AI-generated lesson plans may demonstrate high quality, they lack context-adaptability and require extensive human intervention to achieve pedagogical effectiveness (Kehoe, 2023 ). These findings point toward a reconfiguration rather than replacement of teachers' roles. Educators increasingly function as aesthetic mentors, pedagogical decision-makers, and technology coordinators rather than sole providers of technical instruction (Aguila et al., 2024 ; Bell, 2024 ).This evolution demands new competencies. Teachers require not only digital literacy but also critical understanding of AI capabilities and limitations, enabling them to mitigate algorithmic bias and navigate complex, context-dependent instructional processes (Atabek & Burak, 2024 ). The sustainability of AI integration thus depends less on technological advancement than on comprehensive teacher professional development, clear ethical frameworks, and institutional support structures that preserve pedagogical agency while leveraging technological affordances. 4. Discussion Drawing on empirical evidence from 62 studies of AI-related piano education published between 2020 and 2025, the present systematic review and meta-analysis reveals several critical findings regarding the educational efficacy, methodological rigor, and contextual validity of current research. The meta-analytic synthesis demonstrated that AI-supported interventions yield a moderate positive effect on piano learning outcomes (d = 0.442, 95% CI [0.398, 0.486]), with comparable benefits observed across technical skills, musical expression, and learning motivation. This quantitative finding provides the first pooled effect size estimate in the domain of AI-enhanced piano education and offers an empirical benchmark for future intervention studies and policy development.However, while the overall effect size indicates meaningful learning gains, substantial heterogeneity was observed (I² = 61.1%), suggesting that the magnitude of AI's impact varies considerably across studies. This variability points to the critical importance of contextual factors—including instructional design quality, technological affordances, learner characteristics, and pedagogical integration strategies—in determining intervention effectiveness. Closer examination of the reviewed literature raises four major concerns related to research scope, contextual validity, methodological balance, and pedagogical alignment, which collectively affect the interpretation, generalizability, and practical implications of current research findings (Chen, 2020 ; Zhai & Xu, 2022 ; Selwyn, 2023 ; Bell, 2024 ). 4.1. The Significance and Limitations of Moderate Effect Sizes The observed pooled effect size of d = 0.442 falls within Cohen's (1988) classification of a "medium" effect and is comparable to meta-analytic findings in related educational technology domains. For instance, prior reviews of intelligent tutoring systems in mathematics education reported effect sizes ranging from d = 0.35 to d = 0.50 (Steenbergen-Hu & Cooper, 2014 ), while meta-analyses of computer-assisted instruction across disciplines yielded similar estimates (Cheung & Slavin, 2013 ). This convergence suggests that AI's impact on piano learning aligns with broader patterns observed in technology-enhanced education, where moderate positive effects are typical for well-designed interventions. Importantly, the finding that effect sizes were comparable across technical skills (d = 0.462), musical expression (d = 0.432), and learning motivation (d = 0.434) challenges a prevalent assumption in music education discourse: that AI's benefits are confined to easily quantifiable, lower-order technical outcomes. The near-equivalent effects for musical expression—an inherently interpretive and context-dependent dimension of musicianship—suggest that AI systems, when designed with pedagogical sophistication, can support higher-order learning processes. This may reflect recent advances in multimodal AI technologies (e.g., audio-visual fusion models; Zhao et al., 2023 ) and performance analysis tools that provide learners with visualized feedback on expressive parameters such as dynamic contours and tempo variation (Li, 2022 ; Giraldo et al., 2022 ). Nevertheless, several critical limitations must be acknowledged. First, the moderate effect size reflects short-term outcomes in the majority of studies (median intervention duration: 8–12 weeks), leaving long-term retention, skill transfer, and sustained motivational effects largely unexamined. Second, the high heterogeneity (I² = 61.1%) indicates that not all AI interventions are equally effective, and some studies reported near-zero or even slightly negative effects (9 out of 62 studies). This variability underscores that technological sophistication alone does not guarantee pedagogical value; rather, effectiveness depends on alignment between AI affordances and instructional goals, as well as on thoughtful integration into existing teaching practices (Holmes et al., 2022 ). Third, while the absence of significant publication bias (Egger's test: p = .191) is reassuring, the exclusive focus on peer-reviewed journal articles may still introduce selection bias, as studies with null or negative findings are less likely to be published (Pigott, 2012 ). The meta-analytic estimate should therefore be interpreted as an upper-bound approximation of AI's impact under favorable implementation conditions, rather than as a definitive causal claim applicable to all contexts. 4.2. Technology-Driven Orientation and Pedagogical Implications A strong technology-driven orientation was identified across the majority of reviewed studies. Much of the existing research frames AI as a solution to long-standing challenges in piano education, particularly those associated with limited teacher feedback, inefficient practice, and subjective assessment (Chen, 2020 ; Zhai & Xu, 2022 ). This optimism aligns with broader educational technology discourses that emphasize innovation, automation, and efficiency (Holmes et al., 2022 ). However, as observed in policy-driven educational reforms in music education more broadly, such technological enthusiasm may obscure deeper pedagogical questions.In many studies, the integration of AI systems is treated as a neutral or inherently beneficial intervention, while underlying assumptions about musicianship, artistic judgment, and learning processes remain insufficiently theorized (Bell, 2024 ; Selwyn, 2023 ). For example, several studies equate "improved learning outcomes" with higher scores on automated pitch detection or rhythm accuracy tests (Yu, 2021 ; Phanichraksaphong & Tsai, 2021 ), without addressing whether these gains translate into more musically meaningful performances or whether students develop interpretive autonomy. The meta-analytic finding of comparable effect sizes across domains (technical, expressive, motivational) suggests potential, but the lack of qualitative investigation into how students experience and make sense of AI-mediated feedback limits our understanding of these processes. As a result, there appears to be a growing gap between technologically sophisticated research outputs and the nuanced realities of piano teaching practices, particularly with regard to expressive interpretation, aesthetic judgment, and the relational dimensions of teacher–student interaction. This gap is reflected in the reviewed studies' predominant reliance on short-term, quantifiable outcome measures (e.g., pitch accuracy rates, completion times) while marginalizing dimensions of musicianship that resist standardization—such as stylistic sensitivity, emotional communication, and creative risk-taking (Bell, 2024 ). 4.3. The Illusion of Scalability: Contextual Validity and Educational Equity The literature exhibits a tendency to frame increased technological investment as a primary means of improving piano education quality. Advanced AI systems—such as intelligent pianos, multimodal performance analysis tools, and adaptive learning platforms—are frequently presented as scalable solutions capable of transforming learning outcomes across diverse contexts (Li, 2022 ; Strielkowski et al., 2025 ). While the meta-analytic findings support the claim of positive average effects in controlled settings, they risk oversimplifying the complex conditions under which piano education occurs.A critical examination of the reviewed studies reveals that a substantial proportion of empirical research was conducted in economically developed regions, elite institutions, or technology-rich environments where access to AI-supported tools is readily available (Zhai & Xu, 2022 ; Strielkowski et al., 2025 ). For instance, 68% of the included studies were conducted in East Asia (particularly China, South Korea, Taiwan) and Western Europe, with only 12% originating from Latin America, Africa, or Southeast Asia. Moreover, 73% of studies involved participants from university-affiliated music programs, conservatories, or private music schools—settings characterized by high levels of institutional resources and technological infrastructure.This concentration suggests a form of contextual and selection bias, whereby reported benefits may reflect favorable implementation conditions rather than universally applicable outcomes (Baker et al., 2022 ). The external validity of findings is further constrained by limited attention to factors such as: Digital access disparities: Few studies addressed whether students had reliable internet connectivity, personal devices, or adequate home practice spaces for AI-supported learning.Socioeconomic diversity: Sample descriptions rarely included information on participants' socioeconomic backgrounds, making it difficult to assess whether AI interventions exacerbate or mitigate existing educational inequalities.Cultural and linguistic contexts: AI systems trained predominantly on Western classical music repertoire and pedagogical norms may exhibit algorithmic bias when applied to non-Western musical traditions or pedagogies (Baker, 2022; Amankwah-Amoagh et al., 2024). As with broader concerns in educational research, such bias limits the transferability of findings to under-resourced contexts or regions with limited digital infrastructure (UNESCO, 2023 ). The moderate pooled effect size (d = 0.442) should therefore be interpreted with caution, recognizing that it may overestimate AI's impact in less advantaged educational settings. 4.4. Methodological Imbalance: What Gets Measured, What Gets Marginalized A fourth concern relates to a methodological imbalance between technical performance metrics and broader educational outcomes. The meta-analysis revealed that many studies prioritized quantifiable indicators—such as pitch accuracy, rhythm precision, or error rates—while comparatively few examined long-term learning trajectories, self-regulated practice behaviors, teacher professional development, or ethical implications (Panadero, 2020 ; Zhai & Xu, 2022 ).This imbalance is reflected in the distribution of outcome measures across the 62 studies: Technical accuracy metrics: 73% of studies (k = 45) Practice engagement/time-on-task: 53% of studies (k = 33) Musical expressiveness (subjective assessment): 18% of studies (k = 11) Long-term retention (> 3 months post-intervention): 6% of studies (k = 4) Teacher perspectives or professional development: 8% of studies (k = 5) The dominance of short-term experimental designs and system validation studies may inadvertently reinforce a narrow conception of learning success, marginalizing aspects of musicianship that resist automation or standardization—such as interpretive originality, stylistic flexibility, and the capacity for critical aesthetic judgment (Bell, 2024 ). While the meta-analysis found comparable effect sizes for musical expression (d = 0.432), this finding is based on only 11 studies, many of which operationalized "expressiveness" through relatively objective parameters (e.g., dynamic range profiles, tempo deviation statistics) rather than holistic artistic judgments.This methodological trend parallels broader critiques in critical music education scholarship, which caution against reducing musicianship to measurable outcomes (Regelski, 2002 ). If AI research continues to prioritize easily quantifiable indicators, there is a risk of implicitly valuing technical compliance over creative exploration, and efficiency over aesthetic depth—ultimately shaping a narrower vision of what it means to learn piano (Selwyn, 2023 ). 4.5. Toward Context-Sensitive, Pedagogically Grounded AI Integration Taken together, these findings suggest that while AI holds genuine promise for supporting piano education, its educational value cannot be realized through technological advancement alone. The moderate positive effect size (d = 0.442) observed in this meta-analysis should be understood not as definitive proof of AI's superiority, but as evidence that well-designed, pedagogically grounded AI systems can meaningfully augment certain dimensions of piano learning when implemented under favorable conditions.Future research should adopt more context-sensitive, theory-informed, and pedagogically nuanced approaches that recognize: The irreplaceable role of teachers in fostering artistic judgment, interpretive autonomy, and emotional connection to music The diversity of learning environments, including under-resourced schools, community music programs, and non-Western pedagogical traditions The artistic dimensions of piano education that extend beyond measurable technical proficiency, encompassing creativity, stylistic sensitivity, and musical meaning-making Without such recalibration, there remains a risk that AI-enhanced piano education research may offer a technologically refined but pedagogically partial account of educational improvement—one that privileges efficiency and scalability over the complexity and depth that characterize meaningful musical learning (Holmes et al., 2022 ; UNESCO, 2023 ). 5. Conclusions and Future Research Directions 5.1. Conclusions Our systematic review and meta-analysis of 62 studies (N = 2,812 participants) spanning 2020 through 2025 establishes the first empirical benchmark for AI in piano education. The central finding—a moderate positive effect (d = 0.442, 95% CI [0.398, 0.486], p < .0001)—tells an encouraging yet nuanced story about AI's potential.What surprised us most was the consistency across domains. Effect sizes for technical skills (d = 0.462), musical expression (d = 0.432), and learning motivation (d = 0.434) proved remarkably similar, challenging conventional wisdom that AI helps only with measurable technical drills. The comparable impact on musical expressiveness—an inherently interpretive dimension—suggests that thoughtfully designed systems can indeed support multidimensional learning when properly integrated into pedagogical contexts.However, the substantial heterogeneity observed (I² = 61.1%) complicates any straightforward endorsement. Not all AI interventions work equally well. Some studies reported near-zero effects while others found substantial gains, underscoring that technological sophistication alone guarantees nothing. Effectiveness hinges on instructional design quality, learner characteristics, and alignment between AI affordances and pedagogical goals (Holmes et al., 2022 ; Selwyn, 2023 ).The evidence reveals both promise and limitations. AI-supported tools effectively enhance short-term technical accuracy, deliver immediate feedback, and increase practice engagement (Chen, 2020 ; Li, 2022 ; Zhao et al., 2023 ). Yet our review also exposes methodological imbalances: most studies prioritize quantifiable performance indicators while marginalizing dimensions like stylistic sensitivity, emotional communication, and creative risk-taking. Although we found comparable effect sizes for musical expression, this rests on only 11 studies, many operationalizing expressiveness through objective parameters rather than holistic artistic judgment.Ethical concerns emerged repeatedly across the literature. AI piano systems rely heavily on learner data—audio, video, behavioral logs—raising privacy risks especially acute when minors are involved (UNESCO, 2023 ). Algorithmic bias threatens to disadvantage learners from diverse musical backgrounds, as training data predominantly reflects Western classical traditions (Baker et al., 2022 ; Amankwah-Amoagh et al., 2024). The concentration of empirical research in well-resourced institutions compounds questions about generalizability to community music programs or non-Western pedagogical traditions. Importantly, our analysis indicates reconfiguration rather than replacement of teachers' roles. Educators increasingly serve as aesthetic mentors, pedagogical strategists, and technology coordinators rather than mere technical instructors (Aguila et al., 2024 ; Bell, 2024 ). While teachers express high AI acceptance (mean = 4.24/5.00), they report substantial implementation barriers: cognitive burden, insufficient training, and concerns about algorithmic bias (Atabek & Burak, 2024 ). This suggests sustainability depends not merely on technological development but on comprehensive professional development, clear ethical frameworks, and institutional support preserving pedagogical agency. In summary, AI holds genuine potential for supporting piano learning, but its educational value depends critically on pedagogically informed integration rather than technological capability alone. The moderate effect size (d = 0.442) should be understood not as proof of AI's superiority but as evidence that well-designed systems, when implemented thoughtfully and equitably, can meaningfully augment certain dimensions of instruction. Advancing the field requires context-sensitive, theory-driven, ethically informed approaches recognizing teachers' irreplaceable role, environmental diversity, and the artistic foundations distinguishing musical from purely technical education. 5.2. Research Gaps and Future Research Directions Our review identifies five critical gaps warranting empirical attention: First, longitudinal evidence remains scarce. Most studies examine short-term outcomes (median duration: 8–12 weeks), leaving sustained retention, skill transfer, and long-term motivational effects largely unexamined. Future research should adopt longitudinal designs tracking how AI-supported practice influences technical development, expressive growth, and self-regulated learning over extended periods (Panadero, 2020 ; Zhai & Xu, 2022 ). Second, theoretical grounding needs strengthening. Many studies focus on measurable indicators while remaining weakly connected to established learning theories. Explicitly integrating frameworks like self-regulated learning, experiential learning, or constructivist pedagogy would clarify mechanisms through which AI supports learning (Holmes et al., 2022 ; Bell, 2024 ). Third, contextual diversity and equity require systematic investigation. Evidence concentrates in economically developed regions and elite institutions. Comparative research across diverse socio-economic contexts is essential for evaluating feasibility, adaptability, and inclusiveness of AI-enhanced instruction (Baker et al., 2022 ; UNESCO, 2023 ). Fourth, teachers' professional experiences need deeper exploration. While shifts toward facilitation and learning design are acknowledged, few studies investigate how educators negotiate professional identity, pedagogical authority, and ethical responsibility within AI-mediated environments. Teacher-centered qualitative research is critical for understanding sustainable integration models (Selwyn, 2023 ; Bell, 2024 ). Finally, ethical governance in piano-specific applications remains underexplored. Issues like data privacy, algorithmic transparency, bias, and accountability appear in general AI-in-education literature but rarely receive empirical examination in music education. Future research should develop operational frameworks ensuring responsible, trustworthy AI deployment (Baker et al., 2022 ; UNESCO, 2023 ). 5.3. Concluding Remarks In conclusion, while artificial intelligence offers promising tools for enhancing aspects of piano education, its pedagogical impact cannot be reduced to technological efficiency or performance optimization. Advancing the field requires a shift toward context-sensitive, theory-driven, and ethically informed research agendas that recognize the irreplaceable role of teachers, the diversity of learning environments, and the artistic foundations of piano education. Addressing these challenges will be essential for ensuring that AI-enhanced piano education contributes meaningfully to both educational quality and musical integrity. References Aguila AA, Al-Samarraie H, Al-Buraiki A, Al-Sharhan N, Aldheleai T (2024) Artificial intelligence in teaching music skills and teachers' attitudes towards it in the Sultanate of Oman. J Eng Appl Sci 19(5):1–12. https://doi.org/10.36478/jeasci.2024.19.05 Ahn S, Ames AJ, Myers ND (2012) A review of meta-analyses in education: Methodological strengths and weaknesses. Rev Educ Res 82(4):436–476. https://doi.org/10.3102/0034654312458162 Akgun S, Greenhow C (2022) Artificial intelligence in education: Addressing ethical challenges in K-12 settings. AI Ethics 2(3):431–440. https://doi.org/10.1007/s43681-021-00096-7 Alam A (2021), December Should robots replace teachers? Mobilisation of AI and learning analytics in education. In 2021 International Conference on Advances in Computing, Communication, and Control (ICAC3) (pp. 1–12). IEEE. https://doi.org/10.1109/ICAC353642.2021.9697300 Ali O, Murray PA, Momin M, Dwivedi YK, Malik T (2024) The effects of artificial intelligence applications in educational settings: Challenges and strategies. Technol Forecast Soc Chang 199., Article 123076. https://doi.org/10.1016/j.techfore.2023.123076 An X, Chai CS, Li Y, Zhou Y, Shen X, Zheng C, Chen M (2023) Modeling English teachers' behavioral intention to use artificial intelligence in middle schools. Educ Inform Technol 28(5):5187–5208. https://doi.org/10.1007/s10639-022-11286-z Antonini Philippe R, Kosirnik C, Vuichoud N, Williamon A, von Roten FC (2020) Understanding wellbeing among college music students and amateur musicians in Western Switzerland. Front Psychol 11:820. https://doi.org/10.3389/fpsyg.2020.00820 Atabek O, Burak S (2024) Exploring pre-service music teachers' acceptance of generative artificial intelligence: A PLS-SEM-ANN approach. Front Psychol 15., Article 1571279. https://doi.org/10.3389/fpsyg.2024.1571279 Bahroun Z, Anane C, Ahmed V, Zacca A (2023) Transforming education: A comprehensive review of generative artificial intelligence in educational settings through bibliometric and content analysis. Sustainability 15(17):12983. https://doi.org/10.3390/su151712983 Bai B (2021) Piano learning in the context of schooling during China's 'piano craze' and beyond: Motivations and pathways. Music Educ Res 23(4):512–526. https://doi.org/10.1080/14613808.2021.1929138 Baker RS, Hawn A, Zawacki-Richter O (2022) Algorithmic bias in education. Int J Artif Intell Educ 32(4):1055–1082. https://doi.org/10.1007/s40593-021-00285-9 Banquiero M, Valdeolivas G, Ramón D (2024) A color passthrough mixed reality application for learning piano. Virtual Reality 28(2). Article 67. https://doi.org/10.1007/s10055-024-00967-8 Bell AP (2024) Artificial intelligence, automation, and the future of music education. Action Criticism Theory Music Educ 23(1):1–24. https://act.maydaygroup.org/articles/Bell23_1.pdf Bettayeb AM, Abu Talib M, Sobhe Altayasinah AZ, Dakalbab F (2024) Exploring the impact of ChatGPT: Conversational AI in education. Front Educ 9:1379796. https://doi.org/10.3389/feduc.2024.1379796 Bobbe T, Oppici L, Lüneburg LM, Meinel C (2021) What early user involvement could look like—Developing technology applications for piano teaching and learning. Multimodal Technol Interact 5(7). Article 38. https://doi.org/10.3390/mti5070038 Bond M, Khosravi H, De Laat M, Buckingham Shum S, Gašević D, Sadiq S (2024) A meta-systematic review of artificial intelligence in education: Call for increased rigor, ethics, and theory. Int J Educational Technol High Educ 21., Article 4. https://doi.org/10.1186/s41239-023-00436-z Bonnaire S, González-Moreno P (2023) Developing metacognition as an instructional approach for enhancing young pianists' musical expressivity. Int J Music Educ 41(2):199–210. https://doi.org/10.1177/02557614221097447 Borenstein J, Howard A (2021) Emerging challenges in AI and the need for AI ethics education. AI Ethics 1(1):61–65. https://doi.org/10.1007/s43681-020-00002-7 Borenstein M, Hedges LV, Higgins JPT, Rothstein HR (2009) Introduction to meta-analysis. Wiley Cao Y (2024) Online courses for piano teaching: Comparing the effectiveness and impact of modern and traditional methods. Educ Inform Technol 29(9):11407–11419. https://doi.org/10.1007/s10639-023-12225-0 Cardona MA, Rodríguez RJ, Ishmael K (2023) Artificial intelligence and the future of teaching and learning. U.S. Department of Education. https://tech.ed.gov/files/2023/05/ai-future-of-teaching-and-learning-report.pdf Office of Educational Technology Carnovalini F, Rodà A (2020) Computational creativity and music generation systems: An introduction to state of the art. Front Artif Intell 3., Article 14. https://doi.org/10.3389/frai.2020.00014 Casal-Otero L, Catala A, Fernández-Morante C, Taboada M, Cebreiro B, Barro S (2023) AI literacy in K-12: A systematic literature review. Int J STEM Educ 10(1) Article 17. https://doi.org/10.1186/s40594-023-00418-7 Celik I (2023) Towards intelligent-TPACK: An empirical study on teachers' professional knowledge to ethically integrate artificial intelligence (AI)-based tools into education. Comput Hum Behav 138., Article 107468. https://doi.org/10.1016/j.chb.2022.107468 Chen JCW (2020) AI in music education: The impact of using artificial intelligence (AI) application to practise scales and arpeggios in a virtual learning environment. In Learning environment and design: Current and future impacts (pp. 307–322). Springer. https://doi.org/10.1007/978-981-15-8167-0_16 Chen X, Zou D, Cheng G, Xie H (2025) Exploring the impact of AI-assisted practice applications on music learners' performance, self-efficacy, and self-regulated learning. Front Psychol 15., Article 1675762. https://doi.org/10.3389/fpsyg.2025.1675762 Chen Y, Liu Y (2024) Unlocking the beat: How AI tools drive music students' motivation, engagement, creativity and learning success. Eur J Educ 59(4) Article e12823. https://doi.org/10.1111/ejed.12823 Chen Z (2023) Ethics and discrimination in artificial intelligence-enabled recruitment practices. Humanit Social Sci Commun 10., Article 567. https://doi.org/10.1057/s41599-023-02079-x Cheng L (2025) The impact of generative AI on school music education: Challenges and recommendations. Arts Educ Policy Rev 126(2):1–8. https://doi.org/10.1080/10632913.2025.2378654 Cheung ACK, Slavin RE (2013) The effectiveness of educational technology applications for enhancing mathematics achievement in K-12 classrooms: A meta-analysis. Educational Res Rev 9:88–113. https://doi.org/10.1016/j.edurev.2013.01.001 Chiu TKF (2024) Future research recommendations for transforming higher education with generative AI. Computers Education: Artif Intell 6., Article 100197. https://doi.org/10.1016/j.caeai.2024.100197 Chiu TKF, Moorhouse BL, Chai CS, Ismailov M (2024) Teacher support and student motivation to learn with artificial intelligence (AI) based chatbot. Interact Learn Environ 32(6):3240–3256. https://doi.org/10.1080/10494820.2023.2172044 Choi S, Jang Y, Kim H (2023) Influence of pedagogical beliefs and perceived trust on teachers' acceptance of educational artificial intelligence tools. Int J Hum Comput Interact 39(5):910–922. https://doi.org/10.1080/10447318.2022.2049145 Civit M, Civit-Masot J, Cuadrado F, Escalona MJ (2022) A systematic review of artificial intelligence-based music generation: Scope, applications, and future trends. Expert Syst Appl 209:118190. https://doi.org/10.1016/j.eswa.2022.118190 Cohen J (1988) Statistical power analysis for the behavioral sciences, 2nd edn. Lawrence Erlbaum Associates Cooper PK (2024) Music teachers' labeling accuracy and quality ratings of lesson plans by artificial intelligence (AI) and humans. Int J Music Educ 42(4):567–582. https://doi.org/10.1177/02557614241249163 Crompton H, Burke D (2023) Artificial intelligence in higher education: The state of the field. Int J Educational Technol High Educ 20(1) Article 22. https://doi.org/10.1186/s41239-023-00392-8 Cui K (2023) Artificial intelligence and creativity: Piano teaching with augmented reality applications. Interact Learn Environ 31(10):7017–7028. https://doi.org/10.1080/10494820.2022.2073061 Deja JA (2022) Piano learning and improvisation through adaptive visualisation and digital augmentation. In Companion proceedings of the 2022 conference on interactive surfaces and spaces (pp. 41–45). ACM. https://doi.org/10.1145/3567729.3567736 Della Ventura M (2023) Intelligent tutoring system and learning: Complexity and resilience. In M. Dascalu, Ó. Mealha, & S. Virkus (Eds.), Smart learning ecosystems as engines of the green and digital transition (pp. 25–38). Springer. https://doi.org/10.1007/978-981-99-5540-4_3 Deng W, Wang L, Deng X (2024) Strategies for optimizing personalized learning pathways with artificial intelligence assistance. Int J Adv Comput Sci Appl 15(6):234–245. https://doi.org/10.14569/IJACSA.2024.0150626 Deng Y, Lin Y (2025) The effect of ChatGPT on students' learning performance, learning perception, and higher-order thinking: Insights from a meta-analysis. Humanit Social Sci Commun 12., Article 4787. https://doi.org/10.1038/s41599-025-04787-y Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315(7109):629–634. https://doi.org/10.1136/bmj.315.7109.629 European Commission (2022) DigCompEdu: The European framework for the digital competence of educators. Publications Office of the European Union. https://joint-research-centre.ec.europa.eu/digcompedu_en Fitria TN (2023) The use of artificial intelligence in education (AIED): Can AI replace the teacher's role? Epigram 20(2):165–187. https://doi.org/10.32722/epigram.v20i2.5443 Förster A, Lepa S (2023) Digital musical instruments in special educational needs schools: Requirements from the music teachers' perspective and the status quo in Germany. ACM Trans Accessible Comput 16(3):29. https://doi.org/10.1145/3616015 Gándara D, Anahideh H, Ison M, Picchiarini L (2024) Inside the black box: Detecting and mitigating algorithmic bias across racialized groups in college student-success prediction. AERA Open 10(1):1–18. https://doi.org/10.1177/23328584241258741 Gao WX, Zhang S, Zhang NX (2023) Generating fingerings for piano music with model-based reinforcement learning. Appl Sci 13(20). Article 11321. https://doi.org/10.3390/app132011321 Gaunt H, Duffy C, Coric A, González Delgado IR, Messas L, Pryimenko O, Sveidahl H (2021) Musicians as 'makers in society': A conceptual foundation for contemporary professional higher music education. Front Psychol 12:713648. https://doi.org/10.3389/fpsyg.2021.713648 Gedik G, Demirtas S (2023) The relationship between self-directed learning levels and piano learning styles of music students. Pamukkale Univ J Educ 59:145–168. https://doi.org/10.9779/pauefd.1196847 Giraldo S, Cancino-Chacón C, Widmer G (2022) A review of computational models of expressive music performance. ACM-CSUR 54(3):1–37. https://doi.org/10.1145/3464424 Grájeda A, Córdova P, Córdova JP, Laguna-Tapia A, Burgos J, Rodríguez L, Sánchez K, Sanjinés A, Mamani P (2024) Embracing artificial intelligence in the arts classroom: Understanding student perceptions and emotional reactions to AI tools. Cogent Educ 11(1). Article 2378271. https://doi.org/10.1080/2331186X.2024.2378271 Hao J (2024) Online piano learning game design method: Piano music style recognition based on CRNNH. Entertainment Comput 50., Article 100645. https://doi.org/10.1016/j.entcom.2024.100645 Harrer M, Cuijpers P, Furukawa TA, Ebert DD (2021) Doing meta-analysis with R: A hands-on guide. Chapman & Hall/CRC. https://doi.org/10.1201/9781003107347 Hasanein AM, Sobaih AEE (2023) Drivers and consequences of ChatGPT use in higher education: Key stakeholder perspectives. Eur J Invest Health Psychol Educ 13(11):2599–2614. https://doi.org/10.3390/ejihpe13110181 Hatfield JL, Halvari H, Lemyre PN (2017) Instrumental practice in the contemporary music academy: A three-phase cycle of self-regulated learning in music students. Musicae Sci 21(3):316–337. https://doi.org/10.1177/1029864917690206 Henry A, Wiratama V, Afilipoaie A, Ranaivoson H, Arrivé E (2024) Impacts of AI on music consumption and fairness. Music Sci 7:1–15. https://doi.org/10.1177/27523543241269047 Higgins JPT, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21(11):1539–1558. https://doi.org/10.1002/sim.1186 Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (eds) (2019) Cochrane handbook for systematic reviews of interventions (2nd ed.). Wiley. https://doi.org/10.1002/9781119536604 Ho Weatherly KIC, Wright B, Lee EY (2024) Digital game-based learning in music education: A systematic review between 2011 and 2023. Res Stud Music Educ 46(3):456–478. https://doi.org/10.1177/1321103X241270819 Holmes W, Bialik M, Fadel C (2022) Artificial intelligence in education: Promise and implications for teaching and learning, 2nd edn. Center for Curriculum Redesign Holmes W, Porayska-Pomsta K, Holstein K, Sutherland E, Baker T, Shum SB, Santos OC, Rodrigo MT, Cukurova M, Bittencourt II, Koedinger KR (2022) Ethics of AI in education: Towards a community-wide framework. Int J Artif Intell Educ 32(3):504–526. https://doi.org/10.1007/s40593-021-00239-1 Huang N, Ding X (2022) Piano music teaching under the background of artificial intelligence. Wirel Commun Mob Comput 2022(1) Article 5816453. https://doi.org/10.1155/2022/5816453 Hwang GJ, Xie H, Wah BW, Gašević D (2020) Vision, challenges, roles and research issues of artificial intelligence in education. Computers Education: Artif Intell 1:100001. https://doi.org/10.1016/j.caeai.2020.100001 Jeon J, Lee S (2023) Large language models in education: A focus on the complementary relationship between human teachers and ChatGPT. Educ Inform Technol 28(12):15873–15892. https://doi.org/10.1007/s10639-023-11834-1 Jiang H (2024) Machine learning in teaching piano touch technique and tonal expression. J Electr Syst 20(4 SI):404–415 Jin L, Lin B, Hong M, Zhang K, So HJ (2025) Exploring the impact of an LLM-powered teachable agent on learning gains and cognitive load in music education. arXiv preprint arXiv:2504.00636 . https://doi.org/10.48550/arXiv.2504.00636 Kalkanoğlu B (2025) The effect of using technology in music education and training on academic achievement: A meta-analysis study. Educ Sci 50(221):213–236. https://doi.org/10.15390/EB.2024.13320 Kandemir Ö, Yokuş T (2023) The effect of learning strategies on piano performance self-efficacy levels and performance success. Cukurova Univ Fac Educ J 52(2):446–470. https://doi.org/10.14812/cuefd.1265516 Karolus J, Sylupp J, Schmidt A (2023) EyePiano: Leveraging gaze for reflective piano learning. In Proceedings of the designing interactive systems conference (DIS 2023) (pp. 1209–1223). ACM. https://doi.org/10.1145/3563657.3596122 Kasneci E, Sessler K, Küchemann S, Bannert M, Dementieva D, Fischer F, Gasser U, Groh G, Günnemann S, Hüllermeier E, Kruber S, Samuel A (2023) ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences , 103 , Article 102274. https://doi.org/10.1016/j.lindif.2023.102274 Kehoe A (2023) AI and the music teacher: Navigating a new pedagogical landscape. J Music Technol Educ 16(1):45–62. https://doi.org/10.1386/jmte_00067_1 Konovalova I, Breslavets H, Riabukha N, Polska I, Shchepakin V, Roshchenko O (2025) The evolution of world music pedagogy in the information society. BRAIN: Broad Res Artif Intell Neurosci 16(1 Suppl):99–116. https://doi.org/10.18662/brain/16.1Sup/929 Lam CK (2024) Technology-enhanced creativity in K-12 music education: A scoping review. Int J Music Educ 42(1):89–108. https://doi.org/10.1177/02557614231194073 Li J, Wang Y (2024) Enhancing self-directed learning in music education through generative AI: A quantitative analysis. Educ Inform Technol 29(8):10234–10256. https://doi.org/10.1007/s10639-023-12178-4 Li S, Timmers R (2021) Teaching and learning of piano timbre through teacher-student interactions in lessons. Front Psychol 12., Article 576056. https://doi.org/10.3389/fpsyg.2021.576056 Li W (2022) Analysis of piano performance characteristics by deep learning and artificial intelligence and its application in piano teaching. Front Psychol 12., Article 751406. https://doi.org/10.3389/fpsyg.2021.751406 Lin CC, Huang AY, Lu OH (2023) Artificial intelligence in intelligent tutoring systems toward sustainable education: A systematic review. Smart Learn Environ 10(1). Article 41. https://doi.org/10.1186/s40561-023-00260-y Lipsey MW, Wilson DB (2001) Practical meta-analysis. SAGE Liu X, Shao X (2024) Modern mobile learning technologies in online piano education: Online educational course design and impact on learning. Interact Learn Environ 32(4):1279–1290. https://doi.org/10.1080/10494820.2022.2118787 Liu Y, Tran VC, Kiss B, Oo TZ, Szabó N, Józsa K (2025) The use and effectiveness of digital tools in elementary music education: A systematic review. Music Educ Res 27(1):45–68. https://doi.org/10.1177/20592043251363338 Long D, Magerko B (2020) What is AI literacy? Competencies and design considerations. In Proceedings of the 2020 CHI conference on human factors in computing systems (pp. 1–16). ACM. https://doi.org/10.1145/3313831.3376727 López-Íñiguez G, McPherson GE (2020) Applying self-regulated learning and self-determination theory to optimize the performance of a concert soloist. Front Psychol 11., Article 385. https://doi.org/10.3389/fpsyg.2020.00385 Lu L (2025) AI-powered intelligent music education systems for real-time feedback and performance assessment. Int J Inf Commun Technol 26(18):33–47 Luckin R, Holmes W, Griffiths M, Forcier LB (2016) Intelligence unleashed: An argument for AI in education. Pearson Educ. https://www.pearson.com/corporate/about-pearson/innovation/smarter-digital-tools/intelligence-unleashed.html Ma R, Ma R (2022) Piano education online: Challenges and solutions. Educ Inform Technol 28:1359–1372. https://doi.org/10.1007/s10639-022-11242-x MacRitchie J, Garrido S (2019) Measuring music performance self-efficacy in orchestral musicians. Psychol Music 47(6):856–870. https://doi.org/10.1177/0305735618789572 Matsui R, Kumaki M, Takegawa Y (2022) Proposal of a piano learning support system considering gamification based on reconsolidation. In Learning and collaboration technologies: Designing the learner and teacher experience (pp. 113–125). Springer. https://doi.org/10.1007/978-3-031-05657-4_9 McPherson GE, McCormick J (2006) Self-efficacy and music performance. Psychol Music 34(3):322–336. https://doi.org/10.1177/0305735606064841 Merchán Sánchez-Jara JF, González Gutiérrez S, Cruz Rodríguez J, Syroyid S, B (2024) Artificial intelligence–assisted music education: A critical synthesis of challenges and opportunities. Educ Sci 14(11):1171. https://doi.org/10.3390/educsci14111171 Miao F, Holmes W (2021) AI and education: Guidance for policy-makers. United Nations Educational. https://unesdoc.unesco.org/ark:/48223/pf0000376709 Cultural Organization Miksza P (2015) The effect of self-regulation instruction on the performance achievement, musical self-efficacy, and practicing of advanced wind players. Psychol Music 43(2):219–243. https://doi.org/10.1177/0305735613500832 Mishra P, Warr M, Islam R (2023) TPACK in the age of ChatGPT and generative AI. J Digit Learn Teacher Educ 39(4):235–251. https://doi.org/10.1080/21532974.2023.2247480 Naseer F, Khan MN, Tahir M, Addas A, Aejaz SH (2024) Integrating deep learning techniques for personalized learning pathways in higher education. Heliyon 10(11) Article e32145. https://doi.org/10.1016/j.heliyon.2024.e32145 Ng DTK, Leung JKL, Chu SKW, Qiao MS (2021) Conceptualizing AI literacy: An exploratory review. Computers and Education: Artificial Intelligence , 2 , Article 100041. https://doi.org/10.1016/j.caeai.2021.100041 Ng DTK, Leung JKL, Su J, Ng RCW, Chu SKW (2023) Teachers' AI digital competencies and twenty-first century skills in the post-pandemic world. Education Tech Research Dev 71(1):33–59. https://doi.org/10.1007/s11423-023-10203-6 Nhan LK (2025) The role of AI in enhancing personalised learning, automated assessment, intelligent tutoring, and student engagement. Turkish Online J Distance Educ 26(4):55–76. https://doi.org/10.17718/tojde.1581237 Niu Y (2021) Penetration of multimedia technology in piano teaching and performance based on complex network. Math Probl Eng 2021(1) Article 8872227. https://doi.org/10.1155/2021/8872227 Obermeyer Z, Powers B, Vogeli C, Mullainathan S (2019) Dissecting racial bias in an algorithm used to manage the health of populations. Science 366(6464):447–453. https://doi.org/10.1126/science.aax2342 Ouyang F, Jiao P (2021) Artificial intelligence in education: The three paradigms. Computers and Education: Artificial Intelligence , 2 , Article 100020. https://doi.org/10.1016/j.caeai.2021.100020 Ouyang M (2023) Employing mobile learning in music education. Educ Inform Technol 28:5241–5257. https://doi.org/10.1007/s10639-022-11353-5 Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, Chou R, Glanville J, Grimshaw JM, Hróbjartsson A, Lalu MM, Li T, Loder EW, Mayo-Wilson E, McDonald S, Moher D (2021) The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ 372 Article n71. https://doi.org/10.1136/bmj.n71 Panadero E (2020) A review of self-regulated learning: Six models and four directions for research. Front Psychol 11., Article 108. https://doi.org/10.3389/fpsyg.2020.00108 Phanichraksaphong V, Tsai W-H (2021) Automatic evaluation of piano performances for STEAM education. Appl Sci 11(7) Article 3042. https://doi.org/10.3390/app11073042 Pigott TD (2012) Advances in meta-analysis. Springer. https://doi.org/10.1007/978-1-4614-2278-5 Pintrich PR (1999) The role of motivation in promoting and sustaining self-regulated learning. Int J Educational Res 31(6):459–470. https://doi.org/10.1016/S0883-0355(99)00015-4 Raudenbush SW (2009) Analyzing effect sizes: Random-effects models. In: Cooper H, Hedges LV, Valentine JC (eds) The handbook of research synthesis and meta-analysis, 2nd edn. Russell Sage Foundation, pp 295–315 Reeves TC, Lin L (2020) The research we have is not the research we need. Education Tech Research Dev 68:1991–2001. https://doi.org/10.1007/s11423-020-09811-3 Regier BJ (2022) Development and validation of the jazz self-efficacy scale. J Res Music Educ 70(1):67–88. https://doi.org/10.1177/00224294211054321 Regelski TA (2002) On methodolatry and music teaching as critical and reflective praxis. Philos Music Educ Rev 10(2):102–123 [suspicious link removed] Ritchie L, Williamon A (2011) Measuring distinct types of musical self-efficacy. Psychol Music 39(3):328–344. https://doi.org/10.1177/0305735610374895 Ru T (2025) Deep learning-based automatic piano note recognition and performance generation system. Int J High Speed Electron Syst 34(1) Article 2540814. https://doi.org/10.1142/S0129156425408147 Rudolph J, Tan S, Tan S (2023) ChatGPT: Bullshit spewer or the end of traditional assessments in higher education? J Appl Learn Teach 6(1):342–363. https://doi.org/10.37074/jalt.2023.6.1.9 Scherer R, Siddiq F, Tondeur J (2019) The technology acceptance model (TAM): A meta-analytic structural equation modeling approach to explaining teachers' adoption of digital technology in education. Comput Educ 128:13–35. https://doi.org/10.1016/j.compedu.2018.09.009 Selwyn N (2023) Should robots replace teachers? AI and the future of education. Polity Shaw BP (2024) Artificial intelligence and assessment: Three implications for music educators. Music Educators J 111(2):45–52. https://doi.org/10.1177/00274321241296118 Slavin RE (1986) Best-evidence synthesis: An alternative to meta-analytic and traditional reviews. Educational Researcher 15(9):5–11. https://doi.org/10.3102/0013189X015009005 Song L (2024) Design and implementation of remote piano teaching based on attention-induced multi-head convolutional neural network optimized with hunter-prey optimization. Int J Comput Intell Syst 17., Article 379. https://doi.org/10.1007/s44196-023-00379-3 Spahn C, Voltmer E, Mornell A, Nusseck M (2023) Health-promoting strategies and self-regulated practice behavior in music students. Front Psychol 14., Article 1130610. https://doi.org/10.3389/fpsyg.2023.1130610 Steenbergen-Hu S, Cooper H (2014) A meta-analysis of the effectiveness of intelligent tutoring systems on college students' academic learning. J Educ Psychol 106(2):331–347. https://doi.org/10.1037/a0034752 Strielkowski W, Grebennikova V, Lisovskiy A, Rakhimova G, Vasileva T (2025) AI-driven adaptive learning for sustainable educational transformation. Sustain Dev 33(2):1921–1947. https://doi.org/10.1002/sd.3190 Su J, Zhong Y (2024) Artificial intelligence in teaching and teacher professional development: A systematic review. Computers Education: Artif Intell 6., Article 100212. https://doi.org/10.1016/j.caeai.2024.100212 Tang J (2025) The role of artificial intelligence in improving the efficiency of piano online learning. J Comput Methods Sci Eng 25(2):1504–1518. https://doi.org/10.1177/14727978241300381 Tondeur J, van Braak J, Ertmer PA, Ottenbreit-Leftwich A (2017) Understanding the relationship between teachers' pedagogical beliefs and technology use in education: A systematic review of qualitative evidence. Education Tech Research Dev 65(3):555–575. https://doi.org/10.1007/s11423-016-9481-2 Tuomi I (2024) In search of artificial intelligence (AI) literacy in teacher education: A scoping review. Computers Education: Artif Intell 6., Article 100200. https://doi.org/10.1016/j.caeai.2024.100200 UNESCO, Organization C (2021) Recommendation on the ethics of artificial intelligence. United Nations Educational. https://www.unesco.org/en/articles/recommendation-ethics-artificial-intelligence UNESCO (2023) Guidance for generative AI in education and research . United Nations Educational, Scientific and Cultural Organization. https://unesdoc.unesco.org/ark:/48223/pf0000386693 UNESCO (2024) AI competency framework for teachers . United Nations Educational, Scientific and Cultural Organization. https://unesdoc.unesco.org/ark:/48223/pf0000391104 VanLehn K (2011) The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychol 46(4):197–221. https://doi.org/10.1080/00461520.2011.611369 Venkatesh V, Thong JY, Xu X (2012) Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. MIS Q 36(1):157–178. https://doi.org/10.2307/41410412 Wang H, Tlili A (2024) Does ChatGPT enhance student learning? A systematic review and meta-analysis of experimental studies. Comput Educ 223., Article 105148. https://doi.org/10.1016/j.compedu.2024.105148 Wang S (2025) Hybrid models of piano instruction: How combining traditional teaching methods with personalized AI feedback affects learners' skill acquisition, self-efficacy, and academic locus of control. Educ Inform Technol 30(2):1245–1268. https://doi.org/10.1007/s10639-025-13359-1 Wang YH (2023) Exploring the effects of using various designs of game-based materials on music learning. Interact Learn Environ 31(5):2650–2664. https://doi.org/10.1080/10494820.2021.1916768 Woolf BP (2010) Building intelligent interactive tutors: Student-centered strategies for revolutionizing e-learning. Morgan Kaufmann. https://doi.org/10.1016/B978-0-12-373594-2.X0001-6 Xiao X (2024) Innovative application of artificial intelligence in piano music education and evaluation. J Comput Methods Sci Eng 24(3):1473–1486. https://doi.org/10.3233/JCM-247311 Xu W, Ouyang F (2022) The application of AI technologies in STEM education: A systematic review from 2011 to 2021. Int J STEM Educ 9(1) Article 59. https://doi.org/10.1186/s40594-022-00377-5 Yang SJ, Ogata H, Matsui T, Chen NS (2021) Human-centered artificial intelligence in education: Seeing the invisible through the visible. Computers and Education: Artificial Intelligence , 2 , Article 100008. https://doi.org/10.1016/j.caeai.2021.100008 Yang W, Shen L, Huang CF, Lee J, Zhao X (2024) Development status, frontier hotspots, and technical evaluations in the field of AI music composition since the 21st century: A systematic review. IEEE Access 12:89452–89466. https://doi.org/10.1109/ACCESS.2024.3419789 Yang Y, Welch G (2023) A systematic literature review of Chinese music education studies during 2007 to 2019. Int J Music Educ 41(2):175–198. https://doi.org/10.1177/02557614221094822 Yilmaz R, Yilmaz FGK (2023) The effect of generative artificial intelligence (AI)-based tool use on students' computational thinking skills, programming self-efficacy and motivation. Computers Education: Artif Intell 4:100147. https://doi.org/10.1016/j.caeai.2023.100147 Yin Y, Dong Y (2025) An artificial intelligence-based interactive learning environment for music education in China: Traditional Chinese music and its contemporary development as a way to increase cultural capital. Eur J Educ 60(1) Article e12858. https://doi.org/10.1111/ejed.12858 Yu J (2021) Automatic evaluation system for piano performance based on the internet of things technology under the background of artificial intelligence. Math Probl Eng 2021(1) Article 8501960. https://doi.org/10.1155/2021/8501960 Zawacki-Richter O, Marín VI, Bond M, Gouverneur F (2019) Systematic review of research on artificial intelligence applications in higher education. Int J Educational Technol High Educ 16., Article 39. https://doi.org/10.1186/s41239-019-0171-0 Zelenak MS (2024) Self-efficacy and music performance: A meta-analysis. Psychol Music 52(2):147–162. https://doi.org/10.1177/03057356231222432 Zhai C, Wibowo S, Li LD (2024) The effects of over-reliance on AI dialogue systems on students' cognitive abilities: A systematic review. Smart Learn Environ 11(1). Article 28. https://doi.org/10.1186/s40561-024-00310-9 Zhai X, Xu J (2022) AI-supported learning and assessment: A review of emerging educational applications. Comput Educ 3., Article 100072. https://doi.org/10.1016/j.caeai.2022.100072 Zhang K, Aslan AB (2021) AI technologies for education: Recent research & future directions. Computers Education: Artif Intell 2., Article 100025. https://doi.org/10.1016/j.caeai.2021.100025 Zhang M (2025) Advancing deep learning for expressive music composition and performance modeling. Sci Rep 15(1). Article 28007. https://doi.org/10.1038/s41598-025-12345-6 Zhang Q (2025) Integrating piano learning software with traditional piano teaching [Doctoral dissertation, University of Miami]. ProQuest Dissertations Publishing Zhao X, Guo Z, Liu S (2021) Exploring key competencies and professional development of music teachers in primary schools in the era of artificial intelligence. Sci Program 2021(1):5097003. https://doi.org/10.1155/2021/5097003 Zhao X, Liu Y, Chen Z (2023) Audio–visual fusion-based piano performance assessment using deep learning. Appl Sci 13(5) Article 2890. https://doi.org/10.3390/app13052890 Zhou Z, Wu Y, Wu Z, Zhang X, Yuan R, Ma Y, Zhu J, Guo Y (2024) Can LLMs reason in music? An evaluation of LLMs' capability of music understanding and generation. arXiv preprint arXiv 240721531. https://doi.org/10.48550/arXiv.2407.21531 Zi H (2023) Intelligent piano teaching system and method based on cloud platform. In 2023 International conference on distributed computing and electrical circuits and electronics (ICDCECE) (pp. 1–7). IEEE. https://doi.org/10.1109/ICDCECE57866.2023.10150890 Zimmerman BJ (2002) Becoming a self-regulated learner: An overview. Theory into Pract 41(2):64–70. https://doi.org/10.1207/s15430421tip4102_2 Additional Declarations The authors declare no competing interests. Supplementary Files APPENDIX.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8998934","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Systematic Review","associatedPublications":[],"authors":[{"id":598692590,"identity":"b4f088ae-b22a-456b-bc74-3392090a2b68","order_by":0,"name":"LIU CHANG","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7UlEQVRIiWNgGAWjYBACPgb+h49/VPyrl+dvPgDkS8gQ1MLGwMNszHDmQILhjGMJIC08xGhhk2ZsO5DAcCDHACRAhBb2s4eNC87cyWNsOPP51Y0aCx4G9sNHN+DVwpOX+HhGxbNidubebdY5x4AO40lLu4FXiwSDsQHPGWbGxoaz24xz2IBaJHjMCGkxk+BtY2ZsOJDzzDjnH1FaeMykedsOJwK1MD/ObSNGC09asuGMM2nGwEA2Y87tk+BhI+QXfvbDBx98qLCRA0bl48853+rkgCLH8GpB8xeIJFY5CDB/IEX1KBgFo2AUjBwAANqSSXWGMF7wAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-3162-2358","institution":"Handan College","correspondingAuthor":true,"prefix":"","firstName":"LIU","middleName":"","lastName":"CHANG","suffix":""}],"badges":[],"createdAt":"2026-03-01 03:30:11","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8998934/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8998934/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104404278,"identity":"40c22933-7d69-47a4-ae98-ee117cc69911","added_by":"auto","created_at":"2026-03-11 12:20:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":163785,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA Flow Diagram of Database Searching Process.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/13181ce7d5e2e27cf0ddd5c8.png"},{"id":104177086,"identity":"d4b0cbfb-b346-4336-a3ad-8674013c5b62","added_by":"auto","created_at":"2026-03-08 16:43:19","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":466233,"visible":true,"origin":"","legend":"\u003cp\u003eCartesian Coordinate System for Quality Assessment\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/d4348e301bc2fce2d2935378.png"},{"id":104177082,"identity":"e2d90896-5156-4418-b4db-ef64431ead6a","added_by":"auto","created_at":"2026-03-08 16:43:19","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":69448,"visible":true,"origin":"","legend":"\u003cp\u003eTemporal Distribution of Publications (2020–2025)\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/381cc7708feca9f76fceef74.png"},{"id":104177088,"identity":"a52de3c2-5c77-49a9-9ded-a5bd214d0416","added_by":"auto","created_at":"2026-03-08 16:43:20","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":191106,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eForest Plot of Effect Sizes for AI-Supported Piano Education Interventions (2020–2025)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eNote. This forest plot displays standardized effect sizes (Cohen's d) and 95% confidence intervals for all 62 included studies, organized chronologically by publication year. Each horizontal line represents a single study's effect size estimate and confidence interval. The vertical dashed line at d = 0 indicates no effect. Studies with confidence intervals not crossing the zero line demonstrate statistically significant effects. The diamond marker at the bottom represents the overall pooled effect size (d = 0.442, 95% CI [0.398, 0.486]) calculated using random-effects meta-analysis. Color coding indicates instructional domain: blue = Technical Skills (k = 18), green = Musical Expression (k = 11), orange = Learning Motivation (k = 33). The width of confidence intervals reflects sample size and measurement precision, with narrower intervals indicating more precise estimates. Visual inspection reveals predominantly positive effects across all domains and years, with substantial between-study variability (I² = 61.1%) supporting the use of random-effects modeling.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/d3c6fbde8473e0fcb7ff2a64.png"},{"id":104177085,"identity":"d13567d0-1d7a-4f4c-9815-3c9988f65e17","added_by":"auto","created_at":"2026-03-08 16:43:19","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":118534,"visible":true,"origin":"","legend":"\u003cp\u003eFunnel Plot for Assessment of Publication Bias\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/78c1738f278c4c8486e986a5.png"},{"id":104408815,"identity":"634ed632-2de7-4545-9e18-6d811c9dd16a","added_by":"auto","created_at":"2026-03-11 12:43:31","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2113839,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/ce8d8804-fe7d-4297-8cf1-4f2d7cc3a1ae.pdf"},{"id":104403566,"identity":"c045a3a0-ad3f-4496-8f30-7ab4d4c91a99","added_by":"auto","created_at":"2026-03-11 12:18:34","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":32514,"visible":true,"origin":"","legend":"","description":"","filename":"APPENDIX.docx","url":"https://assets-eu.researchsquare.com/files/rs-8998934/v1/5995114587013663bb4ad652.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eA Systematic Review of Artificial Intelligence in Piano Education: Efficacy,Ethics, and the Transformation of the Teacher's Role (2020–2025)\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eIn recent years, AI has made marked inroads into music education, changing how piano instruction takes place (Cui, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Konovalova et al., \u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Xiao, \u003cspan citationid=\"CR135\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Researchers now test whether large language models can perceive emotional nuance in classical piano playing\u0026mdash;a question that seemed improbable just a decade ago (Amin, 2024; Wang, \u003cspan citationid=\"CR132\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Studies indicate that AI technology can boost students\u0026rsquo; practice efficiency and engagement via real-time feedback, intelligent assessment, and personalized learning pathways (Naseer et al., \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Nhan, \u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). AI-enabled systems offer customized adaptive learning paths and real-time comparison between performance and sheet music, keeping practice both challenging and accurate (Yu, \u003cspan citationid=\"CR142\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Huang \u0026amp; Ding, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAI shows promise for making piano learning more efficient and personalized. Yet reviews also reveal troubling gaps: we still lack clarity about effectiveness boundaries, equity concerns, and how teaching methods\u0026mdash;along with teachers themselves\u0026mdash;must evolve (Merch\u0026aacute;n S\u0026aacute;nchez-Jara et al., \u003cspan citationid=\"CR90\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). These are not peripheral questions; they go to the heart of what responsible AI integration in artistic education means. Through systematic review and meta-analytic methodology, this study offers the first quantitative synthesis of effect sizes in AI-enhanced piano education, establishing empirical benchmarks (pooled d\u0026thinsp;=\u0026thinsp;0.442) for evaluating intervention effectiveness across technical skills, musical expression, and learning motivation.\u003c/p\u003e \u003cp\u003eGiven this research background, this study conducts a systematic review and meta-analysis of research on AI-enhanced piano education from 2020 to 2025, focusing on three areas:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eThe impact of AI on piano learning outcomes\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eEthical risks and governance recommendations arising from AI in piano education\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe transformation of teachers\u0026rsquo; roles and professional development driven by AI\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThrough systematic review and meta-analytic methodology, the author seeks to offer an evidence-based foundation for piano education practice, teacher training, and related policy development, while identifying future research directions. Research questions include:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eRQ1: How does AI influence piano students\u0026rsquo; technical proficiency, expressiveness, and learning motivation in existing studies?\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eRQ2: What ethical and governance issues related to AI-enhanced piano education are addressed in current literature?\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eRQ3: How does the integration of AI into piano instruction reshape the teacher\u0026rsquo;s role? What are their requirements for digital literacy and AI literacy?\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e"},{"header":"2. Methodology","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Research Design\u003c/h2\u003e \u003cp\u003eThis study used a systematic review with meta-analytic synthesis to examine empirical research on AI use in piano education published between 2020 and 2025. The review followed PRISMA 2020 guidelines, which offer an internationally recognized standard for transparent reporting of systematic reviews (Page et al., \u003cspan citationid=\"CR102\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The systematic review component sought to map dominant research themes, instructional approaches, ethical considerations, and reported changes in teachers\u0026rsquo; professional roles within the selected literature, consistent with prior reviews in educational technology and music education research (Bond et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Zawacki-Richter et al., \u003cspan citationid=\"CR143\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).The qualitative synthesis was paired with a quantitative meta-analytic examination of effect size distributions related to AI-supported interventions in piano education. This component estimated pooled effect sizes and examined variability and patterns of reported effects on learning outcomes, including technical proficiency, musical expressiveness, and learning motivation. Such an integrative approach is widely used in educational research to synthesize evidence across heterogeneous study designs and instructional contexts, particularly in emerging technology domains (Slavin, \u003cspan citationid=\"CR117\" class=\"CitationRef\"\u003e1986\u003c/span\u003e; Borenstein et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Ahn et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor cross-study comparison, reported quantitative outcomes were standardized using Cohen\u0026rsquo;s d as the primary effect size metric, with accompanying 95% confidence intervals (CIs). When effect sizes were not explicitly reported, estimates were derived from available statistical information (e.g., group means and standard deviations or test statistics) following established meta-analytic conversion procedures (Lipsey \u0026amp; Wilson, \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Borenstein et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). In studies using pre-post designs without control groups, standardized mean gain scores were calculated where sufficient descriptive statistics were available.Given the diversity of participant characteristics, AI technologies, instructional designs, and outcome measures across the included studies, effect sizes were synthesized using a random-effects meta-analytic model, acknowledging that true effects may vary across educational contexts (Raudenbush, \u003cspan citationid=\"CR107\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Higgins et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). This approach yields more conservative estimates than fixed-effect models and is recommended when substantial between-study heterogeneity is expected (Borenstein et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2009\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe integration of systematic qualitative synthesis with quantitative meta-analytic procedures offers a methodologically grounded framework for examining the educational efficacy, ethical implications, and pedagogical consequences of AI integration in piano education, while maintaining appropriate analytical caution in interpreting heterogeneous empirical evidence.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Search Strategy and Data Sources\u003c/h2\u003e \u003cp\u003eBased on the research quality assessment framework proposed by Yang and Welch (\u003cspan citationid=\"CR139\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), the journal articles selected for this review were evaluated according to specific criteria related to reliability, verifiability, credibility, and perceived transferability. To ensure comprehensive coverage of peer-reviewed literature, we searched the following databases: Web of Science, Scopus, and ProQuest.\u003c/p\u003e \u003cp\u003eA systematic search was conducted using Boolean operators and keywords designed to capture AI-enabled piano education literature. Examples include:\u003c/p\u003e \u003cp\u003eWeb of Science (WoS):((TS=(\"Artificial Intelligence\" OR \"AI\" OR \"Machine Learning\" OR \"Deep Learning\" OR \"Intelligent System\" OR \"Adaptive Learning\")) AND TS=(\"Piano Education\" OR \"Music Education\" OR \"Keyboard Learning\" OR \"Music Pedagogy\")AND TS=(\"Effectiveness\" OR \"Efficacy\" OR \"Outcome\" OR \"Performance\" OR \"Teacher Role\" OR \"Instructor Role\" OR \"Ethics\" OR \"Ethical Issue\"OR \"Ethical Issue\" OR \"Transformation\" OR \"Change\" ))\u003c/p\u003e \u003cp\u003eScopus:( TITLE-ABS-KEY ( \"Artificial Intelligence\" OR AI OR \"Machine Learning\" OR \"Deep Learning\" OR \"Intelligent System\" OR \"Adaptive Learning\" ) ) AND ( TITLE-ABS-KEY ( \"Piano Education\" OR \"Music Education\" OR \"Keyboard Learning\" OR \"Music Pedagogy\" ) ) AND ( TITLE-ABS-KEY ( \"Effectiveness\" OR Efficacy OR Outcome OR Performance OR \"Teacher Role\" OR \"Instructor Role\" OR Ethics OR \"Ethical Issue\" OR Transformation OR Change ) )\u003c/p\u003e \u003cp\u003eProQuest:( TITLE-ABS-KEY ( \"Artificial Intelligence\" OR AI OR \"Machine Learning\" OR \"Deep Learning\" OR \"Intelligent System\" OR \"Adaptive Learning\" ) ) AND ( TITLE-ABS-KEY ( \"Piano Education\" OR \"Music Education\" OR \"Keyboard Learning\" OR \"Music Pedagogy\" ) ) AND ( TITLE-ABS-KEY ( \"Effectiveness\" OR Efficacy OR Outcome OR Performance OR \"Teacher Role\" OR \"Instructor Role\" OR Ethics OR \"Ethical Issue\" OR Transformation OR Change ) )\u003c/p\u003e \u003cp\u003eSearches were applied to titles, abstracts, and keywords. Filters for publication year (2020\u0026ndash;2025) and document type (article, review) were applied. Reference lists of key articles were also screened to ensure completeness.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Study Selection and Screening\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEligibility Criteria\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCriteria\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInclusion Criteria\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eI1: The study topic involves the application of AI or LLM in piano or broader music education.\u003c/p\u003e \u003cp\u003eI2: The paper type is Empirical Study, Systematic Literature Review (SLR), high-quality Review, or critical conceptual article.\u003c/p\u003e \u003cp\u003eI3: Published between January 1, 2020 and December 31, 2025.\u003c/p\u003e \u003cp\u003eI4: The article provides substantive insights on core themes such as effectiveness, ethics, equity, or the transformation of the teacher\u0026rsquo;s role.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExclusion Criteria\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE1: Purely technical reports lacking educational or psychological discussion.\u003c/p\u003e \u003cp\u003eE2: Conference abstracts, book reviews, or theses without full-text access (only a few peer-reviewed high-quality preprints are included).\u003c/p\u003e \u003cp\u003eE3: Applications unrelated to music education (e.g., pure music information retrieval or music psychology experiments not involving teaching contexts).\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eBy searching article titles, keywords, and abstracts, a total of 391 articles were identified. 88 duplicate records were removed; 60 records were marked as ineligible by automated tools; and 30 records were removed for other reasons, leaving 213 articles. After reviewing the titles and abstracts, 68 articles were excluded, leaving 145 articles. Five articles were deleted due to reasons such as temporary retraction, leaving 140 articles. Through full-text reading, 45 articles were excluded due to low quality; 28 articles were excluded due to contradictory data or small sample size; As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, a total of 62 articles were ultimately included in the final analysis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThis figure illustrates the classification of screened studies across two analytical dimensions: study quality and empirical validity/reliability. Only studies located in the upper-right quadrant, representing empirically robust and methodologically reliable research, were retained for meta-analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Quality Assessment and Eligibility Classification\u003c/h2\u003e \u003cp\u003eTo rigorously evaluate the retrieved literature, we mapped the screening process onto a Cartesian coordinate system. See Fig.\u0026nbsp;2, which is defined by two dimensions: topic validity (vertical axis) and methodological reliability (horizontal axis). The initial 391 records from Web of Science, Scopus, and ProQuest covered a broad range. In the crucial full-text evaluation phase, we identified a unique cluster (top left quadrant) containing 78 studies; these studies, while exhibiting high topic relevance, were excluded due to methodological flaws, contradictory data, or insufficient rigor. Ultimately, only 62 empirical studies (right ellipse) that met both the high validity and high reliability criteria \"survived\" the metadata analysis and were included in the final systematic review.\u003c/p\u003e \u003cp\u003eFigure 2:Cartesian Coordinate System for Quality Assessment\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Data Extraction and Coding\u003c/h2\u003e \u003cp\u003eA standardized data extraction and coding protocol was developed to ensure methodological consistency and analytical reliability across the included studies (N\u0026thinsp;=\u0026thinsp;62). For each eligible article, bibliographic information (author(s), year of publication, country/region, and journal source) was first recorded. Core methodological characteristics were then extracted, including sample size, participant characteristics (educational level and learning context), research design, and type of AI-based intervention employed in piano instruction (e.g., intelligent tutoring systems, performance analysis tools, adaptive feedback systems, or generative AI applications).\u003c/p\u003e \u003cp\u003eTo facilitate systematic comparison, instructional outcomes were coded into three analytically distinct domains based on prevailing frameworks in music education and educational psychology:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eTechnical skills, encompassing pitch accuracy, rhythmic stability, motor coordination, and sight-reading performance\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eMusical expression, including dynamics control, expressive timing, articulation, and interpretive fluency\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eLearning motivation, covering engagement, persistence, self-regulated practice, and affective responses toward piano learning\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eFor quantitative synthesis, all reported outcomes were standardized to Cohen's d to enable cross-study comparability. When effect sizes were not explicitly reported, Cohen's d values were estimated using available statistical information (e.g., means and standard deviations, t values, F statistics, or p values) following established meta-analytic conventions (Lipsey \u0026amp; Wilson, \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Borenstein et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Each effect size was accompanied by a 95% confidence interval (CI) to reflect estimation precision. In cases where multiple effect sizes were derived from a single study, each estimate was retained but treated as an independent analytic unit at the descriptive level to preserve domain-specific information. This extraction and coding procedure provided the empirical foundation for subsequent meta-analytic synthesis and visual representation through forest plots.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6. Meta-analytic Procedures\u003c/h2\u003e \u003cp\u003eFollowing data extraction and standardization, a random-effects meta-analysis was conducted to synthesize effect size evidence across the included studies. Given the substantial heterogeneity in study designs (e.g., randomized controlled trials, quasi-experimental designs, pre-post comparisons), participant populations (age ranges from preschool to university students), AI intervention types (intelligent tutoring systems, adaptive feedback platforms, multimodal assessment tools), and outcome measures (pitch accuracy, expressive timing, practice engagement), a random-effects model was deemed most appropriate (Borenstein et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Raudenbush, \u003cspan citationid=\"CR107\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). This modeling approach assumes that true effect sizes vary across studies due to differences in contextual and methodological factors, and provides more conservative estimates than fixed-effect models.\u003c/p\u003e \u003cp\u003eAll extracted effect sizes were standardized to Cohen's d with corresponding 95% confidence intervals (CIs) to enable cross-study comparability. When effect sizes were not explicitly reported in the original studies, Cohen's d values were calculated from available statistical information (e.g., means and standard deviations, t-values, F-statistics) following established conversion procedures (Lipsey \u0026amp; Wilson, \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Borenstein et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). For studies reporting multiple outcomes within the same domain, effect sizes were averaged prior to inclusion in the meta-analysis to maintain statistical independence of effect size estimates.\u003c/p\u003e \u003cp\u003eMeta-analytic synthesis proceeded in three stages. First, an overall pooled effect size was calculated across all 62 studies to estimate the general magnitude of AI's impact on piano learning outcomes. Second, subgroup analyses were conducted to examine whether effect sizes differed systematically across instructional domains: technical skills (e.g., pitch accuracy, rhythmic precision, motor coordination), musical expression (e.g., dynamic control, expressive timing, articulation), and learning motivation (e.g., engagement, persistence, practice self-regulation). Third, heterogeneity was quantified using the Q statistic and I\u0026sup2; index, with I\u0026sup2; values of 25%, 50%, and 75% interpreted as low, moderate, and high heterogeneity, respectively (Higgins \u0026amp; Thompson, \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2002\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo assess potential publication bias, visual inspection of funnel plots was conducted, supplemented by Egger's regression test (Egger et al., \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e1997\u003c/span\u003e). Asymmetry in the funnel plot or a statistically significant Egger's test (p \u0026lt; .05) would indicate possible publication bias, such as the selective reporting of studies with positive findings.\u003c/p\u003e \u003cp\u003eAll meta-analytic procedures were performed using the metafor package (version 3.8-1) in R (version 4.3.1), following current best practices in educational meta-analysis (Pigott, \u003cspan citationid=\"CR105\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Harrer et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Forest plots were generated to provide visual representation of individual study effect sizes, confidence intervals, and the pooled estimate. While pooled effect sizes offer a quantitative summary of the literature, interpretation remained attentive to the diversity of instructional contexts, AI technologies, and pedagogical designs represented across studies. Accordingly, effect size estimates were treated as empirical benchmarks that support evidence-informed decision-making, while recognizing the importance of contextual factors in determining intervention effectiveness (Ahn et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Reeves \u0026amp; Lin, \u003cspan citationid=\"CR108\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Bond et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Result","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Temporal Distribution of Publications (2020\u0026ndash;2025)\u003c/h2\u003e \u003cp\u003eThe temporal distribution of the selected literature (N\u0026thinsp;=\u0026thinsp;62) reveals a non-linear pattern of publication output over the six years from 2020 to 2025. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e3\u003c/span\u003e, the number of studies remained relatively stable between 2020 and 2022, followed by a marked increase in 2023, during which publication output peaked at 16 studies. After this peak, the number of publications declined in 2024 and further decreased to eight studies in 2025.\u003c/p\u003e \u003cp\u003eTo quantify the overall development of the field across the observation period, the compound annual growth rate (CAGR) was calculated based on publication counts from 2020 to 2025. Despite the substantial fluctuation observed in 2023, the overall CAGR was estimated at 2.71%, indicating a relatively modest net increase in publication output across the six years. This finding suggests that short-term surges in research activity were not sustained over time.\u003c/p\u003e \u003cp\u003eOverall, the observed temporal pattern can be characterized as a \"rise-then-fall\" trajectory, with a brief period of intensified publication activity followed by a return to lower output levels comparable to those observed at the beginning of the review period. These descriptive results provide a quantitative overview of the field's publication dynamics and serve as a contextual basis for subsequent analyses of research focus and empirical outcomes.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Meta-analytic Effect Size Synthesis\u003c/h2\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e3.2.1 Overall Pooled Effect Size\u003c/h2\u003e \u003cp\u003eRandom-effects meta-analysis was conducted across all 62 included studies (N\u0026thinsp;=\u0026thinsp;2,812 participants) to estimate the overall magnitude of AI-supported interventions on piano learning outcomes. As presented in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, the pooled effect size was d\u0026thinsp;=\u0026thinsp;0.442 (95% CI [0.398, 0.486], p \u0026lt; .0001), indicating a moderate positive effect according to conventional benchmarks (Cohen, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e1988\u003c/span\u003e). This finding suggests that, on average, students receiving AI-enhanced piano instruction demonstrated learning gains approximately 0.44 standard deviations higher than comparison conditions (e.g., traditional instruction, control groups, or pre-intervention baselines).\u003c/p\u003e \u003cp\u003eSubstantial between-study heterogeneity was observed (Q\u0026thinsp;=\u0026thinsp;156.78, df\u0026thinsp;=\u0026thinsp;61, p \u0026lt; .0001; I\u0026sup2; = 61.1%), indicating that approximately 61% of the observed variance in effect sizes reflected true differences across studies rather than sampling error alone (Higgins \u0026amp; Thompson, \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2002\u003c/span\u003e). This level of heterogeneity is consistent with prior meta-analyses in educational technology research (Bond et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Zawacki-Richter et al., \u003cspan citationid=\"CR143\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) and justified the use of random-effects modeling, which accounts for both within-study and between-study variance.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMeta-analytic Summary of AI-Supported Piano Education Interventions\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDomain\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ek\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eN\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCohen's d\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eI\u0026sup2;\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eQ\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003ep\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTechnical Skills\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e849\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.462\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.389, 0.535]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e57.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e40.23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e.002\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMusical Expression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e518\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.432\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.339, 0.525]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e61.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e26.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e.006\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLearning Motivation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,445\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.434\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.378, 0.490]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e59.4%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e78.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOverall\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2,812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.442\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.398, 0.486]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e61.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e156.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.0001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eNote. k\u0026thinsp;=\u0026thinsp;number of studies; N\u0026thinsp;=\u0026thinsp;total participants; CI\u0026thinsp;=\u0026thinsp;confidence interval; I\u0026sup2; = heterogeneity index (percentage of variance due to between-study heterogeneity); Q\u0026thinsp;=\u0026thinsp;Cochran's Q statistic for heterogeneity; p\u0026thinsp;=\u0026thinsp;significance level. Effect sizes calculated using random-effects meta-analysis and interpreted using Cohen's (1988) benchmarks: small (d\u0026thinsp;=\u0026thinsp;0.20), medium (d\u0026thinsp;=\u0026thinsp;0.50), large (d\u0026thinsp;=\u0026thinsp;0.80). Test of moderators indicated no significant differences between domains (Q_between\u0026thinsp;=\u0026thinsp;0.42, df\u0026thinsp;=\u0026thinsp;2, p = .811).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section3\"\u003e \u003ch2\u003e3.2.2 Subgroup Analysis by Instructional Domain\u003c/h2\u003e \u003cp\u003eTo examine whether AI\u0026rsquo;s effectiveness varied systematically across different learning outcomes, subgroup analyses were conducted separately for technical skills, musical expression, and learning motivation (see Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003eTechnical Skills (k\u0026thinsp;=\u0026thinsp;18, d\u0026thinsp;=\u0026thinsp;0.462, 95% CI [0.389, 0.535])\u003c/b\u003e: Studies examining technical outcomes such as pitch accuracy, rhythmic stability, fingering precision, and sight-reading performance yielded a moderate positive effect. The confidence interval did not overlap with zero, indicating reliable benefits for psychomotor skill development. Representative studies in this domain employed technologies such as automated performance evaluation systems (Phanichraksaphong \u0026amp; Tsai, \u003cspan citationid=\"CR104\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), real-time error detection (Yu, \u003cspan citationid=\"CR142\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), and adaptive difficulty adjustment (Tang, \u003cspan citationid=\"CR123\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003eMusical Expression (k\u0026thinsp;=\u0026thinsp;11, d\u0026thinsp;=\u0026thinsp;0.432, 95% CI [0.339, 0.525])\u003c/b\u003e: Effects for expressive dimensions\u0026mdash;including dynamic control, tempo variation, articulation quality, and interpretive fluency\u0026mdash;were comparable in magnitude to technical skills, contradicting the common assumption that AI is effective only for easily quantifiable outcomes. Studies in this category often utilized multimodal analysis (audio-visual fusion; Zhao et al., \u003cspan citationid=\"CR151\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), performance comparison with expert models (Giraldo et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), and visualized feedback on expressive parameters (Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, the relatively smaller number of studies (k\u0026thinsp;=\u0026thinsp;11) and moderate heterogeneity (I\u0026sup2; = 61.7%) suggest this domain requires further empirical investigation.\u003c/p\u003e \u003cp\u003e \u003cb\u003eLearning Motivation (k\u0026thinsp;=\u0026thinsp;33, d\u0026thinsp;=\u0026thinsp;0.434, 95% CI [0.378, 0.490])\u003c/b\u003e: AI-supported interventions showed moderate positive effects on motivational outcomes, including practice engagement, persistence, self-efficacy, and affective responses toward piano learning. This domain included the largest number of studies (k\u0026thinsp;=\u0026thinsp;33), reflecting growing research interest in non-cognitive outcomes. Common intervention features included gamification elements (Huang \u0026amp; Ding, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), progress visualization (Deng et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), and adaptive challenge adjustment (Strielkowski et al., \u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The consistency of positive effects across diverse motivational measures suggests that AI\u0026rsquo;s capacity to provide immediate, personalized feedback may enhance learners\u0026rsquo; sense of competence and autonomy.\u003c/p\u003e \u003cp\u003eA test of moderators revealed no statistically significant differences between domains (Q_between\u0026thinsp;=\u0026thinsp;0.42, df\u0026thinsp;=\u0026thinsp;2, p = .811), indicating that effect sizes were comparable across technical, expressive, and motivational outcomes. The overlapping confidence intervals further support this interpretation. This finding challenges prevailing assumptions that AI\u0026rsquo;s benefits are confined to lower-order technical skills, suggesting instead that well-designed AI systems can support multidimensional learning outcomes in piano education.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e3.2.3. Effect Size Distribution and Visual Synthesis\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents a forest plot displaying individual study effect sizes in chronological order (2020\u0026ndash;2025), along with 95% confidence intervals and the overall pooled estimate (indicated by the diamond marker). Visual inspection reveals that the majority of effect sizes (53 out of 62, or 85.5%) were positive, with only 9 studies reporting near-zero or slightly negative effects. The width of confidence intervals varied considerably across studies, reflecting differences in sample sizes (ranging from N\u0026thinsp;=\u0026thinsp;30 to N\u0026thinsp;=\u0026thinsp;60) and outcome measurement precision.\u003c/p\u003e \u003cp\u003eSeveral patterns emerge from the visual distribution. First, effect sizes clustered predominantly in the small-to-moderate range (d\u0026thinsp;=\u0026thinsp;0.20\u0026ndash;0.60), with few outliers exceeding d\u0026thinsp;=\u0026thinsp;0.70. Second, studies published in 2023\u0026ndash;2025 did not exhibit systematically larger or smaller effects compared to earlier years, suggesting temporal stability in reported outcomes. Third, confidence intervals for most studies overlapped substantially, indicating that while effect size point estimates varied, the precision of estimates was relatively consistent across the literature.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNote\u003c/strong\u003e \u003cp\u003eThis forest plot displays standardized effect sizes (Cohen's d) and 95% confidence intervals for all 62 included studies, organized chronologically by publication year. Each horizontal line represents a single study's effect size estimate and confidence interval. The vertical dashed line at d\u0026thinsp;=\u0026thinsp;0 indicates no effect. Studies with confidence intervals not crossing the zero line demonstrate statistically significant effects. The diamond marker at the bottom represents the overall pooled effect size (d\u0026thinsp;=\u0026thinsp;0.442, 95% CI [0.398, 0.486]) calculated using random-effects meta-analysis. Color coding indicates instructional domain: blue\u0026thinsp;=\u0026thinsp;Technical Skills (k\u0026thinsp;=\u0026thinsp;18), green\u0026thinsp;=\u0026thinsp;Musical Expression (k\u0026thinsp;=\u0026thinsp;11), orange\u0026thinsp;=\u0026thinsp;Learning Motivation (k\u0026thinsp;=\u0026thinsp;33). The width of confidence intervals reflects sample size and measurement precision, with narrower intervals indicating more precise estimates. Visual inspection reveals predominantly positive effects across all domains and years, with substantial between-study variability (I\u0026sup2; = 61.1%) supporting the use of random-effects modeling.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e3.2.4. Publication Bias Assessment\u003c/h2\u003e \u003cp\u003ePotential publication bias was assessed through visual inspection of a funnel plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003e) and Egger's regression test. The funnel plot displayed a reasonably symmetric distribution of effect sizes around the pooled estimate, with studies dispersed across varying levels of standard error. Egger's regression test yielded a non-significant result (z\u0026thinsp;=\u0026thinsp;1.31, p = .191), suggesting no strong evidence of asymmetry attributable to publication bias. This finding indicates that the observed pooled effect size is unlikely to be substantially inflated by the preferential publication of studies with positive results.\u003c/p\u003e \u003cp\u003eHowever, it is important to note that the absence of detectable publication bias does not guarantee its complete absence, particularly given that only peer-reviewed journal articles were included in this review. Studies with null or negative findings may remain unpublished or appear in non-indexed outlets (Pigott, \u003cspan citationid=\"CR105\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Additionally, small-study effects\u0026mdash;where smaller studies report larger effect sizes\u0026mdash;were not strongly evident in this dataset, further supporting the robustness of the pooled estimate.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003ePublication Bias Assessment:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eEgger's Regression Test: z\u0026thinsp;=\u0026thinsp;1.31, p = .191 (not significant)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePooled Effect Size: d\u0026thinsp;=\u0026thinsp;0.442 (95% CI [0.398, 0.486])\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eNumber of Studies: k\u0026thinsp;=\u0026thinsp;62\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eInterpretation: No significant evidence of publication bias detected\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNote\u003c/strong\u003e \u003cp\u003eThis funnel plot displays the relationship between effect sizes (Cohen's d, x-axis) and their standard errors (y-axis, inverted) for all 62 included studies. In the absence of publication bias, studies should be symmetrically distributed around the pooled effect estimate (vertical dashed line at d\u0026thinsp;=\u0026thinsp;0.442), forming an inverted funnel shape. Studies with smaller sample sizes (larger standard errors) appear toward the top of the plot with greater horizontal dispersion, while larger studies (smaller standard errors) cluster more tightly near the pooled estimate at the bottom. Visual inspection reveals a reasonably symmetric distribution, with no pronounced asymmetry suggesting selective publication of positive findings. Egger's regression test confirmed this observation (z\u0026thinsp;=\u0026thinsp;1.31, p = .191), indicating no significant evidence of publication bias. The triangular reference region (shaded) represents the 95% confidence interval around the pooled effect, within which 95% of studies would be expected to fall in the absence of heterogeneity.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003e3.2.5. Sensitivity Analysis\u003c/h2\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eTo assess the robustness of the pooled effect size, sensitivity analyses were conducted by sequentially removing studies with the largest and smallest effect sizes. Removal of the three largest effect sizes (d\u0026thinsp;\u0026gt;\u0026thinsp;0.65) resulted in a pooled estimate of d\u0026thinsp;=\u0026thinsp;0.425 (95% CI [0.382, 0.468]), while removal of the three smallest effect sizes (d\u0026thinsp;\u0026lt;\u0026thinsp;0.25) yielded d\u0026thinsp;=\u0026thinsp;0.450 (95% CI [0.405, 0.495]). Both estimates remained within the moderate range and closely aligned with the original pooled effect (d\u0026thinsp;=\u0026thinsp;0.442), indicating that the overall finding was not disproportionately influenced by extreme values.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAdditionally, a leave-one-out analysis was performed, in which the meta-analysis was re-run 62 times, each time excluding a different study. Pooled effect sizes ranged from d\u0026thinsp;=\u0026thinsp;0.435 to d\u0026thinsp;=\u0026thinsp;0.449, with all estimates remaining statistically significant (p \u0026lt; .001). This narrow range further supports the stability and reliability of the meta-analytic synthesis.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003e3.2.6. Summary of Meta-analytic Findings\u003c/h2\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eThe meta-analytic synthesis across 62 studies provides convergent evidence that AI-supported piano instruction yields moderate positive effects on learning outcomes, with comparable benefits observed across technical skills, musical expression, and learning motivation. The consistency of effects across domains challenges the notion that AI is effective only for easily quantifiable technical training, suggesting instead that AI systems\u0026mdash;when appropriately designed\u0026mdash;can support holistic musical development. The substantial heterogeneity observed (I\u0026sup2; = 61.1%) underscores the importance of contextual factors, including instructional design quality, learner characteristics, and technological affordances, in determining the effectiveness of AI interventions.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.3. AI Empowers Piano Students' Technical Skills, Expressiveness, Learning Motivation\u003c/h2\u003e \u003cp\u003eIn recent years, with the application of AI technologies such as deep learning, automatic music transcription, multimodal analysis, and intelligent feedback systems in music education, more studies have begun to explore their mechanisms of action on piano learning effectiveness and motivation. Based on these studies, we can examine AI\u0026rsquo;s impact from three dimensions: (1) basic performance skills, (2) musical expressiveness, and (3) learning motivation and participation.\u003c/p\u003e \u003cdiv id=\"Sec19\" class=\"Section3\"\u003e \u003ch2\u003e3.3.1 .Technical Skills\u003c/h2\u003e \u003cp\u003eTraditional piano instruction relies heavily on teacher supervision for developing basic performance skills\u0026mdash;a model constrained by limited class time and infrequent feedback during independent practice (Cao, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Zhang, \u003cspan citationid=\"CR148\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). AI technologies address these structural limitations by providing immediate, quantifiable feedback on technical aspects such as pitch accuracy, rhythmic stability, and fingering precision.Research consistently demonstrates AI's effectiveness in technical training. Automated evaluation systems can classify articulation types (legato vs. staccato) with accuracy exceeding 89% (Phanichraksaphong \u0026amp; Tsai, \u003cspan citationid=\"CR104\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), while real-time error detection enables targeted correction during practice (Yu, \u003cspan citationid=\"CR142\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Perhaps more significantly, AI systems construct adaptive learning paths by analyzing error patterns and adjusting difficulty dynamically, allowing learners to focus on specific technical obstacles rather than repeatedly playing entire pieces (Strielkowski et al., \u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).The precision and objectivity of AI feedback represents a clear advantage over human real-time evaluation in most technical domains (Lu, \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Large language models further enhance this capacity by intelligently matching learning resources to students' skill levels (Deng et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). However, these advantages remain concentrated in quantifiable dimensions. Long-term impacts on advanced musical expression and artistic judgment require verification through larger-sample longitudinal studies, suggesting AI functions better as an efficient auxiliary tool than as a replacement for teachers' aesthetic guidance.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section3\"\u003e \u003ch2\u003e3.3.2. Musical Expressiveness\u003c/h2\u003e \u003cp\u003eMusical expressiveness\u0026mdash;encompassing structural understanding, emotional communication, dynamic control, and stylistic interpretation\u0026mdash;has long been considered resistant to technological intervention due to its subjective, context-dependent nature (Bonnaire \u0026amp; Gonz\u0026aacute;lez-Moreno, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Giraldo et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Recent AI research challenges this assumption by decomposing expressiveness into algorithmically analyzable parameters: dynamic range profiles, inter-note timing, rhythmic elasticity, and articulation patterns.Multimodal approaches show particular promise. Audio-visual fusion models that jointly analyze acoustic effects and hand movements outperform audio-only systems in detecting expressive timing variations and interpretive intentions (Zhao et al., \u003cspan citationid=\"CR151\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). MIDI-based performance analysis enables detailed modeling of dynamic contours and tempo variations, providing learners with quantitative feedback that clarifies abstract concepts like \"layering\" and \"musical breath\" more effectively than verbal instruction alone (Ru, \u003cspan citationid=\"CR112\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zhang, \u003cspan citationid=\"CR148\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Even young learners show heightened attention to dynamic contrast and musical coherence when provided with visualized expressive feedback (Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).Yet significant limitations persist. Current AI systems focus primarily on measurable expressive parameters while struggling with emotional content, stylistic context, and cultural significance (Giraldo et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Empirical research concentrates on short-term interventions with small samples, lacking longitudinal evidence on developmental trajectories and transfer effects (Yang et al., \u003cspan citationid=\"CR138\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Moreover, over-reliance on standardized feedback may inadvertently constrain personalized interpretation and creative expression (Zhai et al., \u003cspan citationid=\"CR145\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). These constraints underscore that while AI can support expressive skill development, it cannot replace the nuanced aesthetic judgment central to artistic piano teaching.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section3\"\u003e \u003ch2\u003e3.3.3. Learning Motivation And Participation\u003c/h2\u003e \u003cp\u003ePiano learning demands sustained autonomous practice, making intrinsic motivation and self-regulation critical for long-term achievement (Panadero, \u003cspan citationid=\"CR103\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Traditional instruction maintains motivation primarily through teacher supervision and external rewards\u0026mdash;mechanisms unavailable during independent practice when learners face vague goals and declining self-efficacy (Zhang, \u003cspan citationid=\"CR148\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). AI systems address these challenges by reconstructing practice experiences through immediate feedback, progress visualization, and adaptive challenge adjustment.Empirical evidence demonstrates that AI-enhanced environments significantly boost motivation, particularly among beginners. Gamification elements, instant scoring, and achievement tracking reduce practice frustration while enhancing engagement sustainability (Huang \u0026amp; Ding, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In one study of preschool children, 84% reported that intelligent systems helped them practice proactively (Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). University-level blended learning incorporating AI feedback shows similar patterns: learners achieve technical improvements while demonstrating heightened engagement and persistence (Tang, \u003cspan citationid=\"CR123\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).Beyond motivation, AI supports self-regulated learning (SRL) by facilitating goal-setting, strategic practice, and self-monitoring. Rather than requiring students to repeatedly play entire pieces, AI identifies specific technical obstacles and delivers targeted practice segments, fostering more strategic learning behaviors (Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Adaptive systems maintain learning within the \"optimal challenge zone\" by dynamically adjusting content based on real-time performance, promoting sustained engagement and self-efficacy development (Strielkowski et al., \u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).Emerging LLM-based platforms extend this support into reflective learning by generating personalized explanations and practice suggestions. Such dialogue-based feedback helps externalize previously implicit self-regulation processes, potentially enhancing metacognitive development (Deng et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Jin et al., \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, LLM feedback quality remains dependent on training data and prompt design, with artistic judgment and stylistic interpretation still requiring expert teacher guidance (Zhou et al., \u003cspan citationid=\"CR152\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e3.4. Ethical Issues in Piano Education Empowered by AI\u003c/h2\u003e \u003cp\u003eThe proliferation of AI in piano instruction raises multiple ethical concerns requiring systematic attention. Foremost among these is data privacy: AI systems collect extensive audio, video, and behavioral data\u0026mdash;including images of minors' home environments\u0026mdash;creating substantial risks for misuse, re-identification, and unauthorized sharing absent robust governance frameworks (UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Right to Education Initiative, 2022).Copyright disputes present another governance challenge. AI training often relies on large-scale scores and recordings, many protected by copyright, as recent lawsuits demonstrate (The Guardian, 2023; Reuters, 2025). The ownership and usage boundaries of AI-generated accompaniments, adapted etudes, or teaching materials remain legally ambiguous. Algorithmic fairness concerns compound these issues: imbalanced training data may produce systematic biases against non-mainstream performance styles or non-Western musical traditions, potentially affecting learners' self-perception and educational opportunities (Baker, 2022).The \"black box\" nature of deep learning models undermines transparency and trust. Automated assessment systems that cannot explain their judgment criteria erode confidence among teachers, students, and parents\u0026mdash;particularly problematic in arts education where evaluations directly shape learning trajectories (Miao \u0026amp; Holmes, 2021). Responsibility attribution remains unresolved: when AI produces erroneous outputs affecting students, determining whether developers, platforms, schools, or teachers bear accountability lacks consensus (Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).Additional concerns include psychological impacts of continuous monitoring, the digital divide exacerbated by high-cost platforms, and risks of corporate control over teaching tools weakening institutional autonomy (UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Existing literature converges on a clear position: effective AI deployment in piano education demands explicit ethical guidelines, transparent data governance, and teacher-centered human-machine collaboration rather than technological substitution (Alam, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Fitria, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Zhao et al., \u003cspan citationid=\"CR150\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e3.5. Transformation of the Teacher's Role\u003c/h2\u003e \u003cp\u003eEmpirical evidence reveals both strong acceptance and significant implementation barriers regarding AI in music teaching. Teachers express considerable willingness to adopt AI tools (mean\u0026thinsp;=\u0026thinsp;4.24/5.00 on Likert scales; Aguila et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), recognizing how generative AI optimizes self-directed learning through immediate feedback (Li \u0026amp; Wang, \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). However, substantial obstacles persist: cognitive burden during implementation, insufficient professional training, and concerns about algorithmic bias limiting pedagogical autonomy (Atabek \u0026amp; Burak, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Kehoe, \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCurrent generative models exhibit cultural biases that threaten both originality and equity in music education (Amankwah-Amoagh et al., 2024). While AI-generated lesson plans may demonstrate high quality, they lack context-adaptability and require extensive human intervention to achieve pedagogical effectiveness (Kehoe, \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). These findings point toward a reconfiguration rather than replacement of teachers' roles. Educators increasingly function as aesthetic mentors, pedagogical decision-makers, and technology coordinators rather than sole providers of technical instruction (Aguila et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).This evolution demands new competencies. Teachers require not only digital literacy but also critical understanding of AI capabilities and limitations, enabling them to mitigate algorithmic bias and navigate complex, context-dependent instructional processes (Atabek \u0026amp; Burak, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The sustainability of AI integration thus depends less on technological advancement than on comprehensive teacher professional development, clear ethical frameworks, and institutional support structures that preserve pedagogical agency while leveraging technological affordances.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eDrawing on empirical evidence from 62 studies of AI-related piano education published between 2020 and 2025, the present systematic review and meta-analysis reveals several critical findings regarding the educational efficacy, methodological rigor, and contextual validity of current research. The meta-analytic synthesis demonstrated that AI-supported interventions yield a moderate positive effect on piano learning outcomes (d\u0026thinsp;=\u0026thinsp;0.442, 95% CI [0.398, 0.486]), with comparable benefits observed across technical skills, musical expression, and learning motivation. This quantitative finding provides the first pooled effect size estimate in the domain of AI-enhanced piano education and offers an empirical benchmark for future intervention studies and policy development.However, while the overall effect size indicates meaningful learning gains, substantial heterogeneity was observed (I\u0026sup2; = 61.1%), suggesting that the magnitude of AI's impact varies considerably across studies. This variability points to the critical importance of contextual factors\u0026mdash;including instructional design quality, technological affordances, learner characteristics, and pedagogical integration strategies\u0026mdash;in determining intervention effectiveness. Closer examination of the reviewed literature raises four major concerns related to research scope, contextual validity, methodological balance, and pedagogical alignment, which collectively affect the interpretation, generalizability, and practical implications of current research findings (Chen, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Selwyn, \u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.1. The Significance and Limitations of Moderate Effect Sizes\u003c/h2\u003e \u003cp\u003eThe observed pooled effect size of d\u0026thinsp;=\u0026thinsp;0.442 falls within Cohen's (1988) classification of a \"medium\" effect and is comparable to meta-analytic findings in related educational technology domains. For instance, prior reviews of intelligent tutoring systems in mathematics education reported effect sizes ranging from d\u0026thinsp;=\u0026thinsp;0.35 to d\u0026thinsp;=\u0026thinsp;0.50 (Steenbergen-Hu \u0026amp; Cooper, \u003cspan citationid=\"CR120\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), while meta-analyses of computer-assisted instruction across disciplines yielded similar estimates (Cheung \u0026amp; Slavin, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). This convergence suggests that AI's impact on piano learning aligns with broader patterns observed in technology-enhanced education, where moderate positive effects are typical for well-designed interventions.\u003c/p\u003e \u003cp\u003eImportantly, the finding that effect sizes were comparable across technical skills (d\u0026thinsp;=\u0026thinsp;0.462), musical expression (d\u0026thinsp;=\u0026thinsp;0.432), and learning motivation (d\u0026thinsp;=\u0026thinsp;0.434) challenges a prevalent assumption in music education discourse: that AI's benefits are confined to easily quantifiable, lower-order technical outcomes. The near-equivalent effects for musical expression\u0026mdash;an inherently interpretive and context-dependent dimension of musicianship\u0026mdash;suggest that AI systems, when designed with pedagogical sophistication, can support higher-order learning processes. This may reflect recent advances in multimodal AI technologies (e.g., audio-visual fusion models; Zhao et al., \u003cspan citationid=\"CR151\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) and performance analysis tools that provide learners with visualized feedback on expressive parameters such as dynamic contours and tempo variation (Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Giraldo et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNevertheless, several critical limitations must be acknowledged. First, the moderate effect size reflects short-term outcomes in the majority of studies (median intervention duration: 8\u0026ndash;12 weeks), leaving long-term retention, skill transfer, and sustained motivational effects largely unexamined. Second, the high heterogeneity (I\u0026sup2; = 61.1%) indicates that not all AI interventions are equally effective, and some studies reported near-zero or even slightly negative effects (9 out of 62 studies). This variability underscores that technological sophistication alone does not guarantee pedagogical value; rather, effectiveness depends on alignment between AI affordances and instructional goals, as well as on thoughtful integration into existing teaching practices (Holmes et al., \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThird, while the absence of significant publication bias (Egger's test: p = .191) is reassuring, the exclusive focus on peer-reviewed journal articles may still introduce selection bias, as studies with null or negative findings are less likely to be published (Pigott, \u003cspan citationid=\"CR105\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). The meta-analytic estimate should therefore be interpreted as an upper-bound approximation of AI's impact under favorable implementation conditions, rather than as a definitive causal claim applicable to all contexts.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e4.2. Technology-Driven Orientation and Pedagogical Implications\u003c/h2\u003e \u003cp\u003eA strong technology-driven orientation was identified across the majority of reviewed studies. Much of the existing research frames AI as a solution to long-standing challenges in piano education, particularly those associated with limited teacher feedback, inefficient practice, and subjective assessment (Chen, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This optimism aligns with broader educational technology discourses that emphasize innovation, automation, and efficiency (Holmes et al., \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, as observed in policy-driven educational reforms in music education more broadly, such technological enthusiasm may obscure deeper pedagogical questions.In many studies, the integration of AI systems is treated as a neutral or inherently beneficial intervention, while underlying assumptions about musicianship, artistic judgment, and learning processes remain insufficiently theorized (Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Selwyn, \u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). For example, several studies equate \"improved learning outcomes\" with higher scores on automated pitch detection or rhythm accuracy tests (Yu, \u003cspan citationid=\"CR142\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Phanichraksaphong \u0026amp; Tsai, \u003cspan citationid=\"CR104\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), without addressing whether these gains translate into more musically meaningful performances or whether students develop interpretive autonomy. The meta-analytic finding of comparable effect sizes across domains (technical, expressive, motivational) suggests potential, but the lack of qualitative investigation into how students experience and make sense of AI-mediated feedback limits our understanding of these processes.\u003c/p\u003e \u003cp\u003eAs a result, there appears to be a growing gap between technologically sophisticated research outputs and the nuanced realities of piano teaching practices, particularly with regard to expressive interpretation, aesthetic judgment, and the relational dimensions of teacher\u0026ndash;student interaction. This gap is reflected in the reviewed studies' predominant reliance on short-term, quantifiable outcome measures (e.g., pitch accuracy rates, completion times) while marginalizing dimensions of musicianship that resist standardization\u0026mdash;such as stylistic sensitivity, emotional communication, and creative risk-taking (Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e4.3. The Illusion of Scalability: Contextual Validity and Educational Equity\u003c/h2\u003e \u003cp\u003eThe literature exhibits a tendency to frame increased technological investment as a primary means of improving piano education quality. Advanced AI systems\u0026mdash;such as intelligent pianos, multimodal performance analysis tools, and adaptive learning platforms\u0026mdash;are frequently presented as scalable solutions capable of transforming learning outcomes across diverse contexts (Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Strielkowski et al., \u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). While the meta-analytic findings support the claim of positive average effects in controlled settings, they risk oversimplifying the complex conditions under which piano education occurs.A critical examination of the reviewed studies reveals that a substantial proportion of empirical research was conducted in economically developed regions, elite institutions, or technology-rich environments where access to AI-supported tools is readily available (Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Strielkowski et al., \u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). For instance, 68% of the included studies were conducted in East Asia (particularly China, South Korea, Taiwan) and Western Europe, with only 12% originating from Latin America, Africa, or Southeast Asia. Moreover, 73% of studies involved participants from university-affiliated music programs, conservatories, or private music schools\u0026mdash;settings characterized by high levels of institutional resources and technological infrastructure.This concentration suggests a form of contextual and selection bias, whereby reported benefits may reflect favorable implementation conditions rather than universally applicable outcomes (Baker et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The external validity of findings is further constrained by limited attention to factors such as:\u003c/p\u003e \u003cp\u003eDigital access disparities: Few studies addressed whether students had reliable internet connectivity, personal devices, or adequate home practice spaces for AI-supported learning.Socioeconomic diversity: Sample descriptions rarely included information on participants' socioeconomic backgrounds, making it difficult to assess whether AI interventions exacerbate or mitigate existing educational inequalities.Cultural and linguistic contexts: AI systems trained predominantly on Western classical music repertoire and pedagogical norms may exhibit algorithmic bias when applied to non-Western musical traditions or pedagogies (Baker, 2022; Amankwah-Amoagh et al., 2024).\u003c/p\u003e \u003cp\u003eAs with broader concerns in educational research, such bias limits the transferability of findings to under-resourced contexts or regions with limited digital infrastructure (UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The moderate pooled effect size (d\u0026thinsp;=\u0026thinsp;0.442) should therefore be interpreted with caution, recognizing that it may overestimate AI's impact in less advantaged educational settings.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e4.4. Methodological Imbalance: What Gets Measured, What Gets Marginalized\u003c/h2\u003e \u003cp\u003eA fourth concern relates to a methodological imbalance between technical performance metrics and broader educational outcomes. The meta-analysis revealed that many studies prioritized quantifiable indicators\u0026mdash;such as pitch accuracy, rhythm precision, or error rates\u0026mdash;while comparatively few examined long-term learning trajectories, self-regulated practice behaviors, teacher professional development, or ethical implications (Panadero, \u003cspan citationid=\"CR103\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).This imbalance is reflected in the distribution of outcome measures across the 62 studies:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eTechnical accuracy metrics: 73% of studies (k\u0026thinsp;=\u0026thinsp;45)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePractice engagement/time-on-task: 53% of studies (k\u0026thinsp;=\u0026thinsp;33)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eMusical expressiveness (subjective assessment): 18% of studies (k\u0026thinsp;=\u0026thinsp;11)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eLong-term retention (\u0026gt;\u0026thinsp;3 months post-intervention): 6% of studies (k\u0026thinsp;=\u0026thinsp;4)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTeacher perspectives or professional development: 8% of studies (k\u0026thinsp;=\u0026thinsp;5)\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe dominance of short-term experimental designs and system validation studies may inadvertently reinforce a narrow conception of learning success, marginalizing aspects of musicianship that resist automation or standardization\u0026mdash;such as interpretive originality, stylistic flexibility, and the capacity for critical aesthetic judgment (Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). While the meta-analysis found comparable effect sizes for musical expression (d\u0026thinsp;=\u0026thinsp;0.432), this finding is based on only 11 studies, many of which operationalized \"expressiveness\" through relatively objective parameters (e.g., dynamic range profiles, tempo deviation statistics) rather than holistic artistic judgments.This methodological trend parallels broader critiques in critical music education scholarship, which caution against reducing musicianship to measurable outcomes (Regelski, \u003cspan citationid=\"CR110\" class=\"CitationRef\"\u003e2002\u003c/span\u003e). If AI research continues to prioritize easily quantifiable indicators, there is a risk of implicitly valuing technical compliance over creative exploration, and efficiency over aesthetic depth\u0026mdash;ultimately shaping a narrower vision of what it means to learn piano (Selwyn, \u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e4.5. Toward Context-Sensitive, Pedagogically Grounded AI Integration\u003c/h2\u003e \u003cp\u003eTaken together, these findings suggest that while AI holds genuine promise for supporting piano education, its educational value cannot be realized through technological advancement alone. The moderate positive effect size (d\u0026thinsp;=\u0026thinsp;0.442) observed in this meta-analysis should be understood not as definitive proof of AI's superiority, but as evidence that well-designed, pedagogically grounded AI systems can meaningfully augment certain dimensions of piano learning when implemented under favorable conditions.Future research should adopt more context-sensitive, theory-informed, and pedagogically nuanced approaches that recognize:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eThe irreplaceable role of teachers in fostering artistic judgment, interpretive autonomy, and emotional connection to music\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe diversity of learning environments, including under-resourced schools, community music programs, and non-Western pedagogical traditions\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe artistic dimensions of piano education that extend beyond measurable technical proficiency, encompassing creativity, stylistic sensitivity, and musical meaning-making\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eWithout such recalibration, there remains a risk that AI-enhanced piano education research may offer a technologically refined but pedagogically partial account of educational improvement\u0026mdash;one that privileges efficiency and scalability over the complexity and depth that characterize meaningful musical learning (Holmes et al., \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusions and Future Research Directions","content":"\u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003e5.1. Conclusions\u003c/h2\u003e \u003cp\u003eOur systematic review and meta-analysis of 62 studies (N\u0026thinsp;=\u0026thinsp;2,812 participants) spanning 2020 through 2025 establishes the first empirical benchmark for AI in piano education. The central finding\u0026mdash;a moderate positive effect (d\u0026thinsp;=\u0026thinsp;0.442, 95% CI [0.398, 0.486], p \u0026lt; .0001)\u0026mdash;tells an encouraging yet nuanced story about AI's potential.What surprised us most was the consistency across domains. Effect sizes for technical skills (d\u0026thinsp;=\u0026thinsp;0.462), musical expression (d\u0026thinsp;=\u0026thinsp;0.432), and learning motivation (d\u0026thinsp;=\u0026thinsp;0.434) proved remarkably similar, challenging conventional wisdom that AI helps only with measurable technical drills. The comparable impact on musical expressiveness\u0026mdash;an inherently interpretive dimension\u0026mdash;suggests that thoughtfully designed systems can indeed support multidimensional learning when properly integrated into pedagogical contexts.However, the substantial heterogeneity observed (I\u0026sup2; = 61.1%) complicates any straightforward endorsement. Not all AI interventions work equally well. Some studies reported near-zero effects while others found substantial gains, underscoring that technological sophistication alone guarantees nothing. Effectiveness hinges on instructional design quality, learner characteristics, and alignment between AI affordances and pedagogical goals (Holmes et al., \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Selwyn, \u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).The evidence reveals both promise and limitations. AI-supported tools effectively enhance short-term technical accuracy, deliver immediate feedback, and increase practice engagement (Chen, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Li, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Zhao et al., \u003cspan citationid=\"CR151\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Yet our review also exposes methodological imbalances: most studies prioritize quantifiable performance indicators while marginalizing dimensions like stylistic sensitivity, emotional communication, and creative risk-taking. Although we found comparable effect sizes for musical expression, this rests on only 11 studies, many operationalizing expressiveness through objective parameters rather than holistic artistic judgment.Ethical concerns emerged repeatedly across the literature. AI piano systems rely heavily on learner data\u0026mdash;audio, video, behavioral logs\u0026mdash;raising privacy risks especially acute when minors are involved (UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Algorithmic bias threatens to disadvantage learners from diverse musical backgrounds, as training data predominantly reflects Western classical traditions (Baker et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Amankwah-Amoagh et al., 2024). The concentration of empirical research in well-resourced institutions compounds questions about generalizability to community music programs or non-Western pedagogical traditions.\u003c/p\u003e \u003cp\u003eImportantly, our analysis indicates reconfiguration rather than replacement of teachers' roles. Educators increasingly serve as aesthetic mentors, pedagogical strategists, and technology coordinators rather than mere technical instructors (Aguila et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). While teachers express high AI acceptance (mean\u0026thinsp;=\u0026thinsp;4.24/5.00), they report substantial implementation barriers: cognitive burden, insufficient training, and concerns about algorithmic bias (Atabek \u0026amp; Burak, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This suggests sustainability depends not merely on technological development but on comprehensive professional development, clear ethical frameworks, and institutional support preserving pedagogical agency.\u003c/p\u003e \u003cp\u003eIn summary, AI holds genuine potential for supporting piano learning, but its educational value depends critically on pedagogically informed integration rather than technological capability alone. The moderate effect size (d\u0026thinsp;=\u0026thinsp;0.442) should be understood not as proof of AI's superiority but as evidence that well-designed systems, when implemented thoughtfully and equitably, can meaningfully augment certain dimensions of instruction. Advancing the field requires context-sensitive, theory-driven, ethically informed approaches recognizing teachers' irreplaceable role, environmental diversity, and the artistic foundations distinguishing musical from purely technical education.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003e5.2. Research Gaps and Future Research Directions\u003c/h2\u003e \u003cp\u003eOur review identifies five critical gaps warranting empirical attention:\u003c/p\u003e \u003cp\u003eFirst, longitudinal evidence remains scarce. Most studies examine short-term outcomes (median duration: 8\u0026ndash;12 weeks), leaving sustained retention, skill transfer, and long-term motivational effects largely unexamined. Future research should adopt longitudinal designs tracking how AI-supported practice influences technical development, expressive growth, and self-regulated learning over extended periods (Panadero, \u003cspan citationid=\"CR103\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhai \u0026amp; Xu, \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSecond, theoretical grounding needs strengthening. Many studies focus on measurable indicators while remaining weakly connected to established learning theories. Explicitly integrating frameworks like self-regulated learning, experiential learning, or constructivist pedagogy would clarify mechanisms through which AI supports learning (Holmes et al., \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThird, contextual diversity and equity require systematic investigation. Evidence concentrates in economically developed regions and elite institutions. Comparative research across diverse socio-economic contexts is essential for evaluating feasibility, adaptability, and inclusiveness of AI-enhanced instruction (Baker et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFourth, teachers' professional experiences need deeper exploration. While shifts toward facilitation and learning design are acknowledged, few studies investigate how educators negotiate professional identity, pedagogical authority, and ethical responsibility within AI-mediated environments. Teacher-centered qualitative research is critical for understanding sustainable integration models (Selwyn, \u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Bell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFinally, ethical governance in piano-specific applications remains underexplored. Issues like data privacy, algorithmic transparency, bias, and accountability appear in general AI-in-education literature but rarely receive empirical examination in music education. Future research should develop operational frameworks ensuring responsible, trustworthy AI deployment (Baker et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; UNESCO, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec33\" class=\"Section2\"\u003e \u003ch2\u003e5.3. Concluding Remarks\u003c/h2\u003e \u003cp\u003eIn conclusion, while artificial intelligence offers promising tools for enhancing aspects of piano education, its pedagogical impact cannot be reduced to technological efficiency or performance optimization. Advancing the field requires a shift toward context-sensitive, theory-driven, and ethically informed research agendas that recognize the irreplaceable role of teachers, the diversity of learning environments, and the artistic foundations of piano education. Addressing these challenges will be essential for ensuring that AI-enhanced piano education contributes meaningfully to both educational quality and musical integrity.\u003c/p\u003e \u003c/div\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAguila AA, Al-Samarraie H, Al-Buraiki A, Al-Sharhan N, Aldheleai T (2024) Artificial intelligence in teaching music skills and teachers' attitudes towards it in the Sultanate of Oman. J Eng Appl Sci 19(5):1\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.36478/jeasci.2024.19.05\u003c/span\u003e\u003cspan address=\"10.36478/jeasci.2024.19.05\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAhn S, Ames AJ, Myers ND (2012) A review of meta-analyses in education: Methodological strengths and weaknesses. Rev Educ Res 82(4):436\u0026ndash;476. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3102/0034654312458162\u003c/span\u003e\u003cspan address=\"10.3102/0034654312458162\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAkgun S, Greenhow C (2022) Artificial intelligence in education: Addressing ethical challenges in K-12 settings. AI Ethics 2(3):431\u0026ndash;440. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s43681-021-00096-7\u003c/span\u003e\u003cspan address=\"10.1007/s43681-021-00096-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlam A (2021), December Should robots replace teachers? Mobilisation of AI and learning analytics in education. In \u003cem\u003e2021 International Conference on Advances in Computing, Communication, and Control (ICAC3)\u003c/em\u003e (pp. 1\u0026ndash;12). IEEE. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICAC353642.2021.9697300\u003c/span\u003e\u003cspan address=\"10.1109/ICAC353642.2021.9697300\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAli O, Murray PA, Momin M, Dwivedi YK, Malik T (2024) The effects of artificial intelligence applications in educational settings: Challenges and strategies. Technol Forecast Soc Chang 199., Article 123076. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.techfore.2023.123076\u003c/span\u003e\u003cspan address=\"10.1016/j.techfore.2023.123076\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAn X, Chai CS, Li Y, Zhou Y, Shen X, Zheng C, Chen M (2023) Modeling English teachers' behavioral intention to use artificial intelligence in middle schools. Educ Inform Technol 28(5):5187\u0026ndash;5208. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-022-11286-z\u003c/span\u003e\u003cspan address=\"10.1007/s10639-022-11286-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAntonini Philippe R, Kosirnik C, Vuichoud N, Williamon A, von Roten FC (2020) Understanding wellbeing among college music students and amateur musicians in Western Switzerland. Front Psychol 11:820. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2020.00820\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2020.00820\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAtabek O, Burak S (2024) Exploring pre-service music teachers' acceptance of generative artificial intelligence: A PLS-SEM-ANN approach. Front Psychol 15., Article 1571279. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2024.1571279\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2024.1571279\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBahroun Z, Anane C, Ahmed V, Zacca A (2023) Transforming education: A comprehensive review of generative artificial intelligence in educational settings through bibliometric and content analysis. Sustainability 15(17):12983. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/su151712983\u003c/span\u003e\u003cspan address=\"10.3390/su151712983\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBai B (2021) Piano learning in the context of schooling during China's 'piano craze' and beyond: Motivations and pathways. Music Educ Res 23(4):512\u0026ndash;526. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/14613808.2021.1929138\u003c/span\u003e\u003cspan address=\"10.1080/14613808.2021.1929138\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaker RS, Hawn A, Zawacki-Richter O (2022) Algorithmic bias in education. Int J Artif Intell Educ 32(4):1055\u0026ndash;1082. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40593-021-00285-9\u003c/span\u003e\u003cspan address=\"10.1007/s40593-021-00285-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBanquiero M, Valdeolivas G, Ram\u0026oacute;n D (2024) A color passthrough mixed reality application for learning piano. Virtual Reality 28(2). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 67. https://doi.org/10.1007/s10055-024-00967-8\u003c/span\u003e\u003cspan address=\"Article 67. 10.1007/s10055-024-00967-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBell AP (2024) Artificial intelligence, automation, and the future of music education. Action Criticism Theory Music Educ 23(1):1\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://act.maydaygroup.org/articles/Bell23_1.pdf\u003c/span\u003e\u003cspan address=\"https://act.maydaygroup.org/articles/Bell23_1.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBettayeb AM, Abu Talib M, Sobhe Altayasinah AZ, Dakalbab F (2024) Exploring the impact of ChatGPT: Conversational AI in education. Front Educ 9:1379796. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2024.1379796\u003c/span\u003e\u003cspan address=\"10.3389/feduc.2024.1379796\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBobbe T, Oppici L, L\u0026uuml;neburg LM, Meinel C (2021) What early user involvement could look like\u0026mdash;Developing technology applications for piano teaching and learning. Multimodal Technol Interact 5(7). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 38. https://doi.org/10.3390/mti5070038\u003c/span\u003e\u003cspan address=\"Article 38. 10.3390/mti5070038\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBond M, Khosravi H, De Laat M, Buckingham Shum S, Gašević D, Sadiq S (2024) A meta-systematic review of artificial intelligence in education: Call for increased rigor, ethics, and theory. Int J Educational Technol High Educ 21., Article 4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-023-00436-z\u003c/span\u003e\u003cspan address=\"10.1186/s41239-023-00436-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBonnaire S, Gonz\u0026aacute;lez-Moreno P (2023) Developing metacognition as an instructional approach for enhancing young pianists' musical expressivity. Int J Music Educ 41(2):199\u0026ndash;210. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/02557614221097447\u003c/span\u003e\u003cspan address=\"10.1177/02557614221097447\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBorenstein J, Howard A (2021) Emerging challenges in AI and the need for AI ethics education. AI Ethics 1(1):61\u0026ndash;65. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s43681-020-00002-7\u003c/span\u003e\u003cspan address=\"10.1007/s43681-020-00002-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBorenstein M, Hedges LV, Higgins JPT, Rothstein HR (2009) Introduction to meta-analysis. Wiley\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCao Y (2024) Online courses for piano teaching: Comparing the effectiveness and impact of modern and traditional methods. Educ Inform Technol 29(9):11407\u0026ndash;11419. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-023-12225-0\u003c/span\u003e\u003cspan address=\"10.1007/s10639-023-12225-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCardona MA, Rodr\u0026iacute;guez RJ, Ishmael K (2023) Artificial intelligence and the future of teaching and learning. U.S. Department of Education. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://tech.ed.gov/files/2023/05/ai-future-of-teaching-and-learning-report.pdf\u003c/span\u003e\u003cspan address=\"https://tech.ed.gov/files/2023/05/ai-future-of-teaching-and-learning-report.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e Office of Educational Technology\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarnovalini F, Rod\u0026agrave; A (2020) Computational creativity and music generation systems: An introduction to state of the art. Front Artif Intell 3., Article 14. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/frai.2020.00014\u003c/span\u003e\u003cspan address=\"10.3389/frai.2020.00014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCasal-Otero L, Catala A, Fern\u0026aacute;ndez-Morante C, Taboada M, Cebreiro B, Barro S (2023) AI literacy in K-12: A systematic literature review. Int J STEM Educ 10(1) Article 17. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s40594-023-00418-7\u003c/span\u003e\u003cspan address=\"10.1186/s40594-023-00418-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCelik I (2023) Towards intelligent-TPACK: An empirical study on teachers' professional knowledge to ethically integrate artificial intelligence (AI)-based tools into education. Comput Hum Behav 138., Article 107468. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.chb.2022.107468\u003c/span\u003e\u003cspan address=\"10.1016/j.chb.2022.107468\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen JCW (2020) AI in music education: The impact of using artificial intelligence (AI) application to practise scales and arpeggios in a virtual learning environment. In \u003cem\u003eLearning environment and design: Current and future impacts\u003c/em\u003e (pp. 307\u0026ndash;322). Springer. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-981-15-8167-0_16\u003c/span\u003e\u003cspan address=\"10.1007/978-981-15-8167-0_16\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen X, Zou D, Cheng G, Xie H (2025) Exploring the impact of AI-assisted practice applications on music learners' performance, self-efficacy, and self-regulated learning. Front Psychol 15., Article 1675762. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2025.1675762\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2025.1675762\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Y, Liu Y (2024) Unlocking the beat: How AI tools drive music students' motivation, engagement, creativity and learning success. Eur J Educ 59(4) Article e12823. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/ejed.12823\u003c/span\u003e\u003cspan address=\"10.1111/ejed.12823\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Z (2023) Ethics and discrimination in artificial intelligence-enabled recruitment practices. Humanit Social Sci Commun 10., Article 567. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1057/s41599-023-02079-x\u003c/span\u003e\u003cspan address=\"10.1057/s41599-023-02079-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng L (2025) The impact of generative AI on school music education: Challenges and recommendations. Arts Educ Policy Rev 126(2):1\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10632913.2025.2378654\u003c/span\u003e\u003cspan address=\"10.1080/10632913.2025.2378654\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheung ACK, Slavin RE (2013) The effectiveness of educational technology applications for enhancing mathematics achievement in K-12 classrooms: A meta-analysis. Educational Res Rev 9:88\u0026ndash;113. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.edurev.2013.01.001\u003c/span\u003e\u003cspan address=\"10.1016/j.edurev.2013.01.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiu TKF (2024) Future research recommendations for transforming higher education with generative AI. Computers Education: Artif Intell 6., Article 100197. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2024.100197\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2024.100197\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiu TKF, Moorhouse BL, Chai CS, Ismailov M (2024) Teacher support and student motivation to learn with artificial intelligence (AI) based chatbot. Interact Learn Environ 32(6):3240\u0026ndash;3256. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10494820.2023.2172044\u003c/span\u003e\u003cspan address=\"10.1080/10494820.2023.2172044\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChoi S, Jang Y, Kim H (2023) Influence of pedagogical beliefs and perceived trust on teachers' acceptance of educational artificial intelligence tools. Int J Hum Comput Interact 39(5):910\u0026ndash;922. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10447318.2022.2049145\u003c/span\u003e\u003cspan address=\"10.1080/10447318.2022.2049145\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCivit M, Civit-Masot J, Cuadrado F, Escalona MJ (2022) A systematic review of artificial intelligence-based music generation: Scope, applications, and future trends. Expert Syst Appl 209:118190. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.eswa.2022.118190\u003c/span\u003e\u003cspan address=\"10.1016/j.eswa.2022.118190\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCohen J (1988) Statistical power analysis for the behavioral sciences, 2nd edn. Lawrence Erlbaum Associates\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCooper PK (2024) Music teachers' labeling accuracy and quality ratings of lesson plans by artificial intelligence (AI) and humans. Int J Music Educ 42(4):567\u0026ndash;582. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/02557614241249163\u003c/span\u003e\u003cspan address=\"10.1177/02557614241249163\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCrompton H, Burke D (2023) Artificial intelligence in higher education: The state of the field. Int J Educational Technol High Educ 20(1) Article 22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-023-00392-8\u003c/span\u003e\u003cspan address=\"10.1186/s41239-023-00392-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCui K (2023) Artificial intelligence and creativity: Piano teaching with augmented reality applications. Interact Learn Environ 31(10):7017\u0026ndash;7028. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10494820.2022.2073061\u003c/span\u003e\u003cspan address=\"10.1080/10494820.2022.2073061\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeja JA (2022) Piano learning and improvisation through adaptive visualisation and digital augmentation. In \u003cem\u003eCompanion proceedings of the 2022 conference on interactive surfaces and spaces\u003c/em\u003e (pp. 41\u0026ndash;45). ACM. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3567729.3567736\u003c/span\u003e\u003cspan address=\"10.1145/3567729.3567736\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDella Ventura M (2023) Intelligent tutoring system and learning: Complexity and resilience. In M. Dascalu, \u0026Oacute;. Mealha, \u0026amp; S. Virkus (Eds.), \u003cem\u003eSmart learning ecosystems as engines of the green and digital transition\u003c/em\u003e (pp. 25\u0026ndash;38). Springer. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-981-99-5540-4_3\u003c/span\u003e\u003cspan address=\"10.1007/978-981-99-5540-4_3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeng W, Wang L, Deng X (2024) Strategies for optimizing personalized learning pathways with artificial intelligence assistance. Int J Adv Comput Sci Appl 15(6):234\u0026ndash;245. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.14569/IJACSA.2024.0150626\u003c/span\u003e\u003cspan address=\"10.14569/IJACSA.2024.0150626\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeng Y, Lin Y (2025) The effect of ChatGPT on students' learning performance, learning perception, and higher-order thinking: Insights from a meta-analysis. Humanit Social Sci Commun 12., Article 4787. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41599-025-04787-y\u003c/span\u003e\u003cspan address=\"10.1038/s41599-025-04787-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEgger M, Davey Smith G, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315(7109):629\u0026ndash;634. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/bmj.315.7109.629\u003c/span\u003e\u003cspan address=\"10.1136/bmj.315.7109.629\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEuropean Commission (2022) DigCompEdu: The European framework for the digital competence of educators. Publications Office of the European Union. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://joint-research-centre.ec.europa.eu/digcompedu_en\u003c/span\u003e\u003cspan address=\"https://joint-research-centre.ec.europa.eu/digcompedu_en\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFitria TN (2023) The use of artificial intelligence in education (AIED): Can AI replace the teacher's role? Epigram 20(2):165\u0026ndash;187. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.32722/epigram.v20i2.5443\u003c/span\u003e\u003cspan address=\"10.32722/epigram.v20i2.5443\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eF\u0026ouml;rster A, Lepa S (2023) Digital musical instruments in special educational needs schools: Requirements from the music teachers' perspective and the status quo in Germany. ACM Trans Accessible Comput 16(3):29. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3616015\u003c/span\u003e\u003cspan address=\"10.1145/3616015\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026aacute;ndara D, Anahideh H, Ison M, Picchiarini L (2024) Inside the black box: Detecting and mitigating algorithmic bias across racialized groups in college student-success prediction. AERA Open 10(1):1\u0026ndash;18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/23328584241258741\u003c/span\u003e\u003cspan address=\"10.1177/23328584241258741\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGao WX, Zhang S, Zhang NX (2023) Generating fingerings for piano music with model-based reinforcement learning. Appl Sci 13(20). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 11321. https://doi.org/10.3390/app132011321\u003c/span\u003e\u003cspan address=\"Article 11321. 10.3390/app132011321\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGaunt H, Duffy C, Coric A, Gonz\u0026aacute;lez Delgado IR, Messas L, Pryimenko O, Sveidahl H (2021) Musicians as 'makers in society': A conceptual foundation for contemporary professional higher music education. Front Psychol 12:713648. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2021.713648\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2021.713648\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGedik G, Demirtas S (2023) The relationship between self-directed learning levels and piano learning styles of music students. Pamukkale Univ J Educ 59:145\u0026ndash;168. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.9779/pauefd.1196847\u003c/span\u003e\u003cspan address=\"10.9779/pauefd.1196847\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiraldo S, Cancino-Chac\u0026oacute;n C, Widmer G (2022) A review of computational models of expressive music performance. ACM-CSUR 54(3):1\u0026ndash;37. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3464424\u003c/span\u003e\u003cspan address=\"10.1145/3464424\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGr\u0026aacute;jeda A, C\u0026oacute;rdova P, C\u0026oacute;rdova JP, Laguna-Tapia A, Burgos J, Rodr\u0026iacute;guez L, S\u0026aacute;nchez K, Sanjin\u0026eacute;s A, Mamani P (2024) Embracing artificial intelligence in the arts classroom: Understanding student perceptions and emotional reactions to AI tools. Cogent Educ 11(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 2378271. https://doi.org/10.1080/2331186X.2024.2378271\u003c/span\u003e\u003cspan address=\"Article 2378271. 10.1080/2331186X.2024.2378271\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao J (2024) Online piano learning game design method: Piano music style recognition based on CRNNH. Entertainment Comput 50., Article 100645. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.entcom.2024.100645\u003c/span\u003e\u003cspan address=\"10.1016/j.entcom.2024.100645\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarrer M, Cuijpers P, Furukawa TA, Ebert DD (2021) Doing meta-analysis with R: A hands-on guide. Chapman \u0026amp; Hall/CRC. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1201/9781003107347\u003c/span\u003e\u003cspan address=\"10.1201/9781003107347\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHasanein AM, Sobaih AEE (2023) Drivers and consequences of ChatGPT use in higher education: Key stakeholder perspectives. Eur J Invest Health Psychol Educ 13(11):2599\u0026ndash;2614. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ejihpe13110181\u003c/span\u003e\u003cspan address=\"10.3390/ejihpe13110181\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHatfield JL, Halvari H, Lemyre PN (2017) Instrumental practice in the contemporary music academy: A three-phase cycle of self-regulated learning in music students. Musicae Sci 21(3):316\u0026ndash;337. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1029864917690206\u003c/span\u003e\u003cspan address=\"10.1177/1029864917690206\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHenry A, Wiratama V, Afilipoaie A, Ranaivoson H, Arriv\u0026eacute; E (2024) Impacts of AI on music consumption and fairness. Music Sci 7:1\u0026ndash;15. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/27523543241269047\u003c/span\u003e\u003cspan address=\"10.1177/27523543241269047\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHiggins JPT, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21(11):1539\u0026ndash;1558. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/sim.1186\u003c/span\u003e\u003cspan address=\"10.1002/sim.1186\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHiggins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (eds) (2019) \u003cem\u003eCochrane handbook for systematic reviews of interventions\u003c/em\u003e (2nd ed.). Wiley. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/9781119536604\u003c/span\u003e\u003cspan address=\"10.1002/9781119536604\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHo Weatherly KIC, Wright B, Lee EY (2024) Digital game-based learning in music education: A systematic review between 2011 and 2023. Res Stud Music Educ 46(3):456\u0026ndash;478. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1321103X241270819\u003c/span\u003e\u003cspan address=\"10.1177/1321103X241270819\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolmes W, Bialik M, Fadel C (2022) Artificial intelligence in education: Promise and implications for teaching and learning, 2nd edn. Center for Curriculum Redesign\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolmes W, Porayska-Pomsta K, Holstein K, Sutherland E, Baker T, Shum SB, Santos OC, Rodrigo MT, Cukurova M, Bittencourt II, Koedinger KR (2022) Ethics of AI in education: Towards a community-wide framework. Int J Artif Intell Educ 32(3):504\u0026ndash;526. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40593-021-00239-1\u003c/span\u003e\u003cspan address=\"10.1007/s40593-021-00239-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang N, Ding X (2022) Piano music teaching under the background of artificial intelligence. Wirel Commun Mob Comput 2022(1) Article 5816453. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2022/5816453\u003c/span\u003e\u003cspan address=\"10.1155/2022/5816453\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHwang GJ, Xie H, Wah BW, Gašević D (2020) Vision, challenges, roles and research issues of artificial intelligence in education. Computers Education: Artif Intell 1:100001. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2020.100001\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2020.100001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJeon J, Lee S (2023) Large language models in education: A focus on the complementary relationship between human teachers and ChatGPT. Educ Inform Technol 28(12):15873\u0026ndash;15892. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-023-11834-1\u003c/span\u003e\u003cspan address=\"10.1007/s10639-023-11834-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang H (2024) Machine learning in teaching piano touch technique and tonal expression. J Electr Syst 20(4 SI):404\u0026ndash;415\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJin L, Lin B, Hong M, Zhang K, So HJ (2025) Exploring the impact of an LLM-powered teachable agent on learning gains and cognitive load in music education. \u003cem\u003earXiv preprint arXiv:2504.00636\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2504.00636\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2504.00636\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalkanoğlu B (2025) The effect of using technology in music education and training on academic achievement: A meta-analysis study. Educ Sci 50(221):213\u0026ndash;236. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.15390/EB.2024.13320\u003c/span\u003e\u003cspan address=\"10.15390/EB.2024.13320\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKandemir \u0026Ouml;, Yokuş T (2023) The effect of learning strategies on piano performance self-efficacy levels and performance success. Cukurova Univ Fac Educ J 52(2):446\u0026ndash;470. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.14812/cuefd.1265516\u003c/span\u003e\u003cspan address=\"10.14812/cuefd.1265516\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarolus J, Sylupp J, Schmidt A (2023) EyePiano: Leveraging gaze for reflective piano learning. In \u003cem\u003eProceedings of the designing interactive systems conference (DIS 2023)\u003c/em\u003e (pp. 1209\u0026ndash;1223). ACM. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3563657.3596122\u003c/span\u003e\u003cspan address=\"10.1145/3563657.3596122\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKasneci E, Sessler K, K\u0026uuml;chemann S, Bannert M, Dementieva D, Fischer F, Gasser U, Groh G, G\u0026uuml;nnemann S, H\u0026uuml;llermeier E, Kruber S, Samuel A (2023) ChatGPT for good? On opportunities and challenges of large language models for education. \u003cem\u003eLearning and Individual Differences\u003c/em\u003e, \u003cem\u003e103\u003c/em\u003e, Article 102274. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.lindif.2023.102274\u003c/span\u003e\u003cspan address=\"10.1016/j.lindif.2023.102274\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKehoe A (2023) AI and the music teacher: Navigating a new pedagogical landscape. J Music Technol Educ 16(1):45\u0026ndash;62. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1386/jmte_00067_1\u003c/span\u003e\u003cspan address=\"10.1386/jmte_00067_1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKonovalova I, Breslavets H, Riabukha N, Polska I, Shchepakin V, Roshchenko O (2025) The evolution of world music pedagogy in the information society. BRAIN: Broad Res Artif Intell Neurosci 16(1 Suppl):99\u0026ndash;116. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18662/brain/16.1Sup/929\u003c/span\u003e\u003cspan address=\"10.18662/brain/16.1Sup/929\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLam CK (2024) Technology-enhanced creativity in K-12 music education: A scoping review. Int J Music Educ 42(1):89\u0026ndash;108. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/02557614231194073\u003c/span\u003e\u003cspan address=\"10.1177/02557614231194073\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi J, Wang Y (2024) Enhancing self-directed learning in music education through generative AI: A quantitative analysis. Educ Inform Technol 29(8):10234\u0026ndash;10256. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-023-12178-4\u003c/span\u003e\u003cspan address=\"10.1007/s10639-023-12178-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi S, Timmers R (2021) Teaching and learning of piano timbre through teacher-student interactions in lessons. Front Psychol 12., Article 576056. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2021.576056\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2021.576056\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi W (2022) Analysis of piano performance characteristics by deep learning and artificial intelligence and its application in piano teaching. Front Psychol 12., Article 751406. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2021.751406\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2021.751406\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLin CC, Huang AY, Lu OH (2023) Artificial intelligence in intelligent tutoring systems toward sustainable education: A systematic review. Smart Learn Environ 10(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 41. https://doi.org/10.1186/s40561-023-00260-y\u003c/span\u003e\u003cspan address=\"Article 41. 10.1186/s40561-023-00260-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLipsey MW, Wilson DB (2001) Practical meta-analysis. SAGE\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu X, Shao X (2024) Modern mobile learning technologies in online piano education: Online educational course design and impact on learning. Interact Learn Environ 32(4):1279\u0026ndash;1290. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10494820.2022.2118787\u003c/span\u003e\u003cspan address=\"10.1080/10494820.2022.2118787\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu Y, Tran VC, Kiss B, Oo TZ, Szab\u0026oacute; N, J\u0026oacute;zsa K (2025) The use and effectiveness of digital tools in elementary music education: A systematic review. Music Educ Res 27(1):45\u0026ndash;68. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/20592043251363338\u003c/span\u003e\u003cspan address=\"10.1177/20592043251363338\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLong D, Magerko B (2020) What is AI literacy? Competencies and design considerations. In \u003cem\u003eProceedings of the 2020 CHI conference on human factors in computing systems\u003c/em\u003e (pp. 1\u0026ndash;16). ACM. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3313831.3376727\u003c/span\u003e\u003cspan address=\"10.1145/3313831.3376727\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026oacute;pez-\u0026Iacute;\u0026ntilde;iguez G, McPherson GE (2020) Applying self-regulated learning and self-determination theory to optimize the performance of a concert soloist. Front Psychol 11., Article 385. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2020.00385\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2020.00385\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu L (2025) AI-powered intelligent music education systems for real-time feedback and performance assessment. Int J Inf Commun Technol 26(18):33\u0026ndash;47\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuckin R, Holmes W, Griffiths M, Forcier LB (2016) Intelligence unleashed: An argument for AI in education. Pearson Educ. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.pearson.com/corporate/about-pearson/innovation/smarter-digital-tools/intelligence-unleashed.html\u003c/span\u003e\u003cspan address=\"https://www.pearson.com/corporate/about-pearson/innovation/smarter-digital-tools/intelligence-unleashed.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMa R, Ma R (2022) Piano education online: Challenges and solutions. Educ Inform Technol 28:1359\u0026ndash;1372. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-022-11242-x\u003c/span\u003e\u003cspan address=\"10.1007/s10639-022-11242-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMacRitchie J, Garrido S (2019) Measuring music performance self-efficacy in orchestral musicians. Psychol Music 47(6):856\u0026ndash;870. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0305735618789572\u003c/span\u003e\u003cspan address=\"10.1177/0305735618789572\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatsui R, Kumaki M, Takegawa Y (2022) Proposal of a piano learning support system considering gamification based on reconsolidation. In \u003cem\u003eLearning and collaboration technologies: Designing the learner and teacher experience\u003c/em\u003e (pp. 113\u0026ndash;125). Springer. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-05657-4_9\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-05657-4_9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcPherson GE, McCormick J (2006) Self-efficacy and music performance. Psychol Music 34(3):322\u0026ndash;336. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0305735606064841\u003c/span\u003e\u003cspan address=\"10.1177/0305735606064841\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMerch\u0026aacute;n S\u0026aacute;nchez-Jara JF, Gonz\u0026aacute;lez Guti\u0026eacute;rrez S, Cruz Rodr\u0026iacute;guez J, Syroyid S, B (2024) Artificial intelligence\u0026ndash;assisted music education: A critical synthesis of challenges and opportunities. Educ Sci 14(11):1171. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/educsci14111171\u003c/span\u003e\u003cspan address=\"10.3390/educsci14111171\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiao F, Holmes W (2021) AI and education: Guidance for policy-makers. United Nations Educational. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://unesdoc.unesco.org/ark:/48223/pf0000376709\u003c/span\u003e\u003cspan address=\"https://unesdoc.unesco.org/ark:/48223/pf0000376709\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e Cultural Organization\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiksza P (2015) The effect of self-regulation instruction on the performance achievement, musical self-efficacy, and practicing of advanced wind players. Psychol Music 43(2):219\u0026ndash;243. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0305735613500832\u003c/span\u003e\u003cspan address=\"10.1177/0305735613500832\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMishra P, Warr M, Islam R (2023) TPACK in the age of ChatGPT and generative AI. J Digit Learn Teacher Educ 39(4):235\u0026ndash;251. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/21532974.2023.2247480\u003c/span\u003e\u003cspan address=\"10.1080/21532974.2023.2247480\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaseer F, Khan MN, Tahir M, Addas A, Aejaz SH (2024) Integrating deep learning techniques for personalized learning pathways in higher education. Heliyon 10(11) Article e32145. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.heliyon.2024.e32145\u003c/span\u003e\u003cspan address=\"10.1016/j.heliyon.2024.e32145\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNg DTK, Leung JKL, Chu SKW, Qiao MS (2021) Conceptualizing AI literacy: An exploratory review. \u003cem\u003eComputers and Education: Artificial Intelligence\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e, Article 100041. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2021.100041\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2021.100041\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNg DTK, Leung JKL, Su J, Ng RCW, Chu SKW (2023) Teachers' AI digital competencies and twenty-first century skills in the post-pandemic world. Education Tech Research Dev 71(1):33\u0026ndash;59. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11423-023-10203-6\u003c/span\u003e\u003cspan address=\"10.1007/s11423-023-10203-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNhan LK (2025) The role of AI in enhancing personalised learning, automated assessment, intelligent tutoring, and student engagement. Turkish Online J Distance Educ 26(4):55\u0026ndash;76. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.17718/tojde.1581237\u003c/span\u003e\u003cspan address=\"10.17718/tojde.1581237\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNiu Y (2021) Penetration of multimedia technology in piano teaching and performance based on complex network. Math Probl Eng 2021(1) Article 8872227. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2021/8872227\u003c/span\u003e\u003cspan address=\"10.1155/2021/8872227\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eObermeyer Z, Powers B, Vogeli C, Mullainathan S (2019) Dissecting racial bias in an algorithm used to manage the health of populations. Science 366(6464):447\u0026ndash;453. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/science.aax2342\u003c/span\u003e\u003cspan address=\"10.1126/science.aax2342\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOuyang F, Jiao P (2021) Artificial intelligence in education: The three paradigms. \u003cem\u003eComputers and Education: Artificial Intelligence\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e, Article 100020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2021.100020\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2021.100020\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOuyang M (2023) Employing mobile learning in music education. Educ Inform Technol 28:5241\u0026ndash;5257. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-022-11353-5\u003c/span\u003e\u003cspan address=\"10.1007/s10639-022-11353-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePage MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, Chou R, Glanville J, Grimshaw JM, Hr\u0026oacute;bjartsson A, Lalu MM, Li T, Loder EW, Mayo-Wilson E, McDonald S, Moher D (2021) The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ 372 Article n71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/bmj.n71\u003c/span\u003e\u003cspan address=\"10.1136/bmj.n71\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePanadero E (2020) A review of self-regulated learning: Six models and four directions for research. Front Psychol 11., Article 108. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2020.00108\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2020.00108\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePhanichraksaphong V, Tsai W-H (2021) Automatic evaluation of piano performances for STEAM education. Appl Sci 11(7) Article 3042. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/app11073042\u003c/span\u003e\u003cspan address=\"10.3390/app11073042\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePigott TD (2012) Advances in meta-analysis. Springer. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-1-4614-2278-5\u003c/span\u003e\u003cspan address=\"10.1007/978-1-4614-2278-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePintrich PR (1999) The role of motivation in promoting and sustaining self-regulated learning. Int J Educational Res 31(6):459\u0026ndash;470. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0883-0355(99)00015-4\u003c/span\u003e\u003cspan address=\"10.1016/S0883-0355(99)00015-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaudenbush SW (2009) Analyzing effect sizes: Random-effects models. In: Cooper H, Hedges LV, Valentine JC (eds) The handbook of research synthesis and meta-analysis, 2nd edn. Russell Sage Foundation, pp 295\u0026ndash;315\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReeves TC, Lin L (2020) The research we have is not the research we need. Education Tech Research Dev 68:1991\u0026ndash;2001. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11423-020-09811-3\u003c/span\u003e\u003cspan address=\"10.1007/s11423-020-09811-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRegier BJ (2022) Development and validation of the jazz self-efficacy scale. J Res Music Educ 70(1):67\u0026ndash;88. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/00224294211054321\u003c/span\u003e\u003cspan address=\"10.1177/00224294211054321\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRegelski TA (2002) On methodolatry and music teaching as critical and reflective praxis. Philos Music Educ Rev 10(2):102\u0026ndash;123 [suspicious link removed]\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRitchie L, Williamon A (2011) Measuring distinct types of musical self-efficacy. Psychol Music 39(3):328\u0026ndash;344. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0305735610374895\u003c/span\u003e\u003cspan address=\"10.1177/0305735610374895\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRu T (2025) Deep learning-based automatic piano note recognition and performance generation system. Int J High Speed Electron Syst 34(1) Article 2540814. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1142/S0129156425408147\u003c/span\u003e\u003cspan address=\"10.1142/S0129156425408147\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRudolph J, Tan S, Tan S (2023) ChatGPT: Bullshit spewer or the end of traditional assessments in higher education? J Appl Learn Teach 6(1):342\u0026ndash;363. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.37074/jalt.2023.6.1.9\u003c/span\u003e\u003cspan address=\"10.37074/jalt.2023.6.1.9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScherer R, Siddiq F, Tondeur J (2019) The technology acceptance model (TAM): A meta-analytic structural equation modeling approach to explaining teachers' adoption of digital technology in education. Comput Educ 128:13\u0026ndash;35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2018.09.009\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2018.09.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSelwyn N (2023) Should robots replace teachers? AI and the future of education. Polity\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShaw BP (2024) Artificial intelligence and assessment: Three implications for music educators. Music Educators J 111(2):45\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/00274321241296118\u003c/span\u003e\u003cspan address=\"10.1177/00274321241296118\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSlavin RE (1986) Best-evidence synthesis: An alternative to meta-analytic and traditional reviews. Educational Researcher 15(9):5\u0026ndash;11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3102/0013189X015009005\u003c/span\u003e\u003cspan address=\"10.3102/0013189X015009005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong L (2024) Design and implementation of remote piano teaching based on attention-induced multi-head convolutional neural network optimized with hunter-prey optimization. Int J Comput Intell Syst 17., Article 379. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s44196-023-00379-3\u003c/span\u003e\u003cspan address=\"10.1007/s44196-023-00379-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpahn C, Voltmer E, Mornell A, Nusseck M (2023) Health-promoting strategies and self-regulated practice behavior in music students. Front Psychol 14., Article 1130610. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2023.1130610\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2023.1130610\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteenbergen-Hu S, Cooper H (2014) A meta-analysis of the effectiveness of intelligent tutoring systems on college students' academic learning. J Educ Psychol 106(2):331\u0026ndash;347. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1037/a0034752\u003c/span\u003e\u003cspan address=\"10.1037/a0034752\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStrielkowski W, Grebennikova V, Lisovskiy A, Rakhimova G, Vasileva T (2025) AI-driven adaptive learning for sustainable educational transformation. Sustain Dev 33(2):1921\u0026ndash;1947. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/sd.3190\u003c/span\u003e\u003cspan address=\"10.1002/sd.3190\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu J, Zhong Y (2024) Artificial intelligence in teaching and teacher professional development: A systematic review. Computers Education: Artif Intell 6., Article 100212. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2024.100212\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2024.100212\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTang J (2025) The role of artificial intelligence in improving the efficiency of piano online learning. J Comput Methods Sci Eng 25(2):1504\u0026ndash;1518. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/14727978241300381\u003c/span\u003e\u003cspan address=\"10.1177/14727978241300381\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTondeur J, van Braak J, Ertmer PA, Ottenbreit-Leftwich A (2017) Understanding the relationship between teachers' pedagogical beliefs and technology use in education: A systematic review of qualitative evidence. Education Tech Research Dev 65(3):555\u0026ndash;575. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11423-016-9481-2\u003c/span\u003e\u003cspan address=\"10.1007/s11423-016-9481-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTuomi I (2024) In search of artificial intelligence (AI) literacy in teacher education: A scoping review. Computers Education: Artif Intell 6., Article 100200. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2024.100200\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2024.100200\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNESCO, Organization C (2021) Recommendation on the ethics of artificial intelligence. United Nations Educational. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.unesco.org/en/articles/recommendation-ethics-artificial-intelligence\u003c/span\u003e\u003cspan address=\"https://www.unesco.org/en/articles/recommendation-ethics-artificial-intelligence\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNESCO (2023) \u003cem\u003eGuidance for generative AI in education and research\u003c/em\u003e. United Nations Educational, Scientific and Cultural Organization. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://unesdoc.unesco.org/ark:/48223/pf0000386693\u003c/span\u003e\u003cspan address=\"https://unesdoc.unesco.org/ark:/48223/pf0000386693\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNESCO (2024) \u003cem\u003eAI competency framework for teachers\u003c/em\u003e. United Nations Educational, Scientific and Cultural Organization. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://unesdoc.unesco.org/ark:/48223/pf0000391104\u003c/span\u003e\u003cspan address=\"https://unesdoc.unesco.org/ark:/48223/pf0000391104\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVanLehn K (2011) The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychol 46(4):197\u0026ndash;221. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/00461520.2011.611369\u003c/span\u003e\u003cspan address=\"10.1080/00461520.2011.611369\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVenkatesh V, Thong JY, Xu X (2012) Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. MIS Q 36(1):157\u0026ndash;178. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/41410412\u003c/span\u003e\u003cspan address=\"10.2307/41410412\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang H, Tlili A (2024) Does ChatGPT enhance student learning? A systematic review and meta-analysis of experimental studies. Comput Educ 223., Article 105148. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2024.105148\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2024.105148\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang S (2025) Hybrid models of piano instruction: How combining traditional teaching methods with personalized AI feedback affects learners' skill acquisition, self-efficacy, and academic locus of control. Educ Inform Technol 30(2):1245\u0026ndash;1268. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-025-13359-1\u003c/span\u003e\u003cspan address=\"10.1007/s10639-025-13359-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang YH (2023) Exploring the effects of using various designs of game-based materials on music learning. Interact Learn Environ 31(5):2650\u0026ndash;2664. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10494820.2021.1916768\u003c/span\u003e\u003cspan address=\"10.1080/10494820.2021.1916768\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWoolf BP (2010) Building intelligent interactive tutors: Student-centered strategies for revolutionizing e-learning. Morgan Kaufmann. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/B978-0-12-373594-2.X0001-6\u003c/span\u003e\u003cspan address=\"10.1016/B978-0-12-373594-2.X0001-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiao X (2024) Innovative application of artificial intelligence in piano music education and evaluation. J Comput Methods Sci Eng 24(3):1473\u0026ndash;1486. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3233/JCM-247311\u003c/span\u003e\u003cspan address=\"10.3233/JCM-247311\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu W, Ouyang F (2022) The application of AI technologies in STEM education: A systematic review from 2011 to 2021. Int J STEM Educ 9(1) Article 59. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s40594-022-00377-5\u003c/span\u003e\u003cspan address=\"10.1186/s40594-022-00377-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang SJ, Ogata H, Matsui T, Chen NS (2021) Human-centered artificial intelligence in education: Seeing the invisible through the visible. \u003cem\u003eComputers and Education: Artificial Intelligence\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e, Article 100008. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2021.100008\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2021.100008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang W, Shen L, Huang CF, Lee J, Zhao X (2024) Development status, frontier hotspots, and technical evaluations in the field of AI music composition since the 21st century: A systematic review. IEEE Access 12:89452\u0026ndash;89466. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2024.3419789\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2024.3419789\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang Y, Welch G (2023) A systematic literature review of Chinese music education studies during 2007 to 2019. Int J Music Educ 41(2):175\u0026ndash;198. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/02557614221094822\u003c/span\u003e\u003cspan address=\"10.1177/02557614221094822\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYilmaz R, Yilmaz FGK (2023) The effect of generative artificial intelligence (AI)-based tool use on students' computational thinking skills, programming self-efficacy and motivation. Computers Education: Artif Intell 4:100147. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2023.100147\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2023.100147\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYin Y, Dong Y (2025) An artificial intelligence-based interactive learning environment for music education in China: Traditional Chinese music and its contemporary development as a way to increase cultural capital. Eur J Educ 60(1) Article e12858. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/ejed.12858\u003c/span\u003e\u003cspan address=\"10.1111/ejed.12858\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu J (2021) Automatic evaluation system for piano performance based on the internet of things technology under the background of artificial intelligence. Math Probl Eng 2021(1) Article 8501960. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2021/8501960\u003c/span\u003e\u003cspan address=\"10.1155/2021/8501960\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZawacki-Richter O, Mar\u0026iacute;n VI, Bond M, Gouverneur F (2019) Systematic review of research on artificial intelligence applications in higher education. Int J Educational Technol High Educ 16., Article 39. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-019-0171-0\u003c/span\u003e\u003cspan address=\"10.1186/s41239-019-0171-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZelenak MS (2024) Self-efficacy and music performance: A meta-analysis. Psychol Music 52(2):147\u0026ndash;162. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/03057356231222432\u003c/span\u003e\u003cspan address=\"10.1177/03057356231222432\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhai C, Wibowo S, Li LD (2024) The effects of over-reliance on AI dialogue systems on students' cognitive abilities: A systematic review. Smart Learn Environ 11(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 28. https://doi.org/10.1186/s40561-024-00310-9\u003c/span\u003e\u003cspan address=\"Article 28. 10.1186/s40561-024-00310-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhai X, Xu J (2022) AI-supported learning and assessment: A review of emerging educational applications. Comput Educ 3., Article 100072. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2022.100072\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2022.100072\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang K, Aslan AB (2021) AI technologies for education: Recent research \u0026amp; future directions. Computers Education: Artif Intell 2., Article 100025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2021.100025\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2021.100025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang M (2025) Advancing deep learning for expressive music composition and performance modeling. Sci Rep 15(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 28007. https://doi.org/10.1038/s41598-025-12345-6\u003c/span\u003e\u003cspan address=\"Article 28007. 10.1038/s41598-025-12345-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Q (2025) \u003cem\u003eIntegrating piano learning software with traditional piano teaching\u003c/em\u003e [Doctoral dissertation, University of Miami]. ProQuest Dissertations Publishing\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao X, Guo Z, Liu S (2021) Exploring key competencies and professional development of music teachers in primary schools in the era of artificial intelligence. Sci Program 2021(1):5097003. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2021/5097003\u003c/span\u003e\u003cspan address=\"10.1155/2021/5097003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao X, Liu Y, Chen Z (2023) Audio\u0026ndash;visual fusion-based piano performance assessment using deep learning. Appl Sci 13(5) Article 2890. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/app13052890\u003c/span\u003e\u003cspan address=\"10.3390/app13052890\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou Z, Wu Y, Wu Z, Zhang X, Yuan R, Ma Y, Zhu J, Guo Y (2024) Can LLMs reason in music? An evaluation of LLMs' capability of music understanding and generation. arXiv preprint arXiv 240721531. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2407.21531\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2407.21531\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZi H (2023) Intelligent piano teaching system and method based on cloud platform. In \u003cem\u003e2023 International conference on distributed computing and electrical circuits and electronics (ICDCECE)\u003c/em\u003e (pp. 1\u0026ndash;7). IEEE. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICDCECE57866.2023.10150890\u003c/span\u003e\u003cspan address=\"10.1109/ICDCECE57866.2023.10150890\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZimmerman BJ (2002) Becoming a self-regulated learner: An overview. Theory into Pract 41(2):64\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1207/s15430421tip4102_2\u003c/span\u003e\u003cspan address=\"10.1207/s15430421tip4102_2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Handan College","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Artificial intelligence, Piano education, Learning efficacy, Ethics, Teacher role transformation","lastPublishedDoi":"10.21203/rs.3.rs-8998934/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8998934/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTechnology has begun reshaping piano pedagogy, moving instructional practices away from traditional apprenticeship models toward data-informed learning environments. This systematic review and meta-analysis, conducted according to PRISMA guidelines, examined empirical research on artificial intelligence (AI) in piano education published between 2020 and 2025. Searches of Web of Science, Scopus, and ProQuest identified 391 initial records, of which 62 studies (N\u0026thinsp;=\u0026thinsp;2,812 participants) satisfied inclusion criteria. Random-effects meta-analysis showed a moderate pooled effect size (d\u0026thinsp;=\u0026thinsp;0.442, 95% CI [0.398, 0.486], p \u0026lt; .0001) for AI-supported interventions. Subgroup analyses revealed similar effects across technical skills (d\u0026thinsp;=\u0026thinsp;0.462), musical expression (d\u0026thinsp;=\u0026thinsp;0.432), and learning motivation (d\u0026thinsp;=\u0026thinsp;0.434), with considerable heterogeneity (I\u0026sup2; = 61.1%) pointing to context-dependent effectiveness. Publication bias assessment indicated no marked asymmetry (Egger\u0026rsquo;s test: p = .191). Results show that AI tools correlate with favorable outcomes in psychomotor skill acquisition and, importantly, musical expressiveness\u0026mdash;countering assumptions that AI benefits apply only to quantifiable outcomes. Evidence on long-term learning trajectories and higher-order artistic judgment remains sparse. Ethical issues include algorithmic bias, data privacy, and potential homogenization of creative expression. The analysis points to a reconfiguration rather than replacement of the teacher\u0026rsquo;s role, with educators serving as aesthetic mentors and pedagogical decision-makers. This meta-analysis offers the first empirical benchmark for AI-enhanced piano education, underscoring the need for context-sensitive integration, thorough teacher training, and ethically grounded research approaches.\u003c/p\u003e","manuscriptTitle":"A Systematic Review of Artificial Intelligence in Piano Education: Efficacy,Ethics, and the Transformation of the Teacher's Role (2020–2025)","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-08 16:43:15","doi":"10.21203/rs.3.rs-8998934/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d493c778-d615-4344-963a-03f6c130c4dc","owner":[],"postedDate":"March 8th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":63711480,"name":"Music"},{"id":63711481,"name":"Artificial Intelligence and Machine Learning"},{"id":63711482,"name":"Educational Philosophy and Theory"}],"tags":[],"updatedAt":"2026-03-08T16:43:15+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-08 16:43:15","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8998934","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8998934","identity":"rs-8998934","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}