Power, Pedagogy, and Algorithms: An Exploratory Study of EAP Instructors' Critical Engagement with Generative AI

Power, Pedagogy, and Algorithms: An Exploratory Study of EAP Instructors' Critical Engagement with Generative AI | 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 Research Article Power, Pedagogy, and Algorithms: An Exploratory Study of EAP Instructors' Critical Engagement with Generative AI Rana Haidar, Athena Tassis This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7014667/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 This qualitative exploratory study examines how nine English for Academic Purposes (EAP) instructors at post-secondary institutions in Ontario, Canada, are engaging with the growing presence of Generative Artificial Intelligence (GenAI) in academic writing instruction. Using a qualitative case study approach, researchers conducted semi-structured interviews with EAP instructors to explore their perceptions, practices, and the institutional and ideological factors influencing GenAI adoption. Data were analyzed using reflexive thematic analysis to ensure rigorous interpretation of participant experiences. The analysis was informed by the PICRAT model, the Unified Theory of Acceptance and Use of Technology (UTAUT), and critical pedagogy. The findings revealed a wide range of engagement levels that were influenced by factors such as pedagogical viability, institutional policy, job precarity, and ethical concerns. Instructors primarily used GenAI to enhance existing pedagogical approaches while developing students' critical literacy. Key concerns addressed by participants included risks to intellectual authenticity, diminished critical thinking, and the marginalization of multilingual learners. Institutional barriers such as unclear policies and limited professional development opportunities were found to hinder meaningful integration. While GenAI was often viewed as a useful supplement, participants emphasized the need for pedagogically sound, student-centered implementation that protects human intellectual agency. This research addresses a critical gap in understanding GenAI integration specifically within Canadian EAP contexts. The study offers practical recommendations for developing comprehensive institutional policies, designing targeted professional development programs, and creating pedagogical frameworks that balance technological integration with critical thinking development. Findings are context-specific to Ontario institutions and warrant investigation across diverse educational settings. This study contributes to ongoing discussions about responsible AI integration in higher education and underscores the importance of addressing broader issues of equity and the educational purpose of GenAI in EAP instruction. Educational Philosophy and Theory Generative AI English for Academic Purposes Instructor Perceptions Critical Pedagogy PICRAT UTAUT Introduction The increasing availability of Generative Artificial Intelligence (GenAI) tools like ChatGPT, DeepSeek, Google Gemini, Perplexity, and Claude presents new challenges and affordances for English for Academic Purposes (EAP) instructors as they navigate potential implications for the instruction of academic writing. Recent scholarship shows Large Language Models (LLMs) can support academic writing by offering interactive, personalized learning experiences that adapt to each user's writing style and pace (Pokrivcakova, 2019; Wang, 2024), which mirrors one-on-one tutoring and helps writers improve through immediate feedback and hands-on practice (Crompton et al., 2024). Also, the GenAI tools’ constant availability makes learning more flexible and accessible (Lin, 2024), and studies show they can minimize language learning anxiety and improve learner motivation (Asio & Suero, 2024; Klimova & Pikhart, 2025). On the flip side, scholars argue that AI-generated content can lead to over-reliance on automation, ultimately reducing students' ability to develop independent writing skills and critical thinking (Marzuki et al., 2023; Barrett & Pack, 2023; Kohnke et al., 2025). Additionally, a major concern for language instructors is the reinforcement of dominant linguistic norms and ideas through GenAI applications that might impact non users of English and marginalize indigenous languages, reducing accessibility (Nyaaba et al., 2024). Specifically, AI tools tend to generate content that is biased toward dominant Western cultural norms ultimately limiting space for non-Western and marginalized perspectives (Creely & Henderson 2025). This seismic shift in the integration of new technologies in writing instruction not only impacts students, but also plays a pivotal role in shaping curriculum, pedagogy, and how educators integrate these applications into their daily practices. Notably, the generative characteristics of LLMs have the potential to transform our conception of authorship, leading to what scholars like Eaton (2023) term a "postplagiarism" paradigm where collaborative creation between humans and artificial intelligence becomes standard practice. Understanding how writing instructors perceive this looming paradigm shift and their strategies for navigating it is essential for developing appropriate responses to human-AI collaborative authorship in academic contexts and defining the evolving role of writing instructors in an AI infused academic world (Baidoo-Anu & Owusu Ansah, 2023; Kinzie, 2024). Chan and Tsi (2024) predict that GenAI can enhance educational efficiency by automating administrative tasks while supporting teachers with course design, information gathering, content generation, and assessment processes, enabling them to focus on higher-level teaching responsibilities, while Cacho (2024) suggests that GenAI can help streamline content creation, curate materials, and support learning preferences. Despite this burgeoning literature on AI in education, there remains limited research examining how academic writing instructors perceive and incorporate generative AI tools into their teaching practices. This study addresses this research gap by investigating two key questions: 1) What perceptions do English for Academic Purposes instructors hold regarding the integration of Generative Artificial Intelligence in education? and 2) To what extent and in what ways are these instructors incorporating GenAI tools into their pedagogical practices and instructional approaches? Through exploring these questions, our research offers valuable insights into the current landscape of GenAI adoption in academic writing instruction in Canadian higher education, revealing both emerging practices and instructors' predominantly cautious responses—characterized by critical evaluation, skepticism about pedagogical value, and tentative acceptance rather than widespread enthusiasm. By critically interrogating whether and where AI tools belong in the EAP classroom, examining the conditions under which these technologies are adopted, and interrogating the assumptions that shape their use, we contribute to ongoing discourse about if and how AI tools can be leveraged responsibly to support, rather than replace, human instruction in EAP contexts. Literature Review Evolving Technologies and the role of AI in Academic Writing The role of AI in academic writing instruction is part of a broader historical trend in language and literacy education, where emerging technologies have repeatedly reshaped teaching and pedagogy. Earlier tools such as computer-assisted language learning (CALL) provided users with structured feedback and grammar correction (Huber, 2018). The development of CALL paved the way for more sophisticated systems in the 1990’s such as Automated Writing Evaluation (AWE), which uses natural language processing to provide immediate, formative feedback on various aspects of student writing and is often used for iterative drafting (Hibert, 2019). However, the feedback provided was often generic and less effective at fostering higher-order writing skills without teacher mediation (Fu et al., 2024). The impact of these early tools was modest, despite claims that these new technologies would significantly disrupt language learning. Traditional teacher-led instruction remained central, partly because the educational structures and pedagogical approaches were not fundamentally disrupted by these new tools (Huber, 2018). By the early 1990s, tools like Microsoft Word introduced built-in grammar checks, and by the 2010s, assistants like Grammarly and WordTune expanded functionality offering real-time suggestions on writing style yet still focusing mainly on surface-level corrections (Gayed et al., 2022). These innovations allowed some instructors to shift their focus away from correcting grammar and spelling, instead spending more instructional time on supporting their students to develop content and higher-order thinking skills (Hibert, 2019). As these foundational tools evolved and instructors adapted their pedagogical focus, the emergence of GenAI marked a new phase in writing instruction. Researchers have begun examining how GenAI tools are used in writing instruction; emerging studies emphasize that AI can support not only efficiency, but also deeper engagement and higher-order thinking in writing, especially when instructors design tasks that prompt students to reflect on and refine AI-generated content (Yatani et al., 2024). Some recent studies show that generative AI tools can enhance students’ cognitive engagement in writing, particularly by supporting activities such as sharing opinions, explaining concepts, and engaging in analysis. This effect is most pronounced when students interact with AI in a reflective and iterative manner, rather than using it only for basic tasks (Hu et al., 2024; Nguyen et al., 2024; Jin et al., 2025). Emergent literature that shows advanced uses of AI in writing—such as transforming and refining text—are linked to improvements in critical thinking, motivation, and writing quality (Malik et al., 2023; Pratama and Sulistiyo, 2024; Jin et al, 2025). However, there are concerns that overreliance on AI could impact creativity and critical thinking, highlighting the need for balanced integration and reflective task design. The quality and depth of student engagement with AI depend heavily on how instructors design writing tasks (Tu et al., 2024). Tasks that require students to analyze, critique, and refine AI-generated content foster deeper engagement and higher-order thinking compared to tasks that simply use AI for technical support. Affordances and Limitations of AI in Writing Instruction Generative AI offers both promising affordances for writing instruction and raises notable limitations. AI can provide benefits of immediate, personalized feedback and scaffolding that might otherwise require significant instructional time and individual attention from instructors - resources that are often constrained in many academic settings. For instance, Solak (2024) suggests that ChatGPT can provide students with real-time, individualized scaffolding for both the generative and refinement phases of academic writing, addressing instructional gaps that often exist due to classroom time constraints. Furthermore, Marzuki et al. (2023) conducted interviews with EFL instructors and found that a variety of AI writing tools (QuillBot, Wordtune, GPT, Paperpal, etc.) were perceived to improve students’ writing quality especially in the content and organization of essays. Other studies have found that integrating ChatGPT into writing instruction offers multiple benefits: it can support in-depth revision processes (Beck & Levine, 2023), enhance grammar and vocabulary development (Crompton et al., 2023), modify the complexity of challenging texts, encourage critical comparison between student-authored and AI-produced content (Tseng & Warschauer, 2023), and elevate the quality of EFL student writing in terms of structure and coherence (Marzuki et al., 2023). In a classroom study, EAP students found ChatGPT especially useful for brainstorming ideas and improving the structure and clarity of their essays (Glahn, 2023): by generating examples or alternative wordings, the AI helped students expand their ideas and vocabulary in ways a limited classroom timeframe might not always allow. GenAI’s affordances also include personalized and adaptive learning where tools like ChatGPT can tailor feedback and learning experiences based on students’ individual writing profiles as well as their language proficiency (Wu, 2024). By leveraging deep learning and natural language processing, these tools can identify students’ specific areas for improvement and provide targeted, individualized guidance to address them. Wu (2024) also notes that ChatGPT supports learners by generating relevant examples and explanations that align with their input, which helps clarify complex ideas and build writing fluency. Furthermore, the research highlights that AI tools can adjust the difficulty level of tasks and recommend targeted resources, making the learning experience more relevant and engaging. This adaptive feedback loop creates a more student-centered environment compared to traditional instruction models (Wu, 2024). Additionally, Kohnke, et al. (2023) highlight that ChatGPT can be used to identify a meaning of a word in context, develop quizzes, annotate texts and provide translations. More specifically, Kohnke et al. (2023) discuss the use of ChatGPT as an aid to support students and instructors in language learning, where teachers provide guided tasks or prompts, and students interact with ChatGPT to clarify meaning, ask follow-up questions, and sustain the conversation to deepen their understanding. Overall, the literature underscores the growing role of generative AI as a valuable support tool in writing instruction. However, alongside these benefits lie important limitations and risks. Accuracy and bias are two major pitfalls in the use of GenAI. Sun (2023) notes that ChatGPT-generated content can omit critical information or even fabricate details including producing false citations and missing key data points. Users must therefore scrutinize AI output carefully. Additionally, Creely et al. (2025) highlight that AI models often reflect the biases of their training data, and for the foreseeable future, learners who are navigating new contexts and new languages or who have limited AI literacy are at risk with regards to their agency as a result of these embedded biases. Furthermore, the integration of generative AI in education presents substantial challenges across three critical domains. Academic integrity concerns have emerged as educators grapple with increased opportunities for misuse, technological plagiarism, and the inconsistency in institutional policies governing appropriate use (Preiksaitis & Rose, 2023) . According to Preiksaitis and Rose’s (2023) scoping review, equally concerning are accuracy and dependability issues, as these systems often operate with outdated information, produce convincing hallucinations, and perpetuate biases from their training data (Lim et al., 2023). Perhaps most worrying are the potential detriments to learning, where overdependence on AI threatens to undermine critical thinking skills, and compromise assessment validity (Ogunleye et al., 2024). This sentiment resonates broadly across higher education, where research indicates that uncritical use of generative AI may undermine students' critical thinking abilities and memory retention (Abbas et al., 2024; Bai et al., 2023; Gerlich, 2025). Navigating this complex landscape requires a two-pronged approach: first, developing essential AI literacies among both educators and students to foster critical engagement with these tools; and second, conducting rigorous empirical research to evaluate AI's actual impacts on learning outcomes and pedagogical effectiveness. Only through this combined focus on literacy development and evidence-based assessment can we ensure that AI tools enhance rather than diminish meaningful learning experiences in higher education. Factors Facilitating or Hindering AI Integration in EAP The degree to which EAP instructors adopt generative AI in their teaching largely depends on a combination of personal, institutional, and contextual factors. Recent studies have begun identifying key drivers and barriers that influence instructors’ acceptance of AI in the classroom. In higher education, instructors are more inclined to integrate GenAI tools when they perceive them as useful and easy to use. According to Al-Abdullatif (2024), perceived ease of use was the strongest predictor of GenAI acceptance, while AI literacy and intelligent Technological Pedagogical and Content Knowledge (TPACK) played important indirect roles. Instructors who were confident using GenAI tools and found them accessible were more likely to integrate them into their teaching. However, having experience with educational technology in general did not automatically translate into readiness to use GenAI. Many instructors lacked direct, hands-on experience with these tools, which limited their ability to recognize their usefulness in enhancing learning outcomes. This suggests that familiarity with GenAI itself is essential for adoption. These findings point to the need for targeted, context-specific training that supports both technical skills and pedagogical application. The supportiveness of the institutional environment is also a significant external factor in an instructor’s willingness to adopt GenAI tools. Research has consistently shown that the presence of clear policies, access to structured training, and sustained professional development opportunities are critical in fostering instructor confidence and readiness (Saihi et al., 2025). A recent study by Ayyoub et al., 2025 found that educators build GenAI literacy through two main pathways: formal professional development and active community support. Drawing on the TPACK framework, the study shows that while professional development shapes perceptions of usefulness, it is community support that directly builds AI competence. The findings also point to the importance of teacher efficacy, strengthened through peer support and mentoring, and call for policy frameworks that can guide the thoughtful integration of AI in teaching. Additionally, Jian et al. (2024) highlight that many educators feel underprepared to integrate GenAI into assessment and instructional practices due to a lack of institutional guidance and discipline-specific frameworks. Their review found that most universities had yet to develop comprehensive policies, and even where such policies existed, they were often vague or perceived as insufficient. Moreover, facilitating conditions—such as infrastructure, training, and administrative support—consistently predict AI adoption (Saihi et al., 2025; Perez, 2024). Social influence and perceived usefulness also shape faculty members’ behavioral intentions, particularly when supported by strong institutional frameworks. Faculty training and peer encouragement further enhance acceptance. In short, the literature points in the direction of job security, strategic organizational support, and cultural alignment as essential criteria for successful AI integration in academic settings. The conditions under which instructors work can also affect AI adoption. In a study of adult English language teaching in Canada, Breshears (2019) identifies low wages, reliance on part-time employment, job insecurity, lack of benefits, limited administrative support, and extensive unpaid labor as persistent challenges in the EAP sector. These factors often combine to create unstable and demanding work environments, making it difficult for instructors to commit time to professional development and may impact the amount of energy they give to learning new technologies. As Breshears notes, these employment conditions are widespread and have long shaped the lived experiences of EAP instructors in Canada. Similar concerns are raised by Kouritzin et al. (2023), who describe how unstable contracts and unclear rehiring practices create uncertainty and limit EAP instructors’ ability to plan ahead. This ongoing insecurity may contribute to the hesitation or inability to engage with new tools and approaches, including the use of educational technology. Theoretical Framework This research investigates how English for Academic Purposes instructors perceive and implement Generative Artificial Intelligence in their teaching practices. The study employs three complementary theoretical frameworks to develop its interview protocol and analyze instructor experiences: the PICRAT technology integration model, the Unified Theory of Acceptance and Use of Technology (UTAUT), and critical pedagogy. While the PICRAT and UTAUT models provide valuable insights into technology integration patterns and adoption factors, they do not adequately address the power dynamics inherent in AI adoption within educational contexts. Therefore, this study incorporates critical pedagogy as an essential framework to examine how GenAI may reinforce or challenge existing hierarchical structures in language education. The PICRAT model, developed by Kimmons, Graham, and West (2020), provides a structured approach to evaluating technology integration through a matrix examining two dimensions. The first dimension considers students' relationship with technology along a continuum from passive reception to interactive engagement to creative production. The second dimension assesses how technology impacts teaching practice, whether merely replacing traditional tools, amplifying existing practices, or transformatively enabling previously impossible learning experiences. This framework helps determine whether instructors view GenAI as simple replacements for existing resources or as transformative elements that fundamentally alter teaching and learning. PICRAT is particularly valuable because it "emphasizes technology as a means to an end" and "focuses on students," ensuring technology serves language learning objectives rather than becoming an end in itself (Kimmons et al., 2020). However, the model's focus on functionality and student-centered outcomes overlooks critical questions about who controls AI technologies, whose voices and perspectives are embedded in AI systems, and how AI adoption might perpetuate or disrupt existing power imbalances between instructors and students, institutions and educators, or dominant and marginalized linguistic communities. The Unified Theory of Acceptance and Use of Technology (UTAUT), developed by Venkatesh and colleagues (2003), identifies four key determinants influencing technology adoption: performance expectancy (beliefs about effectiveness), effort expectancy (perceived ease of use), social influence (impact of colleagues and institutional culture), and facilitating conditions (technical and organizational infrastructure). In the context of GenAI adoption by EAP instructors, this framework helps explore barriers and enablers instructors experience, examining not only individual attitudes but also how institutional contexts and professional networks shape adoption decisions. While UTAUT effectively captures the mechanics of technology acceptance, it operates from a largely positivist perspective that treats technology adoption as a neutral process of optimization. This approach fails to interrogate the political dimensions of AI implementation, such as how institutional mandates for AI adoption may constrain instructor autonomy, or how AI tools developed by profit-driven corporations may not align with pedagogical values of social justice and student empowerment. Critical pedagogy, rooted in Paulo Freire's (1968) seminal work Pedagogy of the Oppressed, offers a theoretical lens for examining power dynamics in educational settings that the PICRAT and UTAUT models overlook. This approach rejects the "banking" concept of education where students passively receive knowledge, instead advocating for education as an emancipatory practice developing critical consciousness. Critical pedagogy views education as inherently political and aimed at fostering social justice, recognizing that GenAI tools are not neutral technologies but products embedded with specific values, biases, and power relations. As Freire argued, critical pedagogy "involves both the recognition that human life is conditioned, not determined, and the crucial necessity of not only reading the world critically but also intervening in the larger social order as part of the responsibility of an informed citizenry" (Freire, 1968). This perspective frames essential questions about how EAP instructors perceive GenAI's role in either reinforcing traditional power structures or enabling more democratic approaches to language learning. By incorporating critical pedagogy alongside PICRAT and UTAUT, this study can examine not only how instructors integrate and adopt AI technologies, but also how they navigate questions of agency, equity, and resistance in AI-mediated educational environments. Together, these three frameworks provide complementary perspectives for investigating GenAI integration in EAP contexts. While PICRAT offers a practical tool for categorizing technology use, UTAUT explains the adoption process itself, and critical pedagogy prompts deeper questions about power, equity, and the purpose of technology in education. The interview protocol addresses both practical aspects of GenAI integration and deeper pedagogical, ethical, and institutional considerations. Questions explore how instructors perceive GenAI's potential to transform existing practices, navigate power dynamics and equity concerns, and what factors influence their adoption decisions. In the data analysis phase, these frameworks serve as interpretive lenses: PICRAT helps categorize and evaluate reported GenAI uses; UTAUT frames understanding of factors influencing varying adoption levels; and critical pedagogy guides analysis of how instructors negotiate issues of power, access, and equity. This multi-theoretical approach acknowledges that educational technology decisions are never merely technical but fundamentally connected to questions of pedagogy, power, equity, and institutional context (Weisberg and Dawson, 2023). Methods This qualitative case study investigated academic writing instructors' perceptions of generative AI integration in teaching and learning contexts, examining both implementation practices and factors influencing adoption or non-adoption decisions. In-depth, semi-structured interviews were conducted with nine experienced academic writing instructors employed at post-secondary institutions in Ontario, Canada. Semi-structured interviews were chosen as the primary data collection method because they allow for systematic exploration of predetermined themes while maintaining the flexibility to pursue unexpected insights that emerge during conversations (Bryman, 2016). The following is a sample of the questions on the interview protocol in this study: “Can you tell me about your experience teaching EAP? How would you describe your familiarity with AI technologies in education? Are you currently using any AI tools in your teaching of academic writing? Why or why not?” This approach was particularly well-suited to the study's objectives as it enabled the researchers to probe deeply into instructors' complex experiences with AI integration while allowing participants to share perspectives that might not have been anticipated in the research design. The semi-structured format also facilitated comparison across interviews while preserving the nuanced, contextual details essential for understanding individual instructor experiences with emerging technologies. Participants were recruited through purposive snowball sampling, a method particularly effective for accessing specialized populations with specific expertise (Creswell and Poth, 2018). Given the relatively recent emergence of generative AI in educational contexts, instructors with substantial experience using these tools in academic writing instruction represent a specialized and somewhat difficult-to-identify population, making snowball sampling an appropriate recruitment strategy. However, this sampling approach carries inherent limitations that must be acknowledged; 1) potentially creating homogeneous samples as participants tend to refer others within their professional networks who likely share similar characteristics, experiences, or perspectives (Noy, 2008). This may limit the diversity of viewpoints captured in the study, potentially excluding instructors with different institutional contexts, pedagogical approaches, or levels of AI engagement and. 2) introducing selection bias, as participants willing to discuss AI integration may represent those who are more comfortable or positive about these technologies, potentially underrepresenting instructors who are skeptical, resistant, or have had negative experiences with AI tools. Despite these limitations, snowball sampling remained the most feasible approach for identifying and accessing this specialized population of early AI adopters in academic writing instruction. Interviews were conducted remotely via video conferencing, recorded with participant consent, and professionally transcribed for analysis. The interview protocol, informed by the theoretical frameworks of PICRAT, critical pedagogy, and UTAUT, explored participants' understanding of generative AI capabilities, their pedagogical approaches to integration, perceived benefits and challenges, concerns about equity and power dynamics, and institutional factors influencing adoption. Thematic analysis was employed to identify patterns across the data corpus, with initial coding conducted independently by two researchers using NVivo to enhance reliability. The coding scheme was iteratively refined through researcher consensus meetings until thematic saturation was achieved. Member checking was conducted by sharing preliminary findings with participants to verify interpretations and enhance validity. All research procedures received approval from the institutional research ethics board prior to participant recruitment. Results and Discussion Participant Technology Engagement Spectrum The participants showed significantly diverse levels of engagement with AI technologies: While some participants have extensively explored advanced tools like ChatGPT and various image generators, integrating them thoughtfully into their teaching practices, others candidly identify themselves as beginners with only cursory or superficial experience with these technologies. This diversity reflects a natural progression of technology adoption in education, with participants generally acknowledging the substantial learning curve associated with AI implementation and viewing it as an ongoing process of discovery. On one side of the continuum, we have P4 who has been exploring GenAI application in teaching and learning since the inception of ChatGPT, and on the other side is P9 who has not yet explored the uses of LLMs in teaching writing because she hasn’t felt the need to call on AI for instructional support. The remaining seven participants fall between those two points in terms of the breadth and depth of exploration and use of AI for teaching. Pedagogical Integration and Evolving Teaching Practices Student-Centered AI Literacy Development When discussing AI use in teaching, all participants primarily reflected on guiding students in utilizing AI tools rather than their own instructional practices. Participants expressed particular concern with developing students' critical AI literacy—their ability to thoughtfully evaluate and critique AI outputs rather than accepting them uncritically. They emphasized teaching students to use AI as supplementary resources enhancing learning rather than replacing knowledge and skill development. This pedagogical priority reflects educators' commitment to ensuring students become discerning AI users maintaining intellectual agency while leveraging technological advantages, aligning with AI literacy best practices (Wolters, 2024; Stolpe & Hallström, 2024 ). P8 offers nuanced responses to student AI questions, foregrounding context: "[I]t depends on the purpose. Is this for an assignment? Are you submitting this where you're saying that this is your language work? Is this a task where the focus is on language? Or is it a task where the focus is on expressing your idea? For me. It really is about context." Participants focus on teaching effective chatbot use to mitigate potential harms (Shanmugasundaram and Tamilarasu, 2023). From a PICRAT perspective, participants' emphasis on student training reflects deliberate GenAI positioning in the Interactive-Amplify quadrant, where technology enhances critical thinking rather than replacing student cognitive processes. This pedagogical choice maintains technology as means to educational ends rather than ends themselves, aligning with PICRAT's student-centered evaluation framework. Several participants have been fielding student questions about ethical and effective GenAI use, making their classroom approach necessarily reactive. P1 suggests clear usage guidelines encourage positive student engagement with these tools. Similarly, P6 guides students on leveraging GenAI to support rather than replace their thinking and learning processes. Acknowledging students already use AI without guidance on effective and ethical use, she teaches critical thinking by emphasizing starting the writing process with their own ideas, using AI for idea organization and brainstorming. P9 argues academic writing courses help graduate students develop disciplinary writing expertise and emphasizes gaining foundational skills before using GenAI tools—otherwise, students struggle to develop field expertise and lack the ability to critique AI outputs. This position resonates with Zhang et al.'s ( 2024 ) recent study postulating that AI users can only verify information when they already know the answer, and as topics become unfamiliar, AI reliance increases even when AI provides explanations. Lindbaum and Fleming (2023) compare academics outsourcing writing to AI to "turkeys voting for Christmas," cheering their own doom. Moreover, they note ChatGPT produces outputs based on derivative, high-probability choices, "has no stake in the knowledge it produces and is thus likely prone to offering irresponsible outputs" (p. 566, emphasis in original). They contrast this with authentic human thought shaped by socio-cultural context and employing unconventional choices within and beyond existing data boundaries, rendering it creative, situated, and contextually relevant. Practical Applications of AI in Writing Instruction Participants offered detailed accounts of their experiences and perspectives about AI in writing instruction. The majority of participants in this study conceptualized chatbots as complementary tools within their broader educational toolkit, where, as P6 observed, the technology currently "fits in kind of interstitially" within established curricula and institutional practices. Some of the participants expressed positive perceptions towards AI as a potential thinking partner: P1 highlighted GenAI's potential as an interactive partner for pedagogical reflection, particularly valuable in the often isolating academic environment. She appreciated how language models can provide feedback on assignment design, helping to assess critical thinking skills. Innovatively, P1 uses ChatGPT as a litmus test for assignment quality - if the AI can generate a passing paper, she views this as a sign that her rubric is too superficial and fails to capture substantive human thought. This approach demonstrates a nuanced method of using AI to enhance educational assignment design by ensuring deeper, more meaningful student engagement. This method underscores the participant's commitment to developing assignments that necessitate human cognition and avoid surface-level responses, while also demonstrating an innovative use of GenAI in pedagogical practice - an emerging practice in the literature on AI&Ed (see for e.g. Bushell, 2024 ; Zepeda et al., 2023 ). Similarly, P2 who works part-time as an educational consultant said that she uses GenAI as a sounding board and a professional development resource given her recent engagement in the field. P7 noted that she uses GenAI to help her brainstorm ideas for classroom activities, using it like she would a search engine to see what has been done on a certain topic. She noted that she never takes the output at face value, rather she assesses the fit of the output to her needs and modifies it accordingly, a sentiment which P3 shares as well. Several participants shared that they use GenAI to create exemplars of different genres of academic writing like abstracts or introductions, and P7 shared that she sometimes uses GenAI to generate paraphrases or to show students the different ways a sentence can be structured, which saves her a lot of time to spend on more sophisticated tasks in the classroom. Several participants shared that GenAI tools are useful for creating exemplars to teach genre awareness or to generate “dummy text” to teach citation and referencing skills. These tasks were described as “rote” or “supplementary” versus more high-stakes tasks like creating assignment instructions, a task that all participants agreed cannot and should not be relegated to AI. The majority of applications fall into the Interactive-Amplify category, where GenAI enhances existing pedagogical practices while maintaining instructor agency and student-centered learning objectives. P1's innovative "litmus test" approach for assignment design exemplifies this integration pattern, using ChatGPT to evaluate whether assignments require genuine human cognition rather than surface-level responses. Similarly, P7's use of GenAI for sentence restructuring demonstrations and P6's application for literature review organization represent amplification of established teaching methods rather than wholesale replacement of instructor expertise. Participants used GenAI-generated content as a teaching tool for critical analysis. P4 noted that students enjoy critiquing GenAI-generated content because “they're not worried about hurting anybody's feelings. They can just trash it, basically.” By examining ChatGPT's outputs, which often contain pragmatic language errors like overly formal or too chummy language, students can learn to identify and correct inappropriate communication styles. This approach allows for teaching language pragmatics through both high-quality and flawed AI-generated examples, encouraging critical thinking about effective communication. Lee and Cook’s ( 2024 ) study on ChatGPT’s pragmatic competence confirms that this is indeed an expedient use of chatbots. The researchers' qualitative analysis revealed ChatGPT's inconsistent performance in language tasks: while the AI struggles with crafting appropriately nuanced apologies, lacking politeness and contextual language sensitivity, it demonstrates more competence in generating requests and refusals. P6 sees similar potential in GenAI that she says can be useful to generate texts with a variety of registers, and P4 shared that ChatGPT might be useful as a teaching tool for generating academic texts that students can critically analyze. She notes that the outputs of ChatGPT often lack specificity, originality, and coherence, making them ideal examples for students to evaluate against academic rubrics. In her teaching approach, P8 uses AI-generated content to help students develop critical writing skills, showing how chatbots "don't add information" but "can make the language pretty." Her goal is to demystify AI use, encouraging open classroom discussion about its capabilities and limitations. P4 reveals that her "ulterior motive" in engaging students with this process is to demonstrate the limitations of ChatGPT's outputs. The goal is to foster students' critical evaluation skills and discourage over-reliance on AI while maintaining their autonomy in learning. The literature indeed warns about the perils of excessive “cognitive offloading” to AI (Gerlich, 2025 ). Cognitive offloading refers to using external tools and agents like AI to reduce cognitive load, which in theory can enhance efficiency without necessarily diminishing cognitive engagement. However, excessive reliance on these tools, especially AI, may potentially compromise deep thinking and critical analysis. These concerns are substantiated by Gerlich's (2025) study, which demonstrated a significant negative correlation between frequent AI tool usage and critical thinking abilities, with cognitive offloading serving as a mediating factor. P6 said GenAI can be useful for teachers and graduate writers to help them sort through massive amounts of literature by creating literature review matrices, the caveat being that the machine would only be responsible for performing “the manual labor” of sorting through ideas, but the ideas themselves would be the purview of the user only, which falls within the Passive-Replace quadrant, where technology substitutes for traditionally manual processes. Image generation seems to be a popular feature among participants who use multimodal texts in their teaching material and find non-copyrighted GenAI generated images convenient and modifiable. P5 shared that one way to use image generators is to create an activity that aims to cultivate critical thinking, language precision, and evaluative skills by having students assess image descriptors generated by AI, some created imperfect by design. Students would be asked to critically evaluate whether the descriptors accurately represent the images; by requiring students to analyze, interpret, and evaluate AI-generated content, the teacher encourages higher-order thinking, active engagement with technology, and practical application of knowledge, “aligning with the university's focus on independent thought and nuanced understanding”. Few truly transformative applications emerged that would qualify for the Creative-Transform quadrant, though P4's approach of having students critique AI outputs to develop metacognitive awareness and P5's critical analysis activities examining AI-generated image descriptors suggest potential for more radical pedagogical innovation. This PICRAT analysis reveals that participants primarily view GenAI as amplifying their existing pedagogical toolkit rather than fundamentally transforming their teaching practices, suggesting a cautious, enhancement-focused approach to AI integration that preserves human agency in educational relationships. GenAI and Language Learners All of the participants in this study teach student populations who can be described as English language learners. When asked if they thought GenAI offered any affordances to multilingual learners, several participants observed that GenAI can be useful for multilingual graduate students, as it might help reduce learning anxiety by improving their diction, syntax, and overall writing accuracy. For instance, P8 maintains a practical perspective, acknowledging that "if it were me in a non-language program, I'd use [AI] to check my vocabulary”. This stance resonates with much of the literature on AI tools and language learning anxiety that posits AI tools can enhance learners' autonomy and self-efficacy, which reflects in lower learning anxiety levels (Yuan, 2024 ; Song, 2023 ; Jubier, 2024). On the other hand, P1, P8 and P9 expressed overlapping concerns regarding multilingual learners use of AI tools as they suggested there might be a risk that the authenticity of multilingual writers' work might be unfairly questioned because the formal style often adopted by many multilingual writers can sometimes resemble AI-generated text in terms of style, potentially leading to unwarranted suspicions of AI use even when none has occurred Increasing bias against multilingual writers. Several of the participants also expressed concern over learning loss; P8 cautioned that if "the purpose is to show your linguistic ability, then that might not be the best idea." Notably, P4’s remarks about the attitudes of the international students in her writing program which caters to engineering students reveal a welcome resistance among high-achieving international engineering students toward adopting AI tools for writing assistance, challenging prevailing assumptions in the literature. P4 explained that the high achieving students in this program “would not deign to use AI in their learning”. However, when these students inquire about AI, they demonstrate sincere interest in understanding ethical and effective implementation. This authentic curiosity motivates P4 and colleagues to explore pedagogical approaches to AI that best address their students' educational needs. In yet another astute observation, P7 observed a nuanced tension in current discussions around AI use in education. On one hand, much of the existing literature emphasizes concerns about academic integrity and the potential erosion or underdevelopment of students’ writing skills due to reliance on AI. On the other hand, P7 highlighted a more subtle but growing issue among multilingual students: a new form of anxiety stemming from the perception that AI might eventually "write better than they can." This perception exacerbates existing insecurities many students already feel when writing in English-medium classrooms. P7 warned that this could lead educators to adopt a "deficit approach"—essentially abandoning efforts to develop students' writing abilities and instead allowing AI to do the work for them. The participants' concerns about multilingual learners reveal what critical pedagogy would recognize as intersectional oppression, where AI adoption may compound existing linguistic marginalization rather than alleviating it. P1, P8, and P9's worry that multilingual writers' work might be "unfairly questioned" because their formal style resembles AI text highlights how technological bias can reinforce linguistic prejudice, creating additional barriers for students already navigating language-based discrimination. This reflects Freire's understanding that educational tools are never neutral but always carry the potential to either humanize or dehumanize learners, with AI technologies risking the further marginalization of students whose linguistic identities are already undervalued in academic contexts. Balancing Innovation and Integrity: Educators’ Perspectives on AI’s Role in Writing, Learning, and Critical Engagement Cognitive Concerns and Learning Loss The participants in this study unanimously emphasized, albeit in varying ways, that the complex cognitive processes involved in producing written work could be undermined if students were to consume too much AI-generated content. Many participants expressed concerns about potential learning loss that could be triggered by over-reliance on AI technologies in academic writing. The literature shows that over-reliance on AI chatbots could increase the risk of critical cognitive skill atrophy and the decline of problem-solving abilities (Dergaa et al., 2024) and might lead to a decline in autonomy and decision-making skills (Klingbeil et al., 2024 )In other words, over-reliance on AI in education could be a recipe for entrenching the banking model of education (Freire, 1970) where students enter a prompt into the machine and accept the output without critical engagement, questioning its assumptions, or understanding the complex processes behind the generated content. This passive consumption of AI-produced content mirrors the very educational dynamics that Freire critiqued, where learners become receptacles of information rather than co-creators of knowledge through active engagement, dialogue and critical reflection. Quality and Authenticity of AI-Generated Writing The participants in this study expressed fundamental skepticism about the rhetorical and intellectual value of chatbot-generated writing, identifying significant limitations in its capacity to produce nuanced, authentic, or contextually appropriate text that would meet standards of academic discourse. Their critiques revealed not merely disappointment with current output quality, but deeper concerns about the algorithmic reproduction of language that lacks genuine rhetorical awareness and critical thought. P1 described the language produced by chatbots as flowery and vapid, lacking in depth and critical thinking; P3 described the style of AI outputs as generic, while P4 shared that the ChatGPT is good at saying things that “sound intelligent” but lack real substance. Diverse Pedagogical Philosophies Toward AI Integration The educators displayed varying attitudes toward AI in education. P3 observed a leveling of student fascination with AI as they recognize its limitations, while P1 adopted a pragmatic stance, advocating for teaching effective and ethical AI use since students were already utilizing it. P2, who teaches adult language learners, expressed skepticism about teaching ethical AI use, stating that "you can't teach trust." P4 demonstrated the most positive outlook toward AI, as her students used the technology responsibly while maintaining personal effort in their work. P5 approached AI with critical concern, focusing on how language models reproduce biases and arguing that critical language analysis should examine who is included, excluded, and what information is privileged - though she noted these considerations often became secondary to practical implementation in English for Academic Purposes contexts. P6 took a utilitarian perspective on AI implementation. she advocated for maintaining AI as a supplementary resource that enhances thinking processes and facilitates brainstorming, rather than allowing it to replace student thinking altogether. Her position emphasized keeping AI "in its place" - establishing clear boundaries around its use to ensure it remains a supportive tool that augments human cognition rather than substituting for it. This approach suggests P6 sees value in AI when properly constrained within a supporting role in the educational process. P7 takes a literacy-focused approach to AI in education, particularly within her writing center context where it is not the writing consultants’ purview to set or enforce AI policies. She emphasizes the importance of student awareness, noting "we cannot really dictate what the students should be doing" but instead guides them to "read the syllabus" and understand course-specific policies. P7’s perspective extends beyond basic skills development to a critical literacy framework where students learn to "evaluate generated text that's produced by AI in terms of biases, in terms of power dynamics, in terms of whose voices are presented." When working with students permitted to use AI, she proposes comparative analysis, having students "compare their own writing with the writing of AI" to develop critical awareness. This approach cultivates "genre awareness" by examining whether AI-generated text "follows genre requirements" and evaluating language choices, particularly for multilingual students. Despite limited personal experience with AI implementation, P7 envisions it serving as "a study buddy" - a supportive tool to be deployed when needed. P8 demonstrates a practical, context-driven approach to AI integration in education. Despite working with a set curriculum that lacks formal AI components, she incorporates AI discussions "interstitially," particularly when teaching academic integrity. P9, who has perhaps engaged the least with AI, presents a balanced view of AI in education, emphasizing the classroom's role as a technology-independent learning space. She positions the classroom as "a place to give students tools so that they can do it without it," recognizing that while AI might offer benefits, students "could surely experiment with it themselves individually." P9 values preserving traditional learning approaches, suggesting instructors should ensure students know "how people have done this without this tool, just in case." When considering potential AI applications, P9 focuses on analytical capabilities rather than content generation. She expresses interest in AI that could "diagnose things for students," such as identifying when "this paragraph lacks coherence" or highlighting structural weaknesses. Her perspective is shaped by her instructional priorities - "argumentation, paragraph structure, coherence" and teaching students to "emphasize what needs to be emphasized and be loud and overt about what it is that they're trying to say." However, P9 remains skeptical about current AI capabilities in this domain, noting "I don't think it's there yet" and characterizing her desired analytical AI tools as "a bit far off." She wants AI that could assist with critical thinking by "pointing out logical flaws" or identifying missing structural elements like topic sentences, suggesting she values AI as a potential diagnostic tool rather than a replacement for student writing. Cognitive Concerns and Learning Loss Participants unanimously emphasized that complex cognitive processes in written work could be undermined by excessive AI-generated content consumption. Many expressed concerns about potential learning loss from over-reliance on AI technologies in academic writing. Literature shows over-reliance on AI chatbots increases risk of critical cognitive skill atrophy and declining problem-solving abilities (Dergaa et al., 2024) and may reduce autonomy and decision-making skills (Klingbeil et al., 2024 ). Over-reliance on AI could entrench Freire's banking model of education (1970) where students enter prompts and accept outputs without critical engagement, questioning assumptions, or understanding complex processes behind generated content. This passive AI consumption mirrors educational dynamics Freire critiqued, where learners become information receptacles rather than knowledge co-creators through active engagement, dialogue, and critical reflection. Quality and Authenticity of AI-Generated Writing Participants expressed fundamental skepticism about chatbot-generated writing's rhetorical and intellectual value, identifying significant limitations in producing nuanced, authentic, or contextually appropriate text meeting academic discourse standards. Their critiques revealed not merely disappointment with output quality, but deeper concerns about algorithmic language reproduction lacking genuine rhetorical awareness and critical thought. P1 described chatbot language as flowery and vapid, lacking depth and critical thinking; P3 called AI outputs generic, while P4 noted ChatGPT excels at saying things that "sound intelligent" but lack real substance. Participants AI Attitudes Educators displayed varying AI attitudes. P3 observed leveling student fascination as they recognize AI limitations, while P1 adopted pragmatic stances, advocating teaching effective and ethical AI use since students already utilize it. P2, teaching adult language learners, expressed skepticism about teaching ethical AI use, stating "you can't teach trust." P4 demonstrated the most positive AI outlook, as her students used technology responsibly while maintaining personal effort. P5 approached AI with critical concern, focusing on how language models reproduce biases and arguing critical language analysis should examine inclusion, exclusion, and privileged information—though noting these considerations often became secondary to practical English for Academic Purposes implementation. P6 took utilitarian perspectives on AI implementation, advocating maintaining AI as supplementary resources enhancing thinking processes and facilitating brainstorming rather than replacing student thinking. Her position emphasized keeping AI "in its place"—establishing clear boundaries ensuring it remains supportive tools augmenting human cognition rather than substituting for it. This approach suggests P6 sees AI value when properly constrained within supporting educational roles. P7 takes literacy-focused AI approaches, particularly within writing center contexts where consultants don't set or enforce AI policies. She emphasizes student awareness importance, noting "we cannot really dictate what the students should be doing" but guides them to "read the syllabus" and understand course-specific policies. P7's perspective extends beyond basic skills to critical literacy frameworks where students learn to "evaluate generated text that's produced by AI in terms of biases, in terms of power dynamics, in terms of whose voices are presented." When working with students permitted to use AI, she proposes comparative analysis, having students "compare their own writing with the writing of AI" to develop critical awareness. This approach cultivates "genre awareness" by examining whether AI-generated text "follows genre requirements" and evaluating language choices, particularly for multilingual students. Despite limited personal AI experience, P7 envisions it serving as "a study buddy"—supportive tools deployed when needed. P8 demonstrates practical, context-driven AI integration approaches. Despite working with set curricula lacking formal AI components, she incorporates AI discussions "interstitially," particularly when teaching academic integrity. P9, who has engaged least with AI, presents balanced AI education views, emphasizing classrooms' roles as technology-independent learning spaces. She positions classrooms as "a place to give students tools so that they can do it without it," recognizing that while AI might offer benefits, students "could surely experiment with it themselves individually." P9 values preserving traditional learning approaches, suggesting instructors ensure students know "how people have done this without this tool, just in case." When considering potential AI applications, P9 focuses on analytical capabilities rather than content generation. She expresses interest in AI that could "diagnose things for students," such as identifying when "this paragraph lacks coherence" or highlighting structural weaknesses. Her perspective is shaped by instructional priorities—"argumentation, paragraph structure, coherence" and teaching students to "emphasize what needs to be emphasized and be loud and overt about what it is that they're trying to say." However, P9 remains skeptical about current AI capabilities, noting "I don't think it's there yet" and characterizing desired analytical AI tools as "a bit far off." She wants AI assisting critical thinking by "pointing out logical flaws" or identifying missing structural elements like topic sentences, suggesting she values AI as potential diagnostic tools rather than student writing replacements. Ethical Considerations and Ethical “Discomfort” Multifaceted Ethics Beyond Academic Integrity The participants in this study displayed sophisticated levels of awareness of the ethical dilemmas associated with the use of GenAI in education. For example, when P5 was asked if she had any ethical concerns about AI, she countered by asking "but which ethics? Is it environmental ethics, or the ethics of labor associated with the training of the models?" This response reveals P5's nuanced understanding of AI ethics as a multifaceted issue rather than a singular concern, demonstrating her awareness that ethical considerations span diverse dimensions from environmental impact to labor practices in AI development. By reframing the question, P5 highlights how discussions of AI ethics often lack specificity and fail to address the complex, interconnected ethical frameworks necessary for meaningful evaluation. In line with most of the literature about the concerns around the use of GenAI in education, the participants in the study are well aware of the likelihood of GenAI being used by students, with a few stating that it is not always possible to tell whether a student has used AI to write a part or all of their paper or if they’ve used it as a thought partner. Notably, many of the participants consistently demonstrated a nuanced understanding that students' engagement with GenAI is inevitable in contemporary educational settings, or as P4 put it, “if you can’t beat them, join them”, in reference to her awareness of her students use of AI in their learning and her attempts at meeting them where they are. Data Privacy and Sovereignty Concerns Many of the participants expressed ethical concerns over losing data privacy and data sovereignty were the practice of using AI to give feedback to students to become widespread. To illustrate, P8 expressed concerns regarding inputting student work in the bot without the students’ consent and showed equal concern about the sources of information the AI provides in its outputs which are scraped from thousands of uncited internet sources. P5 shed light on a profound and often overlooked issue in the discussion of AI ethics in education which is that of indigenous data sovereignty and the double standards that exist when it comes to protecting them: “Indigenous communities have been working for years to protect their knowledges and data sovereignty. Those same people who are up in arms about ethics and cheating - are they defending the data sovereignty of indigenous and other marginalized communities? Are they looking at the built-in data bias of scraping of different communities? Of the reification of misogyny and racism of these large sexist, racist language banks that they're drawing from? If we want to get into ethics, yes, bring it on it's overdue, but make it broad based.” This framing challenges us to broaden our ethical considerations around AI beyond narrow institutional concerns to include questions of power, representation, and justice. It suggests that truly ethical AI engagement requires a more comprehensive approach that centers traditionally marginalized perspectives rather than just protecting established academic norms (cite feminist critique). The call to make ethics discussions "broad based" is essentially asking for consistency in our ethical standards - if we care about proper use of knowledge and information in one context, we should apply similar principles of respect and consent across all contexts, especially regarding communities whose knowledge has historically been appropriated without acknowledgment or compensation. The participants' understanding of the ethical issues surrounding the use of AI tools is very nuanced as is befitting of professional teachers of writing. All of the participants agreed that teachers also have a responsibility to acknowledge their own use of AI in the preparation of any pedagogical material. Power Dynamics and Resistance in AI Adoption Critical pedagogy's focus on power relations illuminates several tensions in participants' AI experiences that functional frameworks like PICRAT and UTAUT cannot adequately address. The dynamics of institutional power versus instructor autonomy emerged clearly as participants navigate between institutional expectations for AI literacy and their pedagogical values. P9's emphasis on preserving "technology-independent learning spaces" represents resistance to technologically deterministic approaches to education, asserting the classroom as a site where human agency and critical thinking can develop without technological mediation. Questions of knowledge authority and student agency reveal complex power negotiations around AI adoption. P4's observation that high-achieving international students "would not deign to use AI" reveals student resistance to technological mediation of their learning, suggesting agency in rejecting tools perceived as undermining their intellectual development. This resistance challenges assumptions about technology adoption as inevitably beneficial, instead revealing how students themselves may critique AI integration as inconsistent with their educational values and goals. Perhaps most significantly, P5's powerful critique of AI ethics highlights how discussions of academic integrity often ignore "the data sovereignty of indigenous and other marginalized communities" whose knowledge systems have been appropriated to train AI models without consent or compensation. This analysis reveals how AI adoption can perpetuate colonial knowledge practices while appearing ethically neutral, embodying what critical pedagogy recognizes as the hidden curriculum of technological implementation. From a critical pedagogy perspective, these dynamics reveal that AI adoption is never merely a technical decision but always involves questions of power, agency, and whose interests are served by technological integration. Barriers to GenAI Integration in Academic Writing P5 offers a critical perspective on how generative AI fits into existing problematic attitudes toward language education where she identifies AI's appeal as rooted in a utilitarian view that sees "language as a means to an end and not its own necessarily knowledge base." P5 connects current AI enthusiasm to historical perspectives of English for Academic Purposes that encourages a service model or what Raimes ( 1991 ) calls “the butler’s stance” (p. 243), where language instructors “get [students’] English ready and then send them to the University where [subject-matter instructors] will do the real work" (p.420). P5 sees generative AI as perpetuating this problematic division, with many hoping AI will "just remove that barrier" to language, feeding into a fantasy that "no one's going to have to study or really engage critically with language anymore" because "the generative AI will produce what's needed", effectively identifying a technosolutionist fantasy that bypasses the critical consciousness-raising that Freire argued was essential to authentic education. P5's critique of GenAI's appeal as rooted in utilitarian views of language directly challenges what Freire (1968) identified as the banking concept of education, where knowledge becomes a commodity to be deposited rather than co-created through critical dialogue. Her analysis suggests AI is being embraced not as an educational tool but as a technosolutionist bypass for meaningful language engagement (Morozov, 2013 ). The appeal of AI as a technological fix for language learning reflects what Freire would recognize as a false solution that addresses symptoms rather than root causes of educational inequality, potentially reinforcing the very power imbalances that critical pedagogy aims to disrupt. Overall, the participants expressed concern that the un-critical adoption of GenAI risks entrenching the very educational dynamics that critical pedagogy seeks to transform, where learners become passive recipients of AI-generated content rather than active co-creators of knowledge through critical engagement and dialogue. As P9 pointed out, graduate academic writing courses facilitate the acquisition of foundational academic writing skills like “how to make an argument, how to synthesize other people's arguments, how to insert your own stance, how to signal your stance”, and these skills should not be bypassed by asking the machine to do the writing, even if the machine is capable of performing those tasks. Questions abound in the literature about the skills we lose when we offload our thinking to the bots (Creely et. al., 2025), and teachers everywhere are expressing similar hesitation and skepticism regarding the impact of chatbots on skill development (Titko et al., 2023; Aljunaid, 2024; Mohammadkarimi, 2023). P9 views the writing classroom as a critical space where students develop expertise in genre conventions and scholarly communication, and subverting this foundational cognitive exercise would deprive students of foundational communication skills. Her skepticism towards AI's current capabilities in academic writing is multifaceted, critically examining the tool's potential through a pedagogical lens. When she remarks, "if it could diagnose things for students... that would be really cool. I don't think it's there yet," she directly challenges AI's diagnostic abilities. Her critique centers on the nuanced aspects of academic writing that require sophisticated understanding: "argumentation, paragraph structure, coherence... How do you integrate a quotation into something that actually has impact?". P9’s stance complicates UTAUT’s notion of performance expectancy by questioning the depth of GenAI’s usefulness in academic writing. While she acknowledges its value for surface-level feedback, P9 critiques its inability to support deeper rhetorical and argumentative skills, which she sees as central to writing pedagogy. This skepticism highlights a limitation in UTAUT: its focus on perceived utility overlooks disciplinary values and pedagogical goals. By emphasizing the cognitive and epistemic dimensions of writing, P9 reframes “usefulness” not as efficiency but as meaningful learning, challenging the assumption that technological adoption in education is inherently beneficial or pedagogically neutral. This perspective resonates with P3's assertion that "AI lacks pedagogical knowledge" and is substantiated by emerging literature that emphasizes GenAI's limitation to statistical probabilities of text generation, rather than genuine reasoning or critical thinking (Yan et al., 2024 ; Daniel et al., 2025 ). Precarity versus Professional Development While the participants in this study displayed keen awareness of the need to stay updated on AI capabilities and applications, eight out of the nine participants in this study shared that they occupy precarious teaching positions in their institutions as PhD candidates with casual employment contracts. This element of job insecurity loomed large in the background of the participants’ responses as an important factor determining the depth and nature of instructor engagement with AI. For instance, when asked what factors might encourage her to adopt AI in her pedagogy, P9 readily said “paid training”. As a contract worker, her week is fully booked with contractual hours, and she doesn’t have the time or the financial resources to seek voluntary AI training. On the flip side, P1 shared that precarity might drive some instructors who are short of paid hours to outsource some of their work to AI, like using it to generate feedback for students or create lessons. Time constraints were also an important factor restricting the participants’ capacity to explore the pedagogical affordances of AI. In general, effort expectancy, a UTAUT construct, emerged as a significant barrier to adoption, with all participants acknowledging the substantial learning curve required for meaningful GenAI integration. Facilitating conditions, another UTAUT construct, presented the most significant constraints on adoption with resource limitations and institutional support gaps creating barriers that individual motivation cannot overcome. This is inline with a lot of literature on the importance of institutional proactivity in encouraging technological adoption (Walter, 2024 ; Keengwe et al., 2009; Porter and Graham, 2016). While all of the participants confirmed their awareness of the presence of professional development opportunities at their institutions, given the “chronic precarity of EAP employment” (Walsh Marr, 2021 , p.139), it is understandable that AI exploration is not top of mind for them. Additionally, the participants in this study who all have substantial teaching experience frequently asserted that their established pedagogical expertise, research background, and professional knowledge-base rendered GenAI superfluous for their practice. Conclusions/Future Directions This study sought to explore the perceptions of 9 EAP instructors about the integration of GenAI in the teaching of academic writing. Several themes have emerged from the data regarding the affordances and limitations of GenAI and the factors that might encourage or hinder the integration of GenAI in EAP pedagogy and instruction. The participants in this study demonstrated high levels of confidence in their pedagogical skills. Their nuanced and cautious attitudes towards the use of GenAI in teaching and learning do not arise from technological incompetence, limited digital literacy, or diminished self-efficacy—contrary to findings in some recent literature (Wang et al., 2024; Shahid et al., 2024 ; Hopcan et al., 2023). Instead, this measured hesitation stems from a genuine, profound concern over the future of human thinking, creativity, and intellectual autonomy as well as sporadic and unfocused institutional support for teaching with GenAI technology. Their reservations reflect a deep philosophical apprehension about the potential erosion of critical thinking skills, the authentic development of student creativity, and the fundamental nature of human cognitive processes when AI becomes increasingly integrated into educational environments. This concern is increasingly reflected in the literature, where scholars have warned that the expanding role of AI in education may compromise students’ critical thinking as well as their intellectual independence (Abbas et al., 2024 ; Bai et al., 2023 ; Gerlich, 2025 ). These educators are not resisting technological innovation, but are instead advocating for a thoughtful, intentional approach that preserves the irreplaceable human elements of teaching and learning—where technology serves as a supportive tool rather than a replacement for human intellectual engagement and pedagogical nuance. The participants’ responses reflect a strategic shift from prohibition toward thoughtful and cautious integration, with a deliberate focus on helping students understand the potential pitfalls of over-reliance on AI. Participants emphasized three key areas of concern: the risk of bypassing essential learning processes when AI is used as a shortcut, the fundamental need for students to develop authentic disciplinary knowledge that AI cannot replace, and the cultivation of sophisticated AI literacy that enables students to critically evaluate AI-generated content. Rather than adopting a defensive stance, these educators are proactively developing pedagogical approaches that acknowledge GenAI as part of the learning landscape while safeguarding the core educational mission of developing independent, critical thinkers capable of navigating an increasingly AI-mediated world. Inconsistent institutional frameworks and resource disparities impact instructor’s access to professional development and weakens their motivation to explore new educational technologies. Educational institutions lack uniform policies regarding AI use, creating a disjointed environment where instructors must navigate conflicting expectations on their own. While many institutions require AI statements in syllabi, they often fail to consider teachers' technological preferences, skill levels, and available resources. This departmental and institutional policy fragmentation creates an atmosphere of uncertainty, with most instructors expressing the need for clearer, more functional guidelines especially on the pedagogical applications of AI. Bureaucratic obstacles further complicate AI implementation, as administrative barriers often prevent necessary changes to course content or teaching methods, particularly when faced with rigid departmental policies. Time constraints severely limit innovation opportunities, especially for part-time or casual instructors who lack sufficient preparation time. Resource disparities exacerbate these challenges, with access to AI tools and training varying dramatically based on institutional wealth. Major universities occupy privileged positions in providing AI-related opportunities, while less-resourced institutions struggle to offer comparable support. These institutional factors collectively create an uneven landscape where AI integration depends largely on institutional privilege rather than educational merit. Finally, analyzing participants' GenAI experiences through these three theoretical lenses reveals complementary insights that individually would remain invisible, demonstrating the value of multi-theoretical approaches to understanding complex educational phenomena. PICRAT's practical focus reveals that most participants position GenAI in amplification rather than transformation roles, suggesting cautious integration that preserves existing pedagogical relationships while enhancing specific teaching functions. This pattern indicates that participants view AI as a tool for strengthening established practices rather than revolutionizing their approach to language instruction, reflecting both pedagogical conservatism and thoughtful resistance to technologically driven change. UTAUT explains why adoption varies significantly despite general awareness of GenAI's potential, revealing how structural constraints rather than individual attitudes primarily determine engagement levels. Institutional barriers such as lack of training support, precarious employment conditions, and time constraints create adoption obstacles that individual motivation cannot overcome, suggesting that successful AI integration requires systemic rather than individual solutions. The framework's emphasis on facilitating conditions particularly illuminates how participants' enthusiasm for AI exploration is constrained by material realities of academic employment that prioritize immediate teaching responsibilities over professional development. However, neither framework adequately addresses the power dynamics that critical pedagogy brings into focus, revealing the limitations of purely functional or adoption-focused approaches to understanding educational technology. P5's critique of the "butler's stance" and concerns about indigenous data sovereignty reveal how seemingly neutral technology integration decisions actually reinforce existing hierarchies and colonial knowledge practices that position language instruction as service rather than critical engagement. Similarly, participants' ethical concerns about student authenticity and multilingual writer bias cannot be understood solely through functional or adoption frameworks but require critical analysis of how AI technologies embody and perpetuate systemic inequalities. Together, these frameworks demonstrate that educational technology decisions operate simultaneously on practical, institutional, and ideological levels, requiring analysis that addresses technical functionality, adoption processes, and power relations. While PICRAT and UTAUT provide valuable insights into the mechanics of AI integration and the factors influencing adoption decisions, critical pedagogy reveals the deeper questions about whose interests are served and what kind of educational relationships are being constructed through these technological choices. This multi-theoretical approach suggests that effective AI integration in EAP contexts requires not only practical strategies and institutional support, but also critical examination of how these tools either advance or undermine emancipatory educational goals that honor student agency and challenge existing power structures. Declarations Ethics approval and consent to participate The study has been approved by the Research Ethics Bureau at the University of Toronto - protocol number 38253 Consent for publication Not applicable Availability of data and materials The dataset analysed during the current study is available from the corresponding author on reasonable request. Funding Not applicable. Acknowledgements Dr. Clare Brett contributed to the revision and editing of the paper. References Abbas M, Jam FA, Khan TI (2024) Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students. Int J Educational Technol High Educ 21(1):10. https://doi.org/10.1186/s41239-024-00444-7 Al-Abdullatif AM (2024) Modeling Teachers’ Acceptance of Generative Artificial Intelligence Use in Higher Education: The Role of AI Literacy, Intelligent TPACK, and Perceived Trust. Educ Sci 14(11). Article 11. https://doi.org/10.3390/educsci14111209 Al-kfairy M, Mustafa D, Kshetri N, Insiew M, Alfandi O (2024) Ethical Challenges and Solutions of Generative AI An Interdisciplinary Perspective. Informatics, 11 (3), Article 3. https://doi.org/10.3390/informatics11030058 Ayyoub AM, Khlaif ZN, Shamali M, Abu Eideh B, Assali A, Hattab MK, Barham KA, Bsharat TRK (2025) Advancing higher education with GenAI: Factors influencing educator AI literacy. Frontiers in Education , 10 . https://doi.org/10.3389/feduc.2025.1530721 Bai L, Liu X, Su J (2023) ChatGPT: The cognitive effects on learning and memory. Brain-X 1(3):e30. https://doi.org/10.1002/brx2.30 Baidoo-Anu D, Owusu Ansah L (2023) Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning (SSRN Scholarly Paper No. 4337484). Social Science Research Network. https://doi.org/10.2139/ssrn.4337484 Barrett A, Pack A (2023) Not quite eye to A.I.: student and teacher perspectives on the use of generative artificial intelligence in the writing process. Int J Educ Technol High Educ 20:59. https://doi.org/10.1186/s41239-023-00427-0 Beck SW, Levine SR (2023) Backtalk: ChatGPT: A powerful technology tool for writing instruction. Phi Delta Kappan 105(1):66–67. https://doi.org/10.1177/00317217231197487 Breshears S (2019) Precarious Work of English Language Teaching in Canada. TESL Can J 36(2):26–47. https://doi.org/10.18806/tesl.v36i2.1312 Bryman A (2016) Social Research Methods. Oxford University Press Bognár L, Khine MS (2025) The shifting landscape of student engagement: A pre-post semester analysis in AI-enhanced classrooms. Computers Education: Artif Intell 8:100395. https://doi.org/10.1016/j.caeai.2025.100395 Bushell C (2024) Investigating the Integration of AI Chat Models to Enhance Critical Thinking Skills through Engineered Prompting (SSRN Scholarly Paper 5017061). Social Science Research Network https://doi.org/10.2139/ssrn.5017061 Cacho R (2024) Integrating Generative AI in University Teaching and Learning. A Model for Balanced Guidelines Online Learn, 28(3). https://doi.org/10.24059/olj.v28i3.4508 Chan CKY (2023) A comprehensive AI policy education framework for university teaching and learning. Int J Educational Technol High Educ 20(1):38. https://doi.org/10.1186/s41239-023-00408-3 Chan CKY, Tsi LHY (2024) Will generative AI replace teachers in higher education? A study of teacher and student perceptions. Stud Educational Evaluation 83:101395. https://doi.org/10.1016/j.stueduc.2024.101395 Creely E (2024) Exploring the Role of Generative AI in Enhancing Language Learning: Opportunities and Challenges. Int J Changes Educ 1(3):158–167. https://doi.org/10.47852/bonviewIJCE42022495 Creely E, Henderson M (2025) Artificial intelligence and the digital dis/empowerment of migrant and refugee learners. In E. Tour, E. Creely, P. Waterhouse, & M. Henderson, Digital Empowerment for Refugee and Migrant Learners (1st ed., pp. 47–64). Routledge. https://doi.org/10.4324/9781003422891-5 Creswell JW (with, Poth CN (2018) Qualitative inquiry & research design: Choosing among five approaches (Fourth edition.). SAGE Crompton H, Edmett A, Ichaporia N, Burke D (2024) AI and English language teaching: Affordances and challenges. Br J Edu Technol 55(6):2503–2529. https://doi.org/10.1111/bjet.13460 Daniel K, Msambwa MM, Wen Z (2025) Can Generative AI Revolutionise Academic Skills D in Higher Education? A Systematic Literature Review. European J Educ, 60 (1), e70036. https://doi.org/10.1111/ejed.70036 Dergaa I, Saad B, Glenn H, Amamou JM, Aissa BB M., Guelmami, N., Fekih- Romdhane F, Chamari K (2024) From tools to threats. A reflection on the impact of artificial-intelligence chatbots on cognitive health Front Psychol, 15 https://doi.org/10.3389/fpsyg.2024.1259845 Eaton SE (2023) Postplagiarism: Transdisciplinary ethics and integrity in the age of artificial intelligence and neurotechnology. International Journal for Educational Integrity , 19 (1), 23, s40979-023-00144–1. https://doi.org/10.1007/s40979-023-00144-1 Fan L, Kim J, Detrick R, Li N (2025) Developing Students’ Higher-Order Thinking Skills With Generative AI Insights and Strategies From a Case Study: In A. ElSayary (Ed.), Advances in Computational Intelligence and Robotics (pp. 411–428). IGI Global. https://doi.org/10.4018/979-8-3693-7332-3.ch024 Fleischer A, Lowman J, Strader K (2025) Understanding Older Adults’ Intention to Use Telehealth A Qualitative Study Using the UTAUT Framework. Journal of Primary Care & Community Health 16, 21501319251320180. https://doi.org/10.1177/21501319251320180 Freire P (2018) Pedagogy of the oppressed (M. B. Ramos, Trans.; 50th anniversary ed.). Bloomsbury Academic Fu Q-K, Zou D, Xie H, Cheng G (2024) A review of AWE feedback: Types, learning outcomes, and implications. Comput Assist Lang Learn 37(1–2):179–221. https://doi.org/10.1080/09588221.2022.2033787 Gayed JM, Carlon MKJ, Oriola AM, Cross JS (2022) Exploring an AI-based writing Assistant’s impact on English language learners. Computers Education: Artif Intell 3:100055. https://doi.org/10.1016/j.caeai.2022.100055 Gerlich M (2025) AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies 15(1). Article 1. https://doi.org/10.3390/soc15010006 Glahn K (2024) Using ChatGPT to Teach English for Academic Purposes: Perspectives from the Classroom. J Educ Innov Communication 5(2):63–75. https://doi.org/10.34097/jeicom-5-2-December-23-5 Hibert AI (2019) Systematic Literature Review of Automated Writing Evaluation as a Formative Learning Tool. In M. Scheffel, J. Broisin, V. Pammer-Schindler, A. Ioannou, & J. Schneider (Eds.), Transforming Learning with Meaningful Technologies (Vol. 11722, pp. 199–212). Springer International Publishing. https://doi.org/10.1007/978-3-030-29736-7_15 Hopcan S, Türkmen G, Polat E (2024) Exploring the artificial intelligence anxiety and machine learning attitudes of teacher candidates. Education and Information Technologies 29 (6), 7281–7301. https://doi.org/10.1007/s10639-023-12086-9 Huber A (2018) Whatever happened to the CALL revolution? Hirao School Manage Rev 8:31–39 Kinzie S (2024) Language Ideologies and English for Academic Purposes Writing: A Case in Ontario: Larry Vandergrift Best Graduate Paper Award 2022. Can Mod Lang Rev 80(1):1–21. https://doi.org/10.3138/cmlr-2023-001 Klimova B, Pikhart M (2025) Exploring The Effects of Artificial Intelligence on Student and Academic Well-being in Higher Education: A Mini-Review. Front Psychol 16:1498132. https://doi.org/10.3389/fpsyg.2025.1498132 Klingbeil A, Grützner C, Schreck P (2024) Trust and reliance on AI — An experimental study on the extent and costs of overreliance on AI Comput Hum Behav, 160 , 108352. https://doi.org/10.1016/j.chb.2024.108352 Kohnke L, Moorhouse BL, Zou D (2023) ChatGPT for Language Teaching and Learning. RELC J 54(2):537–550. https://doi.org/10.1177/00336882231162868 Kouritzin SG, Ellis TF, Ghazani AZ, Nakagawa S (2023) Gigification of English Language Instructor Work in Higher Education: Precarious Employment and Magic Time. TESOL Q 57(4):1518–1544. https://doi.org/10.1002/tesq.3206 Lee BJ, Cook DR (2024) Exploring the potential of AI for pragmatics instruction Technology in Language Teaching & Learning, 6(3), Article 3 https://doi.org/10.29140/tltl.v6n3.1521 Lim JS, Shin D, Lee C, Kim J, Zhang J (2025) The Role of User Empowerment, AI Hallucination, and Privacy Concerns in Continued Use and Premium Subscription Intentions: An Extended Technology Acceptance Model for Generative AI. J Broadcast Electron Media 1–17. https://doi.org/10.1080/08838151.2025.2487679 Lindebaum D, Fleming P (2024) ChatGPT Undermines Human Reflexivity, Scientific Responsibility and Responsible Management Research Br J Manag, 35(2), 566–575. https://doi.org/10.1111/1467-8551.12781 Malik AR, Pratiwi Y, Andajani K, Numertayasa IW, Suharti S, Darwis A, Marzuki (2023) Exploring Artificial Intelligence in Academic Essay. Higher Education Student’s Perspective Int J Educational Res Open, 5, 100296 https://doi.org/10.1016/j.ijedro.2023.100296 Marzuki W, Utami R, Diyenti D, Indrawati I (2023) The impact of AI writing tools on the content and organization of students’ writing: EFL teachers’ perspective. Cogent Educ 10(2):2236469. https://doi.org/10.1080/2331186X.2023.2236469 Melo CO, Luft MCMS, Rocha RO (2021) Influencing elements of technological adoption Case study about management in an educational institution. Contextus – Revista Contemporânea de Economia e Gestão , 19 , 124–145. https://doi.org/10.19094/contextus.2021.61445 Miao F, Holmes W, Huang R, Zhang H et al (2021) AI and education: A guidance for policymakers. UNESCO Publishing Morozov E (2013) To Save Everything. The Folly of Technological Solutionism, Click Here PublicAffairs Noy C (2008) Sampling Knowledge: The Hermeneutics of Snowball Sampling in Qualitative Research. Int J Soc Res Methodol 11(4):327–344. https://doi.org/10.1080/13645570701401305 Nyaaba M, Wright A, Choi GL (2024) Generative AI and Digital Neocolonialism in Global Education: Towards an Equitable Framework (Version 3). arXiv. https://doi.org/10.48550/ARXIV.2406.02966 Ogunleye B, Zakariyyah K, Ajao O, Olayinka O, Sharma H (2024) Higher education assessment practice in the era of generative AI tools. J Appl Learn Teach 7(1):1–11. https://doi.org/10.37074/jalt.2024.7.1.28 Pratama A, Sulistiyo U (2024) A Systematic Review of Artificial Intelligence in Enhancing English Foreign Learners’ Writing Skill. PPSDP Int J Educ 3(2):170–181. https://doi.org/10.59175/pijed.v3i2.299 Preiksaitis C, Rose C (2023) Opportunities, Challenges, and Future Directions of Generative Artificial Intelligence in Medical Education: Scoping Review. JMIR Med Educ 9:e48785. https://doi.org/10.2196/48785 Pokrivcakova S (2019) Preparing teachers for the application of AI-powered technologies in foreign language education. J Lang Cult Educ 7(3):135–153. https://doi.org/10.2478/jolace-2019-0025 Raimes A (1991) Out of the Woods: Emerging Traditions in the Teaching of Writing. TESOL Quarterly 25 (3), 407–430. https://doi.org/10.2307/3586978 Schretzlmaier P, Hecker A, Ammenwerth E (2022) Suitability of the Unified Theory of Acceptance and Use of Technology 2 Model for Predicting mHealth Acceptance Using Diabetes as an Example Qualitative Methods Triangulation Study. JMIR Hum Factors, 9(1), e34918. https://doi.org/10.2196/34918 Shahid MK, Zia T, Bangfan L, Iqbal Z, Ahmad F (2024) Exploring the relationship of psychological factors and adoption readiness in determining university teachers’ attitude on AI-based assessment systems International Journal of Management Education (Elsevier Science), 22 (2), N.PAG-N.PAG. https://doi.org/10.1016/j.ijme.2024.100967 Solak E (2024) Revolutionizing Language Learning: How ChatGPT and AI are changing the Way We Learn Languages. Int J Technol Educ 7(2):353–372. https://doi.org/10.46328/ijte.732 Song C, Song Y (2023) Enhancing academic writing skills and motivation. Assessing the efficacy of ChatGPT in AI-assisted language learning for EFL students. Frontiers in Psychology 14 , 1260843. https://doi.org/10.3389/fpsyg.2023.1260843 Stolpe K, Hallström J (2024) Artificial intelligence literacy for technology education Computers and Education Open 6, 100159. https://doi.org/10.1016/j.caeo.2024.100159 Sun T (2023) Potential Use of Artificial Intelligence in ESL Writing Assessment: A Case Study of IELTS Writing Tasks. Ir J Technol Enhanced Learn 7(2):42–51. https://doi.org/10.22554/ijtel.v7i2.137 Swindell A, Greeley L, Farag A, Verdone B (2024) Against Artificial Education Towards an Ethical Framework for Generative Artificial Intelligence (AI) Use in Education Online Learning 28 (2). https://eric.ed.gov/?id=EJ1428206 Tseng W, Warschauer M (2023) AI-writing tools in education: If you can’t beat them, join them. J China Computer-Assisted Lang Learn 3(2):258–262. https://doi.org/10.1515/jccall-2023-0008 Venkatesh V, Morris MG, Davis GB, Davis FD (2003) User Acceptance of Information Technology: Toward a Unified View. MIS Q 27(3):425–478. https://doi.org/10.2307/30036540 Yan L, Martinez-Maldonado R, Gasevic D (2024) Generative Artificial Intelligence in Learning Analytics: Contextualising Opportunities and Challenges through the Learning Analytics Cycle. Proceedings of the 14th Learning Analytics and Knowledge Conference , 101–111. https://doi.org/10.1145/3636555.3636856 Yatani K, Sramek Z, Yang C-L (2024) AI as Extraherics: Fostering Higher-order Thinking Skills in Human-AI Interaction (No. arXiv:2409.09218). arXiv. https://doi.org/10.48550/arXiv.2409.09218 Yuan B, Hu J (2024) Generative AI as a Tool for Enhancing Reflective Learning in Students (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2412.02603 Walsh Marr J (2021) The Promise and Precarity of Critical Pedagogy in English for Academic Purposes. BC TEAL J 6(1):132–141 Walter Y (2024) Embracing the future of Artificial Intelligence in the classroom: The relevance of AI literacy, prompt engineering, and critical thinking in modern education. Int J Educational Technol High Educ 21(1):15. https://doi.org/10.1186/s41239-024-00448-3 Wang Z (2022) Computer-assisted EFL writing and evaluations based on artificial intelligence: A case from a college reading and writing course. Libr Hi Tech 40(1):80–97. https://doi.org/10.1108/LHT-05-2020-0113 Wang B, Rau P-LP, Yuan T (2023) Measuring user competence in using artificial intelligence Validity and reliability of artificial intelligence literacy scale. Behaviour & Information Technology 42 (9), 1324–1337. https://doi.org/10.1080/0144929X.2022.2072768 Wiboolyasarin W, Wiboolyasarin K, Tiranant P, Jinowat N, Boonyakitanont P (2025) AI-driven chatbots in second language education: A systematic review of their efficacy and pedagogical implications. Ampersand 14:100224. https://doi.org/10.1016/j.amper.2025.100224 Wolters A, Von Straussenburg A, A. F., Riehle DM (2024) AI Literacy in Adult Education—A L Review. https://hdl.handle.net/10125/107211 Wu Y (2024) Critical Thinking Pedagogics Design in an Era of ChatGPT and Other AI Tools—Shifting From Teaching What to Teaching Why and How. J Educ Dev 8(1):1. https://doi.org/10.20849/jed.v8i1.1404 Yuan Z, Lyu T (2024) Association Between AI Chatbot Self-efficacy and EFL Student Class-related Anxiety: A Control-Value Theory Perspective. Proceedings of the (2024) 9th International Conference on Distance Education and Learning , 371–375. https://doi.org/10.1145/3675812.3675822 Zepeda L, Benavides I, Lopez C, Roman O, Dominguez Y (2023) Guided Design Methodology for Activities That Integrate Generative Artificial Intelligence Tools in Education A, Pilot Study Promoting Ethical Use and Fostering Critical Thinking in Graduate Students (2023). ICERI 7283–7292. 16th annual International Conference of Education, Research and Innovation. https://doi.org/10.21125/iceri.2023.1811 Zhang S, Zhao X, Zhou T, Kim JH (2024) Do You Have AI Dependency? The Roles of Academic Self-Efficacy Academic Stress, and Performance Expectations on Problematic AI Usage Behavior International Journal of Educational Technology in Higher Education 21. https://doi.org/10.1186/s41239-024-00467-0 Additional Declarations The authors declare no competing interests. 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-7014667","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":478713117,"identity":"14384dfe-524a-4c4e-ae72-d5d3bf295b28","order_by":0,"name":"Rana Haidar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABFUlEQVRIie2QMUvDQBTH33FwXY7e+oLBz3AlEDPpV7EE4nLVwSWo1Ejh3Nz9Jo4pAbtc6RqogyGQqUNFkE5i7CAWrgQ3h/tN94734//nATgc/xWSAYccyCsA+n9SqGwVDkC3n9ilQKuw761u5eh+Pl2Tp2O/P5s3V+VNxAEvpnWavowF0GptUXxzHiMxMffMWbhUz20xjOnAmAa9jAW2KAQlkWjKZZ6wpWJbhXl3ukCZc2s7FKtgQ/Qtl4uGXarPXYVubAqqsE0puCwTRkd6V2HWFFyF0VDPuFc29GD0gJzxOhhkpvAeJyyMrMVUUL7p65P+IiHv6mN8KHrDqsrSQojepC73Xfr098B+XnTfvsPhcDi6+AI2HFXXuI6CqQAAAABJRU5ErkJggg==","orcid":"","institution":"University of Toronto","correspondingAuthor":true,"prefix":"","firstName":"Rana","middleName":"","lastName":"Haidar","suffix":""},{"id":478713159,"identity":"4c7cf22b-5e5f-457d-a543-5ce919c7ea4e","order_by":1,"name":"Athena Tassis","email":"","orcid":"","institution":"University of Toronto","correspondingAuthor":false,"prefix":"","firstName":"Athena","middleName":"","lastName":"Tassis","suffix":""}],"badges":[],"createdAt":"2025-07-01 00:29:16","currentVersionCode":1,"declarations":{"humanSubjects":true,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":true,"humanSubjectConsent":true,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-7014667/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7014667/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":85809779,"identity":"8215153b-8a3c-4f21-92c8-15c90708b73d","added_by":"auto","created_at":"2025-07-02 03:26:34","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":923940,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7014667/v1/826a79e0-2b5a-4c48-abd9-627f054fd91d.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003ePower, Pedagogy, and Algorithms: An Exploratory Study of EAP Instructors' Critical Engagement with Generative AI\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe increasing availability of Generative Artificial Intelligence (GenAI) tools like ChatGPT, DeepSeek, Google Gemini, Perplexity, and Claude presents new challenges and affordances \u0026nbsp;for English for Academic Purposes (EAP) instructors as they navigate potential implications for the instruction of academic writing. Recent scholarship shows Large Language Models (LLMs) can support academic writing by offering interactive, personalized learning experiences that adapt to each user\u0026apos;s writing style and pace (Pokrivcakova, 2019; Wang, 2024), which mirrors one-on-one tutoring and helps writers improve through immediate feedback and hands-on practice (Crompton et al., 2024). Also, the GenAI tools\u0026rsquo; constant availability makes learning more flexible and accessible (Lin, 2024), and studies show they can minimize language learning anxiety and improve learner motivation (Asio \u0026amp; Suero, 2024; Klimova \u0026amp; Pikhart, 2025). On the flip side, scholars argue that AI-generated content can lead to over-reliance on automation, ultimately reducing students\u0026apos; ability to develop independent writing skills and critical thinking (Marzuki et al., 2023; Barrett \u0026amp; Pack, 2023; Kohnke et al., 2025). Additionally, a major concern for language instructors is the reinforcement of dominant linguistic norms and ideas through GenAI applications that might impact non users of English and marginalize indigenous languages, reducing accessibility (Nyaaba et al., 2024). Specifically, AI tools tend to generate content that is biased toward dominant Western cultural norms ultimately limiting space for non-Western and marginalized perspectives (Creely \u0026amp; Henderson 2025).\u003c/p\u003e\n\u003cp\u003eThis seismic shift in the integration of new technologies in writing instruction not only impacts students, but also plays a pivotal role in shaping curriculum, pedagogy, and how educators integrate these applications into their daily practices. Notably, the generative characteristics of LLMs have the potential to transform our conception of authorship, leading to what scholars like Eaton (2023) term a \u0026quot;postplagiarism\u0026quot; paradigm where collaborative creation between humans and artificial intelligence becomes standard practice. Understanding how writing instructors perceive this looming paradigm shift and their strategies for navigating it is essential for developing appropriate responses to human-AI collaborative authorship in academic contexts and defining the evolving role of writing instructors in an AI infused academic world (Baidoo-Anu \u0026amp; Owusu Ansah, 2023; Kinzie, 2024). Chan and Tsi (2024) predict that GenAI can enhance educational efficiency by automating administrative tasks while supporting teachers with course design, information gathering, content generation, and assessment processes, enabling them to focus on higher-level teaching responsibilities, while Cacho (2024) suggests that GenAI can help streamline content creation, curate materials, and support learning preferences.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDespite this burgeoning literature on AI in education, there remains limited research examining how academic writing instructors perceive and incorporate generative AI tools into their teaching practices. This study addresses this research gap by investigating two key questions: 1) What perceptions do English for Academic Purposes instructors hold regarding the integration of Generative Artificial Intelligence in education? and 2) To what extent and in what ways are these instructors incorporating GenAI tools into their pedagogical practices and instructional approaches? Through exploring these questions, our research offers valuable insights into the current landscape of GenAI adoption in academic writing instruction in Canadian higher education, revealing both emerging practices and instructors\u0026apos; predominantly cautious responses\u0026mdash;characterized by critical evaluation, skepticism about pedagogical value, and tentative acceptance rather than widespread enthusiasm. By critically interrogating whether and where AI tools belong in the EAP classroom, examining the conditions under which these technologies are adopted, and interrogating \u0026nbsp;the assumptions that shape their use, we contribute to ongoing discourse about if and how AI tools can be leveraged responsibly to support, rather than replace, human instruction in EAP contexts.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLiterature Review\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEvolving Technologies and the role of AI in Academic Writing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe role of AI in academic writing instruction is part of a broader historical trend in language and literacy education, where emerging technologies have repeatedly reshaped teaching and pedagogy. Earlier tools such as computer-assisted language learning (CALL) provided users with structured feedback and grammar correction (Huber, 2018). The development of CALL paved the way for more sophisticated systems in the 1990\u0026rsquo;s such as Automated Writing Evaluation (AWE), which uses natural language processing to provide immediate, formative feedback on various aspects of student writing and is often used for iterative drafting (Hibert, 2019). However, the feedback provided was often generic and less effective at fostering higher-order writing skills without teacher mediation (Fu et al., 2024). The impact of these early tools was modest, despite claims that these new technologies would significantly disrupt language learning. Traditional teacher-led instruction remained central, partly because the educational structures and pedagogical approaches were not fundamentally disrupted by these new tools (Huber, 2018). By the early 1990s, tools like Microsoft Word introduced built-in grammar checks, and by the 2010s, assistants like Grammarly and WordTune expanded functionality offering real-time suggestions on writing style yet still focusing mainly on surface-level corrections (Gayed et al., 2022). These innovations allowed some instructors to shift their focus away from correcting grammar and spelling, instead spending more instructional time on supporting their students to develop content and higher-order thinking skills (Hibert, 2019). As these foundational tools evolved and instructors adapted their pedagogical focus, the emergence of GenAI marked a new phase in writing instruction.\u003c/p\u003e\n\u003cp\u003eResearchers have begun examining how GenAI tools are used in writing instruction; emerging studies emphasize that AI can support not only efficiency, but also deeper engagement and higher-order thinking in writing, especially when instructors design tasks that prompt students to reflect on and refine AI-generated content\u0026nbsp;(Yatani et al., 2024). Some recent studies show that generative AI tools can enhance students\u0026rsquo; cognitive engagement in writing, particularly by supporting activities such as sharing opinions, explaining concepts, and engaging in analysis. This effect is most pronounced when students interact with AI in a reflective and iterative manner, rather than using it only for basic tasks (Hu et al., 2024; Nguyen et al., 2024; Jin et al., 2025). Emergent literature that shows advanced uses of AI in writing\u0026mdash;such as transforming and refining text\u0026mdash;are linked to improvements in critical thinking, motivation, and writing quality (Malik et al., 2023; Pratama and Sulistiyo, 2024; Jin et al, 2025). However, there are concerns that overreliance on AI could impact creativity and critical thinking, highlighting the need for balanced integration and reflective task design. The quality and depth of student engagement with AI depend heavily on how instructors design writing tasks (Tu et al., 2024). Tasks that require students to analyze, critique, and refine AI-generated content foster deeper engagement and higher-order thinking compared to tasks that simply use AI for technical support.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAffordances and Limitations of AI in Writing Instruction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGenerative AI offers both promising affordances for writing instruction and raises notable limitations. AI can provide benefits of immediate, personalized feedback and scaffolding that might otherwise require significant instructional time and individual attention from instructors - resources that are often constrained in many academic settings. For instance, Solak (2024) suggests that ChatGPT can provide students with real-time, individualized scaffolding for both the generative and refinement phases of academic writing, addressing instructional gaps that often exist due to classroom time constraints. Furthermore, Marzuki et al. (2023) conducted interviews with EFL instructors and found that a variety of AI writing tools (QuillBot, Wordtune, GPT, Paperpal, etc.) were perceived to improve students\u0026rsquo; writing quality especially in the content and organization of essays. Other studies have found that integrating ChatGPT into writing instruction offers multiple benefits: it can support in-depth revision processes (Beck \u0026amp; Levine, 2023), enhance grammar and vocabulary development (Crompton et al., 2023), modify the complexity of challenging texts, encourage critical comparison between student-authored and AI-produced content (Tseng \u0026amp; Warschauer, 2023), and elevate the quality of EFL student writing in terms of structure and coherence (Marzuki et al., 2023). In a classroom study, EAP students found ChatGPT especially useful for brainstorming ideas and improving the structure and clarity of their essays (Glahn, 2023): by generating examples or alternative wordings, the AI helped students expand their ideas and vocabulary in ways a limited classroom timeframe might not always allow. GenAI\u0026rsquo;s affordances also include personalized and adaptive learning where tools like ChatGPT can tailor feedback and learning experiences based on students\u0026rsquo; individual writing profiles as well as their language proficiency (Wu, 2024). By leveraging deep learning and natural language processing, these tools can identify students\u0026rsquo; specific areas for improvement and provide targeted, individualized guidance to address them. Wu (2024) also notes that ChatGPT supports learners by generating relevant examples and explanations that align with their input, which helps clarify complex ideas and build writing fluency. Furthermore, the research highlights that AI tools can adjust the difficulty level of tasks and recommend targeted resources, making the learning experience more relevant and engaging. This adaptive feedback loop creates a more student-centered environment compared to traditional instruction models (Wu, 2024). Additionally, Kohnke, et al. (2023) highlight that ChatGPT can be used to identify a meaning of a word in context, develop quizzes, annotate texts and provide translations. More specifically, Kohnke et al. (2023) discuss the use of ChatGPT as an aid to support students and instructors in language learning, where teachers provide guided tasks or prompts, and students interact with ChatGPT to clarify meaning, ask follow-up questions, and sustain the conversation to deepen their understanding. Overall, the literature underscores the growing role of generative AI as a valuable support tool in writing instruction.\u003c/p\u003e\n\u003cp\u003eHowever, alongside these benefits lie important limitations and risks. Accuracy and bias are two major pitfalls in the use of GenAI. Sun (2023) notes that ChatGPT-generated content can omit critical information or even fabricate details including producing false citations and missing key data points. Users must therefore scrutinize AI output carefully. Additionally, Creely et al. (2025) highlight that AI models often reflect the biases of their training data, and for the foreseeable future, learners who are navigating new contexts and new languages or who have limited AI literacy are at risk with regards to their agency as a result of these embedded biases.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFurthermore, the integration of generative AI in education presents substantial challenges across three critical domains. Academic integrity concerns have emerged as educators grapple with increased opportunities for misuse, technological plagiarism, and the inconsistency in institutional policies governing appropriate use (Preiksaitis \u0026amp; Rose, 2023) . According to Preiksaitis and Rose\u0026rsquo;s (2023) scoping review,\u0026nbsp;equally concerning are accuracy and dependability issues, as these systems often operate with outdated information, produce convincing hallucinations, and perpetuate biases from their training data (Lim et al., 2023). Perhaps most worrying are the potential detriments to learning, where overdependence on AI threatens to undermine critical thinking skills, and compromise assessment validity (Ogunleye et al., 2024). This sentiment resonates broadly across higher education, where research indicates that uncritical use of generative AI may undermine students\u0026apos; critical thinking abilities and memory retention (Abbas et al., 2024; Bai et al., 2023; Gerlich, 2025). Navigating this complex landscape requires a two-pronged approach: first, developing essential AI literacies among both educators and students to foster critical engagement with these tools; and second, conducting rigorous empirical research to evaluate AI\u0026apos;s actual impacts on learning outcomes and pedagogical effectiveness. Only through this combined focus on literacy development and evidence-based assessment can we ensure that AI tools enhance rather than diminish meaningful learning experiences in higher education.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFactors Facilitating or Hindering AI Integration in EAP\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe degree to which EAP instructors adopt generative AI in their teaching largely depends on a combination of personal, institutional, and contextual factors. Recent studies have begun identifying key drivers and barriers that influence instructors\u0026rsquo; acceptance of AI in the classroom. In higher education, instructors are more inclined to integrate GenAI tools when they perceive them as useful and easy to use. According to Al-Abdullatif (2024), perceived ease of use was the strongest predictor of GenAI acceptance, while AI literacy and intelligent Technological Pedagogical and Content Knowledge (TPACK) played important indirect roles. Instructors who were confident using GenAI tools and found them accessible were more likely to integrate them into their teaching. However, having experience with educational technology in general did not automatically translate into readiness to use GenAI. Many instructors lacked direct, hands-on experience with these tools, which limited their ability to recognize their usefulness in enhancing learning outcomes. This suggests that familiarity with GenAI itself is essential for adoption. These findings point to the need for targeted, context-specific training that supports both technical skills and pedagogical application.\u003c/p\u003e\n\u003cp\u003eThe supportiveness of the institutional environment is also a significant external factor in an instructor\u0026rsquo;s willingness to adopt GenAI tools. Research has consistently shown that the presence of clear policies, access to structured training, and sustained professional development opportunities are critical in fostering instructor confidence and readiness (Saihi et al., 2025). A recent study by Ayyoub et al., 2025 found that educators build GenAI literacy through two main pathways: formal professional development and active community support. Drawing on the TPACK framework, the study shows that while professional development shapes perceptions of usefulness, it is community support that directly builds AI competence. The findings also point to the importance of teacher efficacy, strengthened through peer support and mentoring, and call for policy frameworks that can guide the thoughtful integration of AI in teaching. Additionally, Jian et al. (2024) highlight that many educators feel underprepared to integrate GenAI into assessment and instructional practices due to a lack of institutional guidance and discipline-specific frameworks. Their review found that most universities had yet to develop comprehensive policies, and even where such policies existed, they were often vague or perceived as insufficient. Moreover, facilitating conditions\u0026mdash;such as infrastructure, training, and administrative support\u0026mdash;consistently predict AI adoption (Saihi et al., 2025; Perez, 2024). Social influence and perceived usefulness also shape faculty members\u0026rsquo; behavioral intentions, particularly when supported by strong institutional frameworks. Faculty training and peer encouragement further enhance acceptance. In short, the literature points in the direction of job security, strategic organizational support, and cultural alignment as essential criteria for successful AI integration in academic settings.\u003c/p\u003e\n\u003cp\u003eThe conditions under which instructors work can also affect AI adoption. In a study of adult English language teaching in Canada, Breshears (2019) identifies low wages, reliance on part-time employment, job insecurity, lack of benefits, limited administrative support, and extensive unpaid labor as persistent challenges in the EAP sector. These factors often combine to create unstable and demanding work environments, making it difficult for instructors to commit time to professional development and may impact the amount of energy they give to learning new technologies. As Breshears notes, these employment conditions are widespread and have long shaped the lived experiences of EAP instructors in Canada. Similar concerns are raised by Kouritzin et al. (2023), who describe how unstable contracts and unclear rehiring practices create uncertainty and limit EAP instructors\u0026rsquo; ability to plan ahead. This ongoing insecurity may contribute to the hesitation or inability to engage with new tools and approaches, including the use of educational technology.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTheoretical Framework\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research investigates how English for Academic Purposes instructors perceive and implement Generative Artificial Intelligence in their teaching practices. The study employs three complementary theoretical frameworks to develop its interview protocol and analyze instructor experiences: the PICRAT technology integration model, the Unified Theory of Acceptance and Use of Technology (UTAUT), and critical pedagogy. While the PICRAT and UTAUT models provide valuable insights into technology integration patterns and adoption factors, they do not adequately address the power dynamics inherent in AI adoption within educational contexts. Therefore, this study incorporates critical pedagogy as an essential framework to examine how GenAI may reinforce or challenge existing hierarchical structures in language education.\u003c/p\u003e\n\u003cp\u003eThe PICRAT model, developed by Kimmons, Graham, and West (2020), provides a structured approach to evaluating technology integration through a matrix examining two dimensions. The first dimension considers students\u0026apos; relationship with technology along a continuum from passive reception to interactive engagement to creative production. The second dimension assesses how technology impacts teaching practice, whether merely replacing traditional tools, amplifying existing practices, or transformatively enabling previously impossible learning experiences. This framework helps determine whether instructors view GenAI as simple replacements for existing resources or as transformative elements that fundamentally alter teaching and learning. PICRAT is particularly valuable because it \u0026quot;emphasizes technology as a means to an end\u0026quot; and \u0026quot;focuses on students,\u0026quot; ensuring technology serves language learning objectives rather than becoming an end in itself (Kimmons et al., 2020). However, the model\u0026apos;s focus on functionality and student-centered outcomes overlooks critical questions about who controls AI technologies, whose voices and perspectives are embedded in AI systems, and how AI adoption might perpetuate or disrupt existing power imbalances between instructors and students, institutions and educators, or dominant and marginalized linguistic communities.\u003c/p\u003e\n\u003cp\u003eThe Unified Theory of Acceptance and Use of Technology (UTAUT), developed by Venkatesh and colleagues (2003), identifies four key determinants influencing technology adoption: performance expectancy (beliefs about effectiveness), effort expectancy (perceived ease of use), social influence (impact of colleagues and institutional culture), and facilitating conditions (technical and organizational infrastructure). In the context of GenAI adoption by EAP instructors, this framework helps explore barriers and enablers instructors experience, examining not only individual attitudes but also how institutional contexts and professional networks shape adoption decisions. While UTAUT effectively captures the mechanics of technology acceptance, it operates from a largely positivist perspective that treats technology adoption as a neutral process of optimization. This approach fails to interrogate the political dimensions of AI implementation, such as how institutional mandates for AI adoption may constrain instructor autonomy, or how AI tools developed by profit-driven corporations may not align with pedagogical values of social justice and student empowerment.\u003c/p\u003e\n\u003cp\u003eCritical pedagogy, rooted in Paulo Freire\u0026apos;s (1968) seminal work Pedagogy of the Oppressed, offers a theoretical lens for examining power dynamics in educational settings that the PICRAT and UTAUT models overlook. This approach rejects the \u0026quot;banking\u0026quot; concept of education where students passively receive knowledge, instead advocating for education as an emancipatory practice developing critical consciousness. Critical pedagogy views education as inherently political and aimed at fostering social justice, recognizing that GenAI tools are not neutral technologies but products embedded with specific values, biases, and power relations. As Freire argued, critical pedagogy \u0026quot;involves both the recognition that human life is conditioned, not determined, and the crucial necessity of not only reading the world critically but also intervening in the larger social order as part of the responsibility of an informed citizenry\u0026quot; (Freire, 1968). This perspective frames essential questions about how EAP instructors perceive GenAI\u0026apos;s role in either reinforcing traditional power structures or enabling more democratic approaches to language learning. By incorporating critical pedagogy alongside PICRAT and UTAUT, this study can examine not only how instructors integrate and adopt AI technologies, but also how they navigate questions of agency, equity, and resistance in AI-mediated educational environments.\u003c/p\u003e\n\u003cp\u003eTogether, these three frameworks provide complementary perspectives for investigating GenAI integration in EAP contexts. While PICRAT offers a practical tool for categorizing technology use, UTAUT explains the adoption process itself, and critical pedagogy prompts deeper questions about power, equity, and the purpose of technology in education. The interview protocol addresses both practical aspects of GenAI integration and deeper pedagogical, ethical, and institutional considerations. Questions explore how instructors perceive GenAI\u0026apos;s potential to transform existing practices, navigate power dynamics and equity concerns, and what factors influence their adoption decisions.\u003c/p\u003e\n\u003cp\u003eIn the data analysis phase, these frameworks serve as interpretive lenses: PICRAT helps categorize and evaluate reported GenAI uses; UTAUT frames understanding of factors influencing varying adoption levels; and critical pedagogy guides analysis of how instructors negotiate issues of power, access, and equity. This multi-theoretical approach acknowledges that educational technology decisions are never merely technical but fundamentally connected to questions of pedagogy, power, equity, and institutional context (Weisberg and Dawson, 2023).\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThis qualitative case study investigated academic writing instructors\u0026apos; perceptions of generative AI integration in teaching and learning contexts, examining both implementation practices and factors influencing adoption or non-adoption decisions. In-depth, semi-structured interviews were conducted with nine experienced academic writing instructors employed at post-secondary institutions in Ontario, Canada. Semi-structured interviews were chosen as the primary data collection method because they allow for systematic exploration of predetermined themes while maintaining the flexibility to pursue unexpected insights that emerge during conversations (Bryman, 2016). The following is a sample of the questions on the interview protocol in this study: \u0026ldquo;Can you tell me about your experience teaching EAP? How would you describe your familiarity with AI technologies in education? Are you currently using any AI tools in your teaching of academic writing? Why or why not?\u0026rdquo;\u003c/p\u003e\n\u003cp\u003eThis approach was particularly well-suited to the study\u0026apos;s objectives as it enabled the researchers to probe deeply into instructors\u0026apos; complex experiences with AI integration while allowing participants to share perspectives that might not have been anticipated in the research design. The semi-structured format also facilitated comparison across interviews while preserving the nuanced, contextual details essential for understanding individual instructor experiences with emerging technologies.\u003c/p\u003e\n\u003cp\u003eParticipants were recruited through purposive snowball sampling, a method particularly effective for accessing specialized populations with specific expertise (Creswell and Poth, 2018). Given the relatively recent emergence of generative AI in educational contexts, instructors with substantial experience using these tools in academic writing instruction represent a specialized and somewhat difficult-to-identify population, making snowball sampling an appropriate recruitment strategy. However, this sampling approach carries inherent limitations that must be acknowledged; 1) potentially creating homogeneous samples as participants tend to refer others within their professional networks who likely share similar characteristics, experiences, or perspectives (Noy, 2008). This may limit the diversity of viewpoints captured in the study, potentially excluding instructors with different institutional contexts, pedagogical approaches, or levels of AI engagement and. 2) introducing selection bias, as participants willing to discuss AI integration may represent those who are more comfortable or positive about these technologies, potentially underrepresenting instructors who are skeptical, resistant, or have had negative experiences with AI tools. Despite these limitations, snowball sampling remained the most feasible approach for identifying and accessing this specialized population of early AI adopters in academic writing instruction.\u003c/p\u003e\n\u003cp\u003eInterviews were conducted remotely via video conferencing, recorded with participant consent, and professionally transcribed for analysis. The interview protocol, informed by the theoretical frameworks of PICRAT, critical pedagogy, and UTAUT, explored participants\u0026apos; understanding of generative AI capabilities, their pedagogical approaches to integration, perceived benefits and challenges, concerns about equity and power dynamics, and institutional factors influencing adoption.\u003c/p\u003e\n\u003cp\u003eThematic analysis was employed to identify patterns across the data corpus, with initial coding conducted independently by two researchers using NVivo to enhance reliability. The coding scheme was iteratively refined through researcher consensus meetings until thematic saturation was achieved. Member checking was conducted by sharing preliminary findings with participants to verify interpretations and enhance validity. All research procedures received approval from the institutional research ethics board prior to participant recruitment.\u003c/p\u003e"},{"header":"Results and Discussion","content":"\u003cp\u003e \u003cb\u003eParticipant Technology Engagement Spectrum\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe participants showed significantly diverse levels of engagement with AI technologies: While some participants have extensively explored advanced tools like ChatGPT and various image generators, integrating them thoughtfully into their teaching practices, others candidly identify themselves as beginners with only cursory or superficial experience with these technologies. This diversity reflects a natural progression of technology adoption in education, with participants generally acknowledging the substantial learning curve associated with AI implementation and viewing it as an ongoing process of discovery. On one side of the continuum, we have P4 who has been exploring GenAI application in teaching and learning since the inception of ChatGPT, and on the other side is P9 who has not yet explored the uses of LLMs in teaching writing because she hasn’t felt the need to call on AI for instructional support. The remaining seven participants fall between those two points in terms of the breadth and depth of exploration and use of AI for teaching.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePedagogical Integration and Evolving Teaching Practices\u003c/b\u003e \u003c/p\u003e\u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003e \u003cb\u003eStudent-Centered AI Literacy Development\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003eWhen discussing AI use in teaching, all participants primarily reflected on guiding students in utilizing AI tools rather than their own instructional practices. Participants expressed particular concern with developing students' critical AI literacy—their ability to thoughtfully evaluate and critique AI outputs rather than accepting them uncritically. They emphasized teaching students to use AI as supplementary resources enhancing learning rather than replacing knowledge and skill development. This pedagogical priority reflects educators' commitment to ensuring students become discerning AI users maintaining intellectual agency while leveraging technological advantages, aligning with AI literacy best practices (Wolters, 2024; Stolpe \u0026amp; Hallström, \u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). P8 offers nuanced responses to student AI questions, foregrounding context: \"[I]t depends on the purpose. Is this for an assignment? Are you submitting this where you're saying that this is your language work? Is this a task where the focus is on language? Or is it a task where the focus is on expressing your idea? For me. It really is about context.\"\u003c/p\u003e \u003cp\u003eParticipants focus on teaching effective chatbot use to mitigate potential harms (Shanmugasundaram and Tamilarasu, 2023). From a PICRAT perspective, participants' emphasis on student training reflects deliberate GenAI positioning in the Interactive-Amplify quadrant, where technology enhances critical thinking rather than replacing student cognitive processes. This pedagogical choice maintains technology as means to educational ends rather than ends themselves, aligning with PICRAT's student-centered evaluation framework. Several participants have been fielding student questions about ethical and effective GenAI use, making their classroom approach necessarily reactive. P1 suggests clear usage guidelines encourage positive student engagement with these tools. Similarly, P6 guides students on leveraging GenAI to support rather than replace their thinking and learning processes. Acknowledging students already use AI without guidance on effective and ethical use, she teaches critical thinking by emphasizing starting the writing process with their own ideas, using AI for idea organization and brainstorming.\u003c/p\u003e \u003cp\u003eP9 argues academic writing courses help graduate students develop disciplinary writing expertise and emphasizes gaining foundational skills before using GenAI tools—otherwise, students struggle to develop field expertise and lack the ability to critique AI outputs. This position resonates with Zhang et al.'s (\u003cspan citationid=\"CR112\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) recent study postulating that AI users can only verify information when they already know the answer, and as topics become unfamiliar, AI reliance increases even when AI provides explanations. Lindbaum and Fleming (2023) compare academics outsourcing writing to AI to \"turkeys voting for Christmas,\" cheering their own doom. Moreover, they note ChatGPT produces outputs based on derivative, high-probability choices, \"has no stake in the knowledge it produces and is thus likely prone to offering irresponsible outputs\" (p. 566, emphasis in original). They contrast this with authentic human thought shaped by socio-cultural context and employing unconventional choices within and beyond existing data boundaries, rendering it creative, situated, and contextually relevant.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePractical Applications of AI in Writing Instruction\u003c/b\u003e \u003c/p\u003e \u003cp\u003eParticipants offered detailed accounts of their experiences and perspectives about AI in writing instruction. The majority of participants in this study conceptualized chatbots as complementary tools within their broader educational toolkit, where, as P6 observed, the technology currently \"fits in kind of interstitially\" within established curricula and institutional practices. Some of the participants expressed positive perceptions towards AI as a potential thinking partner: P1 highlighted GenAI's potential as an interactive partner for pedagogical reflection, particularly valuable in the often isolating academic environment. She appreciated how language models can provide feedback on assignment design, helping to assess critical thinking skills. Innovatively, P1 uses ChatGPT as a litmus test for assignment quality - if the AI can generate a passing paper, she views this as a sign that her rubric is too superficial and fails to capture substantive human thought. This approach demonstrates a nuanced method of using AI to enhance educational assignment design by ensuring deeper, more meaningful student engagement. This method underscores the participant's commitment to developing assignments that necessitate human cognition and avoid surface-level responses, while also demonstrating an innovative use of GenAI in pedagogical practice - an emerging practice in the literature on AI\u0026amp;Ed (see for e.g. Bushell, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Zepeda et al., \u003cspan citationid=\"CR109\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Similarly, P2 who works part-time as an educational consultant said that she uses GenAI as a sounding board and a professional development resource given her recent engagement in the field. P7 noted that she uses GenAI to help her brainstorm ideas for classroom activities, using it like she would a search engine to see what has been done on a certain topic. She noted that she never takes the output at face value, rather she assesses the fit of the output to her needs and modifies it accordingly, a sentiment which P3 shares as well. Several participants shared that they use GenAI to create exemplars of different genres of academic writing like abstracts or introductions, and P7 shared that she sometimes uses GenAI to generate paraphrases or to show students the different ways a sentence can be structured, which saves her a lot of time to spend on more sophisticated tasks in the classroom. Several participants shared that GenAI tools are useful for creating exemplars to teach genre awareness or to generate “dummy text” to teach citation and referencing skills. These tasks were described as “rote” or “supplementary” versus more high-stakes tasks like creating assignment instructions, a task that all participants agreed cannot and should not be relegated to AI.\u003c/p\u003e \u003cp\u003eThe majority of applications fall into the Interactive-Amplify category, where GenAI enhances existing pedagogical practices while maintaining instructor agency and student-centered learning objectives. P1's innovative \"litmus test\" approach for assignment design exemplifies this integration pattern, using ChatGPT to evaluate whether assignments require genuine human cognition rather than surface-level responses. Similarly, P7's use of GenAI for sentence restructuring demonstrations and P6's application for literature review organization represent amplification of established teaching methods rather than wholesale replacement of instructor expertise.\u003c/p\u003e \u003cp\u003eParticipants used GenAI-generated content as a teaching tool for critical analysis. P4 noted that students enjoy critiquing GenAI-generated content because “they're not worried about hurting anybody's feelings. They can just trash it, basically.” By examining ChatGPT's outputs, which often contain pragmatic language errors like overly formal or too chummy language, students can learn to identify and correct inappropriate communication styles. This approach allows for teaching language pragmatics through both high-quality and flawed AI-generated examples, encouraging critical thinking about effective communication. Lee and Cook’s (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) study on ChatGPT’s pragmatic competence confirms that this is indeed an expedient use of chatbots. The researchers' qualitative analysis revealed ChatGPT's inconsistent performance in language tasks: while the AI struggles with crafting appropriately nuanced apologies, lacking politeness and contextual language sensitivity, it demonstrates more competence in generating requests and refusals. P6 sees similar potential in GenAI that she says can be useful to generate texts with a variety of registers, and P4 shared that ChatGPT might be useful as a teaching tool for generating academic texts that students can critically analyze. She notes that the outputs of ChatGPT often lack specificity, originality, and coherence, making them ideal examples for students to evaluate against academic rubrics. In her teaching approach, P8 uses AI-generated content to help students develop critical writing skills, showing how chatbots \"don't add information\" but \"can make the language pretty.\" Her goal is to demystify AI use, encouraging open classroom discussion about its capabilities and limitations.\u003c/p\u003e \u003cp\u003eP4 reveals that her \"ulterior motive\" in engaging students with this process is to demonstrate the limitations of ChatGPT's outputs. The goal is to foster students' critical evaluation skills and discourage over-reliance on AI while maintaining their autonomy in learning. The literature indeed warns about the perils of excessive “cognitive offloading” to AI (Gerlich, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Cognitive offloading refers to using external tools and agents like AI to reduce cognitive load, which in theory can enhance efficiency without necessarily diminishing cognitive engagement. However, excessive reliance on these tools, especially AI, may potentially compromise deep thinking and critical analysis. These concerns are substantiated by Gerlich's (2025) study, which demonstrated a significant negative correlation between frequent AI tool usage and critical thinking abilities, with cognitive offloading serving as a mediating factor.\u003c/p\u003e \u003cp\u003eP6 said GenAI can be useful for teachers and graduate writers to help them sort through massive amounts of literature by creating literature review matrices, the caveat being that the machine would only be responsible for performing “the manual labor” of sorting through ideas, but the ideas themselves would be the purview of the user only, which falls within the Passive-Replace quadrant, where technology substitutes for traditionally manual processes.\u003c/p\u003e \u003cp\u003eImage generation seems to be a popular feature among participants who use multimodal texts in their teaching material and find non-copyrighted GenAI generated images convenient and modifiable. P5 shared that one way to use image generators is to create an activity that aims to cultivate critical thinking, language precision, and evaluative skills by having students assess image descriptors generated by AI, some created imperfect by design. Students would be asked to critically evaluate whether the descriptors accurately represent the images; by requiring students to analyze, interpret, and evaluate AI-generated content, the teacher encourages higher-order thinking, active engagement with technology, and practical application of knowledge, “aligning with the university's focus on independent thought and nuanced understanding”.\u003c/p\u003e \u003cp\u003eFew truly transformative applications emerged that would qualify for the Creative-Transform quadrant, though P4's approach of having students critique AI outputs to develop metacognitive awareness and P5's critical analysis activities examining AI-generated image descriptors suggest potential for more radical pedagogical innovation. This PICRAT analysis reveals that participants primarily view GenAI as amplifying their existing pedagogical toolkit rather than fundamentally transforming their teaching practices, suggesting a cautious, enhancement-focused approach to AI integration that preserves human agency in educational relationships.\u003c/p\u003e \u003cp\u003e \u003cb\u003eGenAI and Language Learners\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAll of the participants in this study teach student populations who can be described as English language learners. When asked if they thought GenAI offered any affordances to multilingual learners, several participants observed that GenAI can be useful for multilingual graduate students, as it might help reduce learning anxiety by improving their diction, syntax, and overall writing accuracy. For instance, P8 maintains a practical perspective, acknowledging that \"if it were me in a non-language program, I'd use [AI] to check my vocabulary”. This stance resonates with much of the literature on AI tools and language learning anxiety that posits AI tools can enhance learners' autonomy and self-efficacy, which reflects in lower learning anxiety levels (Yuan, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Song, \u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Jubier, 2024).\u003c/p\u003e \u003cp\u003eOn the other hand, P1, P8 and P9 expressed overlapping concerns regarding multilingual learners use of AI tools as they suggested there might be a risk that the authenticity of multilingual writers' work might be unfairly questioned because the formal style often adopted by many multilingual writers can sometimes resemble AI-generated text in terms of style, potentially leading to unwarranted suspicions of AI use even when none has occurred Increasing bias against multilingual writers. Several of the participants also expressed concern over learning loss; P8 cautioned that if \"the purpose is to show your linguistic ability, then that might not be the best idea.\"\u003c/p\u003e \u003cp\u003eNotably, P4’s remarks about the attitudes of the international students in her writing program which caters to engineering students reveal a welcome resistance among high-achieving international engineering students toward adopting AI tools for writing assistance, challenging prevailing assumptions in the literature. P4 explained that the high achieving students in this program “would not deign to use AI in their learning”. However, when these students inquire about AI, they demonstrate sincere interest in understanding ethical and effective implementation. This authentic curiosity motivates P4 and colleagues to explore pedagogical approaches to AI that best address their students' educational needs.\u003c/p\u003e \u003cp\u003eIn yet another astute observation, P7 observed a nuanced tension in current discussions around AI use in education. On one hand, much of the existing literature emphasizes concerns about academic integrity and the potential erosion or underdevelopment of students’ writing skills due to reliance on AI. On the other hand, P7 highlighted a more subtle but growing issue among multilingual students: a new form of anxiety stemming from the perception that AI might eventually \"write better than they can.\" This perception exacerbates existing insecurities many students already feel when writing in English-medium classrooms. P7 warned that this could lead educators to adopt a \"deficit approach\"—essentially abandoning efforts to develop students' writing abilities and instead allowing AI to do the work for them.\u003c/p\u003e \u003cp\u003eThe participants' concerns about multilingual learners reveal what critical pedagogy would recognize as intersectional oppression, where AI adoption may compound existing linguistic marginalization rather than alleviating it. P1, P8, and P9's worry that multilingual writers' work might be \"unfairly questioned\" because their formal style resembles AI text highlights how technological bias can reinforce linguistic prejudice, creating additional barriers for students already navigating language-based discrimination. This reflects Freire's understanding that educational tools are never neutral but always carry the potential to either humanize or dehumanize learners, with AI technologies risking the further marginalization of students whose linguistic identities are already undervalued in academic contexts.\u003c/p\u003e \u003cp\u003e \u003cb\u003eBalancing Innovation and Integrity: Educators’ Perspectives on AI’s Role in Writing, Learning, and Critical Engagement\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eCognitive Concerns and Learning Loss\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe participants in this study unanimously emphasized, albeit in varying ways, that the complex cognitive processes involved in producing written work could be undermined if students were to consume too much AI-generated content. Many participants expressed concerns about potential learning loss that could be triggered by over-reliance on AI technologies in academic writing. The literature shows that over-reliance on AI chatbots could increase the risk of critical cognitive skill atrophy and the decline of problem-solving abilities (Dergaa et al., 2024) and might lead to a decline in autonomy and decision-making skills (Klingbeil et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)In other words, over-reliance on AI in education could be a recipe for entrenching the banking model of education (Freire, 1970) where students enter a prompt into the machine and accept the output without critical engagement, questioning its assumptions, or understanding the complex processes behind the generated content. This passive consumption of AI-produced content mirrors the very educational dynamics that Freire critiqued, where learners become receptacles of information rather than co-creators of knowledge through active engagement, dialogue and critical reflection.\u003c/p\u003e \u003cp\u003e \u003cb\u003eQuality and Authenticity of AI-Generated Writing\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe participants in this study expressed fundamental skepticism about the rhetorical and intellectual value of chatbot-generated writing, identifying significant limitations in its capacity to produce nuanced, authentic, or contextually appropriate text that would meet standards of academic discourse. Their critiques revealed not merely disappointment with current output quality, but deeper concerns about the algorithmic reproduction of language that lacks genuine rhetorical awareness and critical thought. P1 described the language produced by chatbots as flowery and vapid, lacking in depth and critical thinking; P3 described the style of AI outputs as generic, while P4 shared that the ChatGPT is good at saying things that “sound intelligent” but lack real substance.\u003c/p\u003e \u003cp\u003e \u003cb\u003eDiverse Pedagogical Philosophies Toward AI Integration\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe educators displayed varying attitudes toward AI in education. P3 observed a leveling of student fascination with AI as they recognize its limitations, while P1 adopted a pragmatic stance, advocating for teaching effective and ethical AI use since students were already utilizing it. P2, who teaches adult language learners, expressed skepticism about teaching ethical AI use, stating that \"you can't teach trust.\" P4 demonstrated the most positive outlook toward AI, as her students used the technology responsibly while maintaining personal effort in their work. P5 approached AI with critical concern, focusing on how language models reproduce biases and arguing that critical language analysis should examine who is included, excluded, and what information is privileged - though she noted these considerations often became secondary to practical implementation in English for Academic Purposes contexts. P6 took a utilitarian perspective on AI implementation. she advocated for maintaining AI as a supplementary resource that enhances thinking processes and facilitates brainstorming, rather than allowing it to replace student thinking altogether. Her position emphasized keeping AI \"in its place\" - establishing clear boundaries around its use to ensure it remains a supportive tool that augments human cognition rather than substituting for it. This approach suggests P6 sees value in AI when properly constrained within a supporting role in the educational process. P7 takes a literacy-focused approach to AI in education, particularly within her writing center context where it is not the writing consultants’ purview to set or enforce AI policies. She emphasizes the importance of student awareness, noting \"we cannot really dictate what the students should be doing\" but instead guides them to \"read the syllabus\" and understand course-specific policies. P7’s perspective extends beyond basic skills development to a critical literacy framework where students learn to \"evaluate generated text that's produced by AI in terms of biases, in terms of power dynamics, in terms of whose voices are presented.\" When working with students permitted to use AI, she proposes comparative analysis, having students \"compare their own writing with the writing of AI\" to develop critical awareness. This approach cultivates \"genre awareness\" by examining whether AI-generated text \"follows genre requirements\" and evaluating language choices, particularly for multilingual students. Despite limited personal experience with AI implementation, P7 envisions it serving as \"a study buddy\" - a supportive tool to be deployed when needed. P8 demonstrates a practical, context-driven approach to AI integration in education. Despite working with a set curriculum that lacks formal AI components, she incorporates AI discussions \"interstitially,\" particularly when teaching academic integrity. P9, who has perhaps engaged the least with AI, presents a balanced view of AI in education, emphasizing the classroom's role as a technology-independent learning space. She positions the classroom as \"a place to give students tools so that they can do it without it,\" recognizing that while AI might offer benefits, students \"could surely experiment with it themselves individually.\" P9 values preserving traditional learning approaches, suggesting instructors should ensure students know \"how people have done this without this tool, just in case.\"\u003c/p\u003e \u003cp\u003eWhen considering potential AI applications, P9 focuses on analytical capabilities rather than content generation. She expresses interest in AI that could \"diagnose things for students,\" such as identifying when \"this paragraph lacks coherence\" or highlighting structural weaknesses. Her perspective is shaped by her instructional priorities - \"argumentation, paragraph structure, coherence\" and teaching students to \"emphasize what needs to be emphasized and be loud and overt about what it is that they're trying to say.\"\u003c/p\u003e \u003cp\u003eHowever, P9 remains skeptical about current AI capabilities in this domain, noting \"I don't think it's there yet\" and characterizing her desired analytical AI tools as \"a bit far off.\" She wants AI that could assist with critical thinking by \"pointing out logical flaws\" or identifying missing structural elements like topic sentences, suggesting she values AI as a potential diagnostic tool rather than a replacement for student writing.\u003c/p\u003e \u003cp\u003e \u003cb\u003eCognitive Concerns and Learning Loss\u003c/b\u003e \u003c/p\u003e \u003cp\u003eParticipants unanimously emphasized that complex cognitive processes in written work could be undermined by excessive AI-generated content consumption. Many expressed concerns about potential learning loss from over-reliance on AI technologies in academic writing. Literature shows over-reliance on AI chatbots increases risk of critical cognitive skill atrophy and declining problem-solving abilities (Dergaa et al., 2024) and may reduce autonomy and decision-making skills (Klingbeil et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Over-reliance on AI could entrench Freire's banking model of education (1970) where students enter prompts and accept outputs without critical engagement, questioning assumptions, or understanding complex processes behind generated content. This passive AI consumption mirrors educational dynamics Freire critiqued, where learners become information receptacles rather than knowledge co-creators through active engagement, dialogue, and critical reflection.\u003c/p\u003e \u003cp\u003e \u003cb\u003eQuality and Authenticity of AI-Generated Writing\u003c/b\u003e \u003c/p\u003e \u003cp\u003eParticipants expressed fundamental skepticism about chatbot-generated writing's rhetorical and intellectual value, identifying significant limitations in producing nuanced, authentic, or contextually appropriate text meeting academic discourse standards. Their critiques revealed not merely disappointment with output quality, but deeper concerns about algorithmic language reproduction lacking genuine rhetorical awareness and critical thought. P1 described chatbot language as flowery and vapid, lacking depth and critical thinking; P3 called AI outputs generic, while P4 noted ChatGPT excels at saying things that \"sound intelligent\" but lack real substance.\u003c/p\u003e \u003cp\u003e \u003cb\u003eParticipants AI Attitudes\u003c/b\u003e \u003c/p\u003e \u003cp\u003eEducators displayed varying AI attitudes. P3 observed leveling student fascination as they recognize AI limitations, while P1 adopted pragmatic stances, advocating teaching effective and ethical AI use since students already utilize it. P2, teaching adult language learners, expressed skepticism about teaching ethical AI use, stating \"you can't teach trust.\" P4 demonstrated the most positive AI outlook, as her students used technology responsibly while maintaining personal effort. P5 approached AI with critical concern, focusing on how language models reproduce biases and arguing critical language analysis should examine inclusion, exclusion, and privileged information—though noting these considerations often became secondary to practical English for Academic Purposes implementation.\u003c/p\u003e \u003cp\u003eP6 took utilitarian perspectives on AI implementation, advocating maintaining AI as supplementary resources enhancing thinking processes and facilitating brainstorming rather than replacing student thinking. Her position emphasized keeping AI \"in its place\"—establishing clear boundaries ensuring it remains supportive tools augmenting human cognition rather than substituting for it. This approach suggests P6 sees AI value when properly constrained within supporting educational roles.\u003c/p\u003e \u003cp\u003eP7 takes literacy-focused AI approaches, particularly within writing center contexts where consultants don't set or enforce AI policies. She emphasizes student awareness importance, noting \"we cannot really dictate what the students should be doing\" but guides them to \"read the syllabus\" and understand course-specific policies. P7's perspective extends beyond basic skills to critical literacy frameworks where students learn to \"evaluate generated text that's produced by AI in terms of biases, in terms of power dynamics, in terms of whose voices are presented.\" When working with students permitted to use AI, she proposes comparative analysis, having students \"compare their own writing with the writing of AI\" to develop critical awareness. This approach cultivates \"genre awareness\" by examining whether AI-generated text \"follows genre requirements\" and evaluating language choices, particularly for multilingual students. Despite limited personal AI experience, P7 envisions it serving as \"a study buddy\"—supportive tools deployed when needed.\u003c/p\u003e \u003cp\u003eP8 demonstrates practical, context-driven AI integration approaches. Despite working with set curricula lacking formal AI components, she incorporates AI discussions \"interstitially,\" particularly when teaching academic integrity. P9, who has engaged least with AI, presents balanced AI education views, emphasizing classrooms' roles as technology-independent learning spaces. She positions classrooms as \"a place to give students tools so that they can do it without it,\" recognizing that while AI might offer benefits, students \"could surely experiment with it themselves individually.\" P9 values preserving traditional learning approaches, suggesting instructors ensure students know \"how people have done this without this tool, just in case.\"\u003c/p\u003e \u003cp\u003eWhen considering potential AI applications, P9 focuses on analytical capabilities rather than content generation. She expresses interest in AI that could \"diagnose things for students,\" such as identifying when \"this paragraph lacks coherence\" or highlighting structural weaknesses. Her perspective is shaped by instructional priorities—\"argumentation, paragraph structure, coherence\" and teaching students to \"emphasize what needs to be emphasized and be loud and overt about what it is that they're trying to say.\"\u003c/p\u003e \u003cp\u003eHowever, P9 remains skeptical about current AI capabilities, noting \"I don't think it's there yet\" and characterizing desired analytical AI tools as \"a bit far off.\" She wants AI assisting critical thinking by \"pointing out logical flaws\" or identifying missing structural elements like topic sentences, suggesting she values AI as potential diagnostic tools rather than student writing replacements.\u003c/p\u003e \u003cp\u003e \u003cb\u003eEthical Considerations and Ethical “Discomfort”\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eMultifaceted Ethics Beyond Academic Integrity\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe participants in this study displayed sophisticated levels of awareness of the ethical dilemmas associated with the use of GenAI in education. For example, when P5 was asked if she had any ethical concerns about AI, she countered by asking \"but which ethics? Is it environmental ethics, or the ethics of labor associated with the training of the models?\" This response reveals P5's nuanced understanding of AI ethics as a multifaceted issue rather than a singular concern, demonstrating her awareness that ethical considerations span diverse dimensions from environmental impact to labor practices in AI development. By reframing the question, P5 highlights how discussions of AI ethics often lack specificity and fail to address the complex, interconnected ethical frameworks necessary for meaningful evaluation.\u003c/p\u003e \u003cp\u003eIn line with most of the literature about the concerns around the use of GenAI in education, the participants in the study are well aware of the likelihood of GenAI being used by students, with a few stating that it is not always possible to tell whether a student has used AI to write a part or all of their paper or if they’ve used it as a thought partner. Notably, many of the participants consistently demonstrated a nuanced understanding that students' engagement with GenAI is inevitable in contemporary educational settings, or as P4 put it, “if you can’t beat them, join them”, in reference to her awareness of her students use of AI in their learning and her attempts at meeting them where they are.\u003c/p\u003e \u003cp\u003e \u003cb\u003eData Privacy and Sovereignty Concerns\u003c/b\u003e \u003c/p\u003e \u003cp\u003eMany of the participants expressed ethical concerns over losing data privacy and data sovereignty were the practice of using AI to give feedback to students to become widespread. To illustrate, P8 expressed concerns regarding inputting student work in the bot without the students’ consent and showed equal concern about the sources of information the AI provides in its outputs which are scraped from thousands of uncited internet sources. P5 shed light on a profound and often overlooked issue in the discussion of AI ethics in education which is that of indigenous data sovereignty and the double standards that exist when it comes to protecting them: “Indigenous communities have been working for years to protect their knowledges and data sovereignty. Those same people who are up in arms about ethics and cheating - are they defending the data sovereignty of indigenous and other marginalized communities? Are they looking at the built-in data bias of scraping of different communities? Of the reification of misogyny and racism of these large sexist, racist language banks that they're drawing from? If we want to get into ethics, yes, bring it on it's overdue, but make it broad based.” This framing challenges us to broaden our ethical considerations around AI beyond narrow institutional concerns to include questions of power, representation, and justice. It suggests that truly ethical AI engagement requires a more comprehensive approach that centers traditionally marginalized perspectives rather than just protecting established academic norms (cite feminist critique). The call to make ethics discussions \"broad based\" is essentially asking for consistency in our ethical standards - if we care about proper use of knowledge and information in one context, we should apply similar principles of respect and consent across all contexts, especially regarding communities whose knowledge has historically been appropriated without acknowledgment or compensation.\u003c/p\u003e \u003cp\u003eThe participants' understanding of the ethical issues surrounding the use of AI tools is very nuanced as is befitting of professional teachers of writing. All of the participants agreed that teachers also have a responsibility to acknowledge their own use of AI in the preparation of any pedagogical material.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePower Dynamics and Resistance in AI Adoption\u003c/b\u003e \u003c/p\u003e \u003cp\u003eCritical pedagogy's focus on power relations illuminates several tensions in participants' AI experiences that functional frameworks like PICRAT and UTAUT cannot adequately address. The dynamics of institutional power versus instructor autonomy emerged clearly as participants navigate between institutional expectations for AI literacy and their pedagogical values. P9's emphasis on preserving \"technology-independent learning spaces\" represents resistance to technologically deterministic approaches to education, asserting the classroom as a site where human agency and critical thinking can develop without technological mediation.\u003c/p\u003e \u003cp\u003eQuestions of knowledge authority and student agency reveal complex power negotiations around AI adoption. P4's observation that high-achieving international students \"would not deign to use AI\" reveals student resistance to technological mediation of their learning, suggesting agency in rejecting tools perceived as undermining their intellectual development. This resistance challenges assumptions about technology adoption as inevitably beneficial, instead revealing how students themselves may critique AI integration as inconsistent with their educational values and goals.\u003c/p\u003e \u003cp\u003ePerhaps most significantly, P5's powerful critique of AI ethics highlights how discussions of academic integrity often ignore \"the data sovereignty of indigenous and other marginalized communities\" whose knowledge systems have been appropriated to train AI models without consent or compensation. This analysis reveals how AI adoption can perpetuate colonial knowledge practices while appearing ethically neutral, embodying what critical pedagogy recognizes as the hidden curriculum of technological implementation. From a critical pedagogy perspective, these dynamics reveal that AI adoption is never merely a technical decision but always involves questions of power, agency, and whose interests are served by technological integration.\u003c/p\u003e \u003cp\u003e \u003cb\u003eBarriers to GenAI Integration in Academic Writing\u003c/b\u003e \u003c/p\u003e \u003cp\u003eP5 offers a critical perspective on how generative AI fits into existing problematic attitudes toward language education where she identifies AI's appeal as rooted in a utilitarian view that sees \"language as a means to an end and not its own necessarily knowledge base.\" P5 connects current AI enthusiasm to historical perspectives of English for Academic Purposes that encourages a service model or what Raimes (\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e1991\u003c/span\u003e) calls “the butler’s stance” (p. 243), where language instructors “get [students’] English ready and then send them to the University where [subject-matter instructors] will do the real work\" (p.420). P5 sees generative AI as perpetuating this problematic division, with many hoping AI will \"just remove that barrier\" to language, feeding into a fantasy that \"no one's going to have to study or really engage critically with language anymore\" because \"the generative AI will produce what's needed\", effectively identifying a technosolutionist fantasy that bypasses the critical consciousness-raising that Freire argued was essential to authentic education. P5's critique of GenAI's appeal as rooted in utilitarian views of language directly challenges what Freire (1968) identified as the banking concept of education, where knowledge becomes a commodity to be deposited rather than co-created through critical dialogue. Her analysis suggests AI is being embraced not as an educational tool but as a technosolutionist bypass for meaningful language engagement (Morozov, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). The appeal of AI as a technological fix for language learning reflects what Freire would recognize as a false solution that addresses symptoms rather than root causes of educational inequality, potentially reinforcing the very power imbalances that critical pedagogy aims to disrupt.\u003c/p\u003e \u003cp\u003eOverall, the participants expressed concern that the un-critical adoption of GenAI risks entrenching the very educational dynamics that critical pedagogy seeks to transform, where learners become passive recipients of AI-generated content rather than active co-creators of knowledge through critical engagement and dialogue. As P9 pointed out, graduate academic writing courses facilitate the acquisition of foundational academic writing skills like “how to make an argument, how to synthesize other people's arguments, how to insert your own stance, how to signal your stance”, and these skills should not be bypassed by asking the machine to do the writing, even if the machine is capable of performing those tasks. Questions abound in the literature about the skills we lose when we offload our thinking to the bots (Creely et. al., 2025), and teachers everywhere are expressing similar hesitation and skepticism regarding the impact of chatbots on skill development (Titko et al., 2023; Aljunaid, 2024; Mohammadkarimi, 2023). P9 views the writing classroom as a critical space where students develop expertise in genre conventions and scholarly communication, and subverting this foundational cognitive exercise would deprive students of foundational communication skills. Her skepticism towards AI's current capabilities in academic writing is multifaceted, critically examining the tool's potential through a pedagogical lens. When she remarks, \"if it could diagnose things for students... that would be really cool. I don't think it's there yet,\" she directly challenges AI's diagnostic abilities. Her critique centers on the nuanced aspects of academic writing that require sophisticated understanding: \"argumentation, paragraph structure, coherence... How do you integrate a quotation into something that actually has impact?\". P9’s stance complicates UTAUT’s notion of performance expectancy by questioning the depth of GenAI’s usefulness in academic writing. While she acknowledges its value for surface-level feedback, P9 critiques its inability to support deeper rhetorical and argumentative skills, which she sees as central to writing pedagogy. This skepticism highlights a limitation in UTAUT: its focus on perceived utility overlooks disciplinary values and pedagogical goals. By emphasizing the cognitive and epistemic dimensions of writing, P9 reframes “usefulness” not as efficiency but as meaningful learning, challenging the assumption that technological adoption in education is inherently beneficial or pedagogically neutral. This perspective resonates with P3's assertion that \"AI lacks pedagogical knowledge\" and is substantiated by emerging literature that emphasizes GenAI's limitation to statistical probabilities of text generation, rather than genuine reasoning or critical thinking (Yan et al., \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Daniel et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003ePrecarity versus Professional Development\u003c/b\u003e \u003c/p\u003e \u003cp\u003eWhile the participants in this study displayed keen awareness of the need to stay updated on AI capabilities and applications, eight out of the nine participants in this study shared that they occupy precarious teaching positions in their institutions as PhD candidates with casual employment contracts. This element of job insecurity loomed large in the background of the participants’ responses as an important factor determining the depth and nature of instructor engagement with AI. For instance, when asked what factors might encourage her to adopt AI in her pedagogy, P9 readily said “paid training”. As a contract worker, her week is fully booked with contractual hours, and she doesn’t have the time or the financial resources to seek voluntary AI training. On the flip side, P1 shared that precarity might drive some instructors who are short of paid hours to outsource some of their work to AI, like using it to generate feedback for students or create lessons. Time constraints were also an important factor restricting the participants’ capacity to explore the pedagogical affordances of AI. In general, effort expectancy, a UTAUT construct, emerged as a significant barrier to adoption, with all participants acknowledging the substantial learning curve required for meaningful GenAI integration.\u003c/p\u003e \u003cp\u003eFacilitating conditions, another UTAUT construct, presented the most significant constraints on adoption with resource limitations and institutional support gaps creating barriers that individual motivation cannot overcome. This is inline with a lot of literature on the importance of institutional proactivity in encouraging technological adoption (Walter, \u003cspan citationid=\"CR98\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Keengwe et al., 2009; Porter and Graham, 2016). While all of the participants confirmed their awareness of the presence of professional development opportunities at their institutions, given the “chronic precarity of EAP employment” (Walsh Marr, \u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e2021\u003c/span\u003e, p.139), it is understandable that AI exploration is not top of mind for them. Additionally, the participants in this study who all have substantial teaching experience frequently asserted that their established pedagogical expertise, research background, and professional knowledge-base rendered GenAI superfluous for their practice.\u003c/p\u003e "},{"header":"Conclusions/Future Directions","content":"\u003cp\u003eThis study sought to explore the perceptions of 9 EAP instructors about the integration of GenAI in the teaching of academic writing. Several themes have emerged from the data regarding the affordances and limitations of GenAI and the factors that might encourage or hinder the integration of GenAI in EAP pedagogy and instruction.\u003c/p\u003e\u003cp\u003eThe participants in this study demonstrated high levels of confidence in their pedagogical skills. Their nuanced and cautious attitudes towards the use of GenAI in teaching and learning do not arise from technological incompetence, limited digital literacy, or diminished self-efficacy—contrary to findings in some recent literature (Wang et al., 2024; Shahid et al., \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Hopcan et al., 2023). Instead, this measured hesitation stems from a genuine, profound concern over the future of human thinking, creativity, and intellectual autonomy as well as sporadic and unfocused institutional support for teaching with GenAI technology. Their reservations reflect a deep philosophical apprehension about the potential erosion of critical thinking skills, the authentic development of student creativity, and the fundamental nature of human cognitive processes when AI becomes increasingly integrated into educational environments. This concern is increasingly reflected in the literature, where scholars have warned that the expanding role of AI in education may compromise students’ critical thinking as well as their intellectual independence (Abbas et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Bai et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Gerlich, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). These educators are not resisting technological innovation, but are instead advocating for a thoughtful, intentional approach that preserves the irreplaceable human elements of teaching and learning—where technology serves as a supportive tool rather than a replacement for human intellectual engagement and pedagogical nuance.\u003c/p\u003e\u003cp\u003eThe participants’ responses reflect a strategic shift from prohibition toward thoughtful and cautious integration, with a deliberate focus on helping students understand the potential pitfalls of over-reliance on AI. Participants emphasized three key areas of concern: the risk of bypassing essential learning processes when AI is used as a shortcut, the fundamental need for students to develop authentic disciplinary knowledge that AI cannot replace, and the cultivation of sophisticated AI literacy that enables students to critically evaluate AI-generated content. Rather than adopting a defensive stance, these educators are proactively developing pedagogical approaches that acknowledge GenAI as part of the learning landscape while safeguarding the core educational mission of developing independent, critical thinkers capable of navigating an increasingly AI-mediated world.\u003c/p\u003e\u003cp\u003eInconsistent institutional frameworks and resource disparities impact instructor’s access to professional development and weakens their motivation to explore new educational technologies. Educational institutions lack uniform policies regarding AI use, creating a disjointed environment where instructors must navigate conflicting expectations on their own. While many institutions require AI statements in syllabi, they often fail to consider teachers' technological preferences, skill levels, and available resources. This departmental and institutional policy fragmentation creates an atmosphere of uncertainty, with most instructors expressing the need for clearer, more functional guidelines especially on the pedagogical applications of AI. Bureaucratic obstacles further complicate AI implementation, as administrative barriers often prevent necessary changes to course content or teaching methods, particularly when faced with rigid departmental policies. Time constraints severely limit innovation opportunities, especially for part-time or casual instructors who lack sufficient preparation time. Resource disparities exacerbate these challenges, with access to AI tools and training varying dramatically based on institutional wealth. Major universities occupy privileged positions in providing AI-related opportunities, while less-resourced institutions struggle to offer comparable support. These institutional factors collectively create an uneven landscape where AI integration depends largely on institutional privilege rather than educational merit.\u003c/p\u003e\u003cp\u003eFinally, analyzing participants' GenAI experiences through these three theoretical lenses reveals complementary insights that individually would remain invisible, demonstrating the value of multi-theoretical approaches to understanding complex educational phenomena. PICRAT's practical focus reveals that most participants position GenAI in amplification rather than transformation roles, suggesting cautious integration that preserves existing pedagogical relationships while enhancing specific teaching functions. This pattern indicates that participants view AI as a tool for strengthening established practices rather than revolutionizing their approach to language instruction, reflecting both pedagogical conservatism and thoughtful resistance to technologically driven change.\u003c/p\u003e\u003cp\u003eUTAUT explains why adoption varies significantly despite general awareness of GenAI's potential, revealing how structural constraints rather than individual attitudes primarily determine engagement levels. Institutional barriers such as lack of training support, precarious employment conditions, and time constraints create adoption obstacles that individual motivation cannot overcome, suggesting that successful AI integration requires systemic rather than individual solutions. The framework's emphasis on facilitating conditions particularly illuminates how participants' enthusiasm for AI exploration is constrained by material realities of academic employment that prioritize immediate teaching responsibilities over professional development.\u003c/p\u003e\u003cp\u003eHowever, neither framework adequately addresses the power dynamics that critical pedagogy brings into focus, revealing the limitations of purely functional or adoption-focused approaches to understanding educational technology. P5's critique of the \"butler's stance\" and concerns about indigenous data sovereignty reveal how seemingly neutral technology integration decisions actually reinforce existing hierarchies and colonial knowledge practices that position language instruction as service rather than critical engagement. Similarly, participants' ethical concerns about student authenticity and multilingual writer bias cannot be understood solely through functional or adoption frameworks but require critical analysis of how AI technologies embody and perpetuate systemic inequalities.\u003c/p\u003e\u003cp\u003eTogether, these frameworks demonstrate that educational technology decisions operate simultaneously on practical, institutional, and ideological levels, requiring analysis that addresses technical functionality, adoption processes, and power relations. While PICRAT and UTAUT provide valuable insights into the mechanics of AI integration and the factors influencing adoption decisions, critical pedagogy reveals the deeper questions about whose interests are served and what kind of educational relationships are being constructed through these technological choices. This multi-theoretical approach suggests that effective AI integration in EAP contexts requires not only practical strategies and institutional support, but also critical examination of how these tools either advance or undermine emancipatory educational goals that honor student agency and challenge existing power structures.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study has been approved by the Research Ethics Bureau at the University of Toronto - protocol number 38253\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset analysed during the current study is available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDr. Clare Brett contributed to the revision and editing of the paper.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003e\u003cspan\u003eAbbas M, Jam FA, Khan TI (2024) Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students. Int J Educational Technol High Educ 21(1):10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-024-00444-7\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAl-Abdullatif AM (2024) Modeling Teachers\u0026rsquo; Acceptance of Generative Artificial Intelligence Use in Higher Education: The Role of AI Literacy, Intelligent TPACK, and Perceived Trust. Educ Sci 14(11). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 11. https://doi.org/10.3390/educsci14111209\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAl-kfairy M, Mustafa D, Kshetri N, Insiew M, Alfandi O (2024) Ethical Challenges\u0026nbsp;\u003c/span\u003e\u003cspan\u003eand Solutions of Generative AI An Interdisciplinary Perspective. Informatics, \u003cem\u003e11\u003c/em\u003e(3), Article 3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/informatics11030058\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAyyoub AM, Khlaif ZN, Shamali M, Abu Eideh B, Assali A, Hattab MK, Barham KA, Bsharat TRK (2025) Advancing higher education with GenAI: Factors influencing educator AI literacy. \u003cem\u003eFrontiers in Education\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2025.1530721\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBai L, Liu X, Su J (2023) ChatGPT: The cognitive effects on learning and memory. Brain-X 1(3):e30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/brx2.30\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBaidoo-Anu D, Owusu Ansah L (2023) \u003cem\u003eEducation in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning\u003c/em\u003e (SSRN Scholarly Paper No. 4337484). Social Science Research Network. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2139/ssrn.4337484\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBarrett A, Pack A (2023) Not quite eye to A.I.: student and teacher perspectives on the use of generative artificial intelligence in the writing process. Int J Educ Technol High Educ 20:59. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-023-00427-0\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBeck SW, Levine SR (2023) Backtalk: ChatGPT: A powerful technology tool for writing instruction. Phi Delta Kappan 105(1):66\u0026ndash;67. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/00317217231197487\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBreshears S (2019) Precarious Work of English Language Teaching in Canada. TESL Can J 36(2):26\u0026ndash;47. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18806/tesl.v36i2.1312\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBryman A (2016) Social Research Methods. Oxford University Press\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBogn\u0026aacute;r L, Khine MS (2025) The shifting landscape of student engagement: A pre-post semester analysis in AI-enhanced classrooms. Computers Education: Artif Intell 8:100395. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2025.100395\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBushell C (2024) Investigating the Integration of AI Chat Models to Enhance Critical Thinking\u0026nbsp;\u003c/span\u003e\u003cspan\u003eSkills through Engineered Prompting (SSRN Scholarly Paper 5017061). Social Science\u0026nbsp;\u003c/span\u003e\u003cspan\u003eResearch Network \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2139/ssrn.5017061\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCacho R (2024) Integrating Generative AI in University Teaching and Learning. A Model for\u0026nbsp;\u003c/span\u003e\u003cspan\u003eBalanced Guidelines Online Learn, 28(3). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.24059/olj.v28i3.4508\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eChan CKY (2023) A comprehensive AI policy education framework for university teaching and learning. Int J Educational Technol High Educ 20(1):38. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-023-00408-3\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eChan CKY, Tsi LHY (2024) Will generative AI replace teachers in higher education? A study of teacher and student perceptions. Stud Educational Evaluation 83:101395. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.stueduc.2024.101395\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCreely E (2024) Exploring the Role of Generative AI in Enhancing Language Learning: Opportunities and Challenges. Int J Changes Educ 1(3):158\u0026ndash;167. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.47852/bonviewIJCE42022495\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCreely E, Henderson M (2025) Artificial intelligence and the digital dis/empowerment of migrant and refugee learners. In E. Tour, E. Creely, P. Waterhouse, \u0026amp; M. Henderson, \u003cem\u003eDigital Empowerment for Refugee and Migrant Learners\u003c/em\u003e (1st ed., pp. 47\u0026ndash;64). Routledge. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9781003422891-5\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCreswell JW (with, Poth CN (2018) Qualitative inquiry \u0026amp; research design: Choosing among five approaches (Fourth edition.). SAGE\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCrompton H, Edmett A, Ichaporia N, Burke D (2024) AI and English language teaching: Affordances and challenges. Br J Edu Technol 55(6):2503\u0026ndash;2529. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/bjet.13460\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDaniel K, Msambwa MM, Wen Z (2025) Can Generative AI Revolutionise Academic\u0026nbsp;\u003c/span\u003e\u003cspan\u003eSkills D in Higher Education? A Systematic Literature Review. European\u0026nbsp;\u003c/span\u003e\u003cspan\u003eJ Educ, \u003cem\u003e60\u003c/em\u003e(1), e70036. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/ejed.70036\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDergaa I, Saad B, Glenn H, Amamou JM, Aissa BB M., Guelmami, N., Fekih-\u003c/span\u003e\u003cspan\u003eRomdhane F, Chamari K (2024) From tools to threats. A reflection on the impact of\u0026nbsp;\u003c/span\u003e\u003cspan\u003eartificial-intelligence chatbots on cognitive health Front Psychol, 15\u0026nbsp;\u003c/span\u003e\u003cspan\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2024.1259845\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eEaton SE (2023) Postplagiarism: Transdisciplinary ethics and integrity in the age of artificial intelligence and neurotechnology. \u003cem\u003eInternational Journal for Educational Integrity\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(1), 23, s40979-023-00144\u0026ndash;1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40979-023-00144-1\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFan L, Kim J, Detrick R, Li N (2025) Developing Students\u0026rsquo; Higher-Order Thinking Skills With Generative AI Insights and Strategies From a Case Study: In A. ElSayary (Ed.), \u003cem\u003eAdvances in Computational Intelligence and Robotics\u003c/em\u003e (pp. 411\u0026ndash;428). IGI Global. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4018/979-8-3693-7332-3.ch024\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFleischer A, Lowman J, Strader K (2025) Understanding Older Adults\u0026rsquo; Intention to Use\u0026nbsp;\u003c/span\u003e\u003cspan\u003eTelehealth A Qualitative Study Using the UTAUT Framework. Journal of Primary Care \u0026amp;\u0026nbsp;\u003c/span\u003e\u003cspan\u003eCommunity Health 16, 21501319251320180. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/21501319251320180\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFreire P (2018) Pedagogy of the oppressed (M. B. Ramos, Trans.; 50th anniversary ed.). Bloomsbury Academic\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFu Q-K, Zou D, Xie H, Cheng G (2024) A review of AWE feedback: Types, learning outcomes, and implications. Comput Assist Lang Learn 37(1\u0026ndash;2):179\u0026ndash;221. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/09588221.2022.2033787\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eGayed JM, Carlon MKJ, Oriola AM, Cross JS (2022) Exploring an AI-based writing Assistant\u0026rsquo;s impact on English language learners. Computers Education: Artif Intell 3:100055. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2022.100055\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eGerlich M (2025) AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies 15(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eArticle 1. https://doi.org/10.3390/soc15010006\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eGlahn K (2024) Using ChatGPT to Teach English for Academic Purposes: Perspectives from the Classroom. J Educ Innov Communication 5(2):63\u0026ndash;75. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.34097/jeicom-5-2-December-23-5\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHibert AI (2019) Systematic Literature Review of Automated Writing Evaluation as a Formative Learning Tool. In M. Scheffel, J. Broisin, V. Pammer-Schindler, A. Ioannou, \u0026amp; J. Schneider (Eds.), \u003cem\u003eTransforming Learning with Meaningful Technologies\u003c/em\u003e (Vol. 11722, pp. 199\u0026ndash;212). Springer International Publishing. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-030-29736-7_15\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHopcan S, T\u0026uuml;rkmen G, Polat E (2024) Exploring the artificial intelligence anxiety and machine learning attitudes of teacher candidates. Education and Information\u0026nbsp;\u003c/span\u003e\u003cspan\u003eTechnologies \u003cem\u003e29\u003c/em\u003e(6), 7281\u0026ndash;7301. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-023-12086-9\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHuber A (2018) Whatever happened to the CALL revolution? Hirao School Manage Rev 8:31\u0026ndash;39\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKinzie S (2024) Language Ideologies and English for Academic Purposes Writing: A Case in Ontario: Larry Vandergrift Best Graduate Paper Award 2022. Can Mod Lang Rev 80(1):1\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3138/cmlr-2023-001\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKlimova B, Pikhart M (2025) Exploring The Effects of Artificial Intelligence on Student and Academic Well-being in Higher Education: A Mini-Review. Front Psychol 16:1498132. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2025.1498132\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKlingbeil A, Gr\u0026uuml;tzner C, Schreck P (2024) Trust and reliance on AI \u0026mdash; An experimental\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003estudy on the extent and costs of overreliance on AI Comput Hum Behav, \u003cem\u003e160\u003c/em\u003e, 108352. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.chb.2024.108352\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKohnke L, Moorhouse BL, Zou D (2023) ChatGPT for Language Teaching and Learning. RELC J 54(2):537\u0026ndash;550. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/00336882231162868\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKouritzin SG, Ellis TF, Ghazani AZ, Nakagawa S (2023) Gigification of English Language Instructor Work in Higher Education: Precarious Employment and Magic Time. TESOL Q 57(4):1518\u0026ndash;1544. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/tesq.3206\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLee BJ, Cook DR (2024) Exploring the potential of AI for pragmatics instruction\u0026nbsp;\u003c/span\u003e\u003cspan\u003eTechnology in Language Teaching \u0026amp; Learning, 6(3), Article 3\u0026nbsp;\u003c/span\u003e\u003cspan\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.29140/tltl.v6n3.1521\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLim JS, Shin D, Lee C, Kim J, Zhang J (2025) The Role of User Empowerment, AI Hallucination, and Privacy Concerns in Continued Use and Premium Subscription Intentions: An Extended Technology Acceptance Model for Generative AI. J Broadcast Electron Media 1\u0026ndash;17. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/08838151.2025.2487679\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLindebaum D, Fleming P (2024) ChatGPT Undermines Human Reflexivity, Scientific\u0026nbsp;\u003c/span\u003e\u003cspan\u003eResponsibility and Responsible Management Research Br J Manag, 35(2), 566\u0026ndash;575. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/1467-8551.12781\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMalik AR, Pratiwi Y, Andajani K, Numertayasa IW, Suharti S, Darwis A, Marzuki (2023) Exploring Artificial Intelligence in Academic Essay. Higher Education Student\u0026rsquo;s\u0026nbsp;\u003c/span\u003e\u003cspan\u003ePerspective Int J Educational Res Open, 5, 100296\u0026nbsp;\u003c/span\u003e\u003cspan\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ijedro.2023.100296\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMarzuki W, Utami R, Diyenti D, Indrawati I (2023) The impact of AI writing tools on the content and organization of students\u0026rsquo; writing: EFL teachers\u0026rsquo; perspective. Cogent Educ 10(2):2236469. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/2331186X.2023.2236469\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMelo CO, Luft MCMS, Rocha RO (2021) Influencing elements of technological\u0026nbsp;\u003c/span\u003e\u003cspan\u003eadoption Case study about management in an educational institution. Contextus \u0026ndash; Revista\u0026nbsp;\u003c/span\u003e\u003cspan\u003e\u003cem\u003eContempor\u0026acirc;nea de Economia e Gest\u0026atilde;o\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e, 124\u0026ndash;145. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.19094/contextus.2021.61445\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMiao F, Holmes W, Huang R, Zhang H et al (2021) AI and education: A guidance for policymakers. UNESCO Publishing\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMorozov E (2013) To Save Everything. The Folly of Technological Solutionism, Click Here\u0026nbsp;\u003c/span\u003e\u003cspan\u003ePublicAffairs\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNoy C (2008) Sampling Knowledge: The Hermeneutics of Snowball Sampling in Qualitative Research. Int J Soc Res Methodol 11(4):327\u0026ndash;344. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/13645570701401305\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNyaaba M, Wright A, Choi GL (2024) \u003cem\u003eGenerative AI and Digital Neocolonialism in Global Education: Towards an Equitable Framework\u003c/em\u003e (Version 3). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2406.02966\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eOgunleye B, Zakariyyah K, Ajao O, Olayinka O, Sharma H (2024) Higher education assessment practice in the era of generative AI tools. J Appl Learn Teach 7(1):1\u0026ndash;11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.37074/jalt.2024.7.1.28\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003ePratama A, Sulistiyo U (2024) A Systematic Review of Artificial Intelligence in Enhancing English Foreign Learners\u0026rsquo; Writing Skill. PPSDP Int J Educ 3(2):170\u0026ndash;181. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.59175/pijed.v3i2.299\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003ePreiksaitis C, Rose C (2023) Opportunities, Challenges, and Future Directions of Generative Artificial Intelligence in Medical Education: Scoping Review. JMIR Med Educ 9:e48785. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2196/48785\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003ePokrivcakova S (2019) Preparing teachers for the application of AI-powered technologies in foreign language education. J Lang Cult Educ 7(3):135\u0026ndash;153. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2478/jolace-2019-0025\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eRaimes A (1991) Out of the Woods: Emerging Traditions in the Teaching of Writing. TESOL\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eQuarterly \u003cem\u003e25\u003c/em\u003e(3), 407\u0026ndash;430. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/3586978\u0026nbsp;\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan\u003eSchretzlmaier P, Hecker A, Ammenwerth E (2022) Suitability of the Unified Theory of\u0026nbsp;\u003c/span\u003e\u003cspan\u003eAcceptance and Use of Technology 2 Model for Predicting mHealth Acceptance Using\u0026nbsp;\u003c/span\u003e\u003cspan\u003eDiabetes as an Example Qualitative Methods Triangulation Study. JMIR Hum Factors, 9(1), e34918. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2196/34918\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eShahid MK, Zia T, Bangfan L, Iqbal Z, Ahmad F (2024) Exploring the relationship of\u0026nbsp;\u003c/span\u003e\u003cspan\u003epsychological factors and adoption readiness in determining university teachers\u0026rsquo; attitude on\u0026nbsp;\u003c/span\u003e\u003cspan\u003eAI-based assessment systems International Journal of Management Education (Elsevier\u0026nbsp;\u003c/span\u003e\u003cspan\u003eScience), \u003cem\u003e22\u003c/em\u003e(2), N.PAG-N.PAG. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ijme.2024.100967\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSolak E (2024) Revolutionizing Language Learning: How ChatGPT and AI are changing the Way We Learn Languages. Int J Technol Educ 7(2):353\u0026ndash;372. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.46328/ijte.732\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSong C, Song Y (2023) Enhancing academic writing skills and motivation. Assessing the\u0026nbsp;\u003c/span\u003e\u003cspan\u003eefficacy of ChatGPT in AI-assisted language learning for EFL students. Frontiers in\u0026nbsp;\u003c/span\u003e\u003cspan\u003ePsychology \u003cem\u003e14\u003c/em\u003e, 1260843. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2023.1260843\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eStolpe K, Hallstr\u0026ouml;m J (2024) Artificial intelligence literacy for technology education\u0026nbsp;\u003c/span\u003e\u003cspan\u003eComputers and Education Open 6, 100159. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeo.2024.100159\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSun T (2023) Potential Use of Artificial Intelligence in ESL Writing Assessment: A Case Study of IELTS Writing Tasks. Ir J Technol Enhanced Learn 7(2):42\u0026ndash;51. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.22554/ijtel.v7i2.137\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSwindell A, Greeley L, Farag A, Verdone B (2024) Against Artificial Education\u0026nbsp;\u003c/span\u003e\u003cspan\u003eTowards an Ethical Framework for Generative Artificial Intelligence (AI) Use in Education\u0026nbsp;\u003c/span\u003e\u003cspan\u003eOnline Learning \u003cem\u003e28\u003c/em\u003e(2). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://eric.ed.gov/?id=EJ1428206\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eTseng W, Warschauer M (2023) AI-writing tools in education: If you can\u0026rsquo;t beat them, join them. J China Computer-Assisted Lang Learn 3(2):258\u0026ndash;262. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1515/jccall-2023-0008\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eVenkatesh V, Morris MG, Davis GB, Davis FD (2003) User Acceptance of Information Technology: Toward a Unified View. MIS Q 27(3):425\u0026ndash;478. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/30036540\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eYan L, Martinez-Maldonado R, Gasevic D (2024) Generative Artificial Intelligence in Learning Analytics: Contextualising Opportunities and Challenges through the Learning Analytics Cycle. \u003cem\u003eProceedings of the 14th Learning Analytics and Knowledge Conference\u003c/em\u003e, 101\u0026ndash;111. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3636555.3636856\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eYatani K, Sramek Z, Yang C-L (2024) \u003cem\u003eAI as Extraherics: Fostering Higher-order Thinking Skills in Human-AI Interaction\u003c/em\u003e (No. arXiv:2409.09218). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2409.09218\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eYuan B, Hu J (2024) \u003cem\u003eGenerative AI as a Tool for Enhancing Reflective Learning in Students\u003c/em\u003e (Version 1). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2412.02603\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWalsh Marr J (2021) The Promise and Precarity of Critical Pedagogy in English for Academic Purposes. BC TEAL J 6(1):132\u0026ndash;141\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWalter Y (2024) Embracing the future of Artificial Intelligence in the classroom: The relevance of AI literacy, prompt engineering, and critical thinking in modern education. Int J Educational Technol High Educ 21(1):15. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-024-00448-3\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWang Z (2022) Computer-assisted EFL writing and evaluations based on artificial intelligence: A case from a college reading and writing course. Libr Hi Tech 40(1):80\u0026ndash;97. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1108/LHT-05-2020-0113\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWang B, Rau P-LP, Yuan T (2023) Measuring user competence in using artificial\u0026nbsp;\u003c/span\u003e\u003cspan\u003eintelligence Validity and reliability of artificial intelligence literacy scale. Behaviour \u0026amp;amp\u0026nbsp;\u003c/span\u003e\u003cspan\u003eInformation Technology \u003cem\u003e42\u003c/em\u003e(9), 1324\u0026ndash;1337. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/0144929X.2022.2072768\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWiboolyasarin W, Wiboolyasarin K, Tiranant P, Jinowat N, Boonyakitanont P (2025) AI-driven chatbots in second language education: A systematic review of their efficacy and pedagogical implications. Ampersand 14:100224. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.amper.2025.100224\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWolters A, Von Straussenburg A, A. F., Riehle DM (2024) AI Literacy in Adult\u0026nbsp;\u003c/span\u003e\u003cspan\u003eEducation\u0026mdash;A L Review. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://hdl.handle.net/10125/107211\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWu Y (2024) Critical Thinking Pedagogics Design in an Era of ChatGPT and Other AI Tools\u0026mdash;Shifting From Teaching What to Teaching Why and How. J Educ Dev 8(1):1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.20849/jed.v8i1.1404\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eYuan Z, Lyu T (2024) Association Between AI Chatbot Self-efficacy and EFL Student\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eClass-related Anxiety: A Control-Value Theory Perspective. Proceedings of the (2024) \u003cem\u003e9th International Conference on Distance Education and Learning\u003c/em\u003e, 371\u0026ndash;375. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3675812.3675822\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eZepeda L, Benavides I, Lopez C, Roman O, Dominguez Y (2023) Guided Design\u0026nbsp;\u003c/span\u003e\u003cspan\u003eMethodology for Activities That Integrate Generative Artificial Intelligence Tools in\u0026nbsp;\u003c/span\u003e\u003cspan\u003eEducation A, Pilot Study Promoting Ethical Use and Fostering Critical Thinking in Graduate Students (2023). \u003cem\u003eICERI\u003c/em\u003e 7283\u0026ndash;7292. 16th annual International Conference of Education, Research and Innovation. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21125/iceri.2023.1811\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eZhang S, Zhao X, Zhou T, Kim JH (2024) Do You Have AI Dependency? The Roles of\u0026nbsp;\u003c/span\u003e\u003cspan\u003eAcademic Self-Efficacy Academic Stress, and Performance Expectations on Problematic\u0026nbsp;\u003c/span\u003e\u003cspan\u003eAI Usage Behavior \u003cem\u003eInternational Journal of Educational Technology in Higher\u0026nbsp;\u003c/em\u003e\u003c/span\u003e\u003cspan\u003eEducation 21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-024-00467-0\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"University of Toronto","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":"Generative AI, English for Academic Purposes, Instructor Perceptions, Critical Pedagogy, PICRAT, UTAUT","lastPublishedDoi":"10.21203/rs.3.rs-7014667/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7014667/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis qualitative exploratory study examines how nine English for Academic Purposes (EAP) instructors at post-secondary institutions in Ontario, Canada, are engaging with the growing presence of Generative Artificial Intelligence (GenAI) in academic writing instruction. Using a qualitative case study approach, researchers conducted semi-structured interviews with EAP instructors to explore their perceptions, practices, and the institutional and ideological factors influencing GenAI adoption. Data were analyzed using reflexive thematic analysis to ensure rigorous interpretation of participant experiences. The analysis was informed by the PICRAT model, the Unified Theory of Acceptance and Use of Technology (UTAUT), and critical pedagogy. The findings revealed a wide range of engagement levels that were influenced by factors such as pedagogical viability, institutional policy, job precarity, and ethical concerns. Instructors primarily used GenAI to enhance existing pedagogical approaches while developing students' critical literacy. Key concerns addressed by participants included risks to intellectual authenticity, diminished critical thinking, and the marginalization of multilingual learners. Institutional barriers such as unclear policies and limited professional development opportunities were found to hinder meaningful integration. While GenAI was often viewed as a useful supplement, participants emphasized the need for pedagogically sound, student-centered implementation that protects human intellectual agency. This research addresses a critical gap in understanding GenAI integration specifically within Canadian EAP contexts. The study offers practical recommendations for developing comprehensive institutional policies, designing targeted professional development programs, and creating pedagogical frameworks that balance technological integration with critical thinking development. Findings are context-specific to Ontario institutions and warrant investigation across diverse educational settings. This study contributes to ongoing discussions about responsible AI integration in higher education and underscores the importance of addressing broader issues of equity and the educational purpose of GenAI in EAP instruction.\u003c/p\u003e","manuscriptTitle":"Power, Pedagogy, and Algorithms: An Exploratory Study of EAP Instructors' Critical Engagement with Generative AI","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-02 03:18:28","doi":"10.21203/rs.3.rs-7014667/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":"06beba81-e54b-4efd-a1f6-05a2c28450d5","owner":[],"postedDate":"July 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":50827983,"name":"Educational Philosophy and Theory"}],"tags":[],"updatedAt":"2025-07-02T03:18:29+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-02 03:18:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7014667","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7014667","identity":"rs-7014667","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}