Writing with AI at the Margins: Student Voice and Authenticity at a Minority-Serving Institution

Writing with AI at the Margins: Student Voice and Authenticity at a Minority-Serving Institution | 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 Writing with AI at the Margins: Student Voice and Authenticity at a Minority-Serving Institution John C. Chick This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8427622/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 The rapid rise in generative artificial intelligence programs like ChatGPT and Claude has prompted important questions around authorship, student voice, and academic integrity. This mixed-methods study surveyed 102 students and 25 faculty at a minority-serving institution to explore perceptions of AI writing tools and their impacts on writing authenticity. The surveys included 16–18 quantitative items on five-point Likert scales and 5 qualitative open-ended questions, gathering information on patterns of AI use, confidence, ethics, and institutional supports. Few students (21%) used AI to complete assignments, but 64% used it for revisions and 43% for clarity support. Faculty and students viewed grammar support as AI's most positive use, though students expressed concerns about originality. A significant majority (76%) used AI without disclosure, constituting an academic integrity violation. Responses about ethics were split between "neither agree nor disagree" (54%) and those acknowledging violations (36%). Multilingual students valued AI assistance with Standard Academic English grammar, viewing it as a positive learning addition. However, students worried these tools could diminish student voice and homogenize written perspectives across cultures. The 47% perception gap between faculty estimates (68%) and student self-reports (21%) of AI use for complete drafting suggests prohibitionist policies may address faculty concerns more than student realities. Findings support distinguishing instrumental support (grammar, mechanics) from expressive support (ideas, voice) in developing AI policies that preserve authentic student perspective while acknowledging AI's legitimate uses. Educational Philosophy and Theory generative AI student voice writing authenticity educational technology minority-serving institutions academic integrity large language models writing assessment Figures Figure 1 Introduction The introduction of generative artificial intelligence tools in education settings at an accelerated pace is a phenomenon that has no historical parallel with respect to writing instruction and assessment. Large language models such as ChatGPT, Claude, Gemini, and other AI writing assistants are now able to produce lengthy academic prose on par with or higher than the quality of a typical undergraduate student's essay on multiple levels, from the overall structure and organization to coherence, argument quality, and stylistic sophistication. The unprecedented speed at which generative AI has entered educational technology and the prospect of its imminent ubiquity in education settings has led to several issues rising to the top of the agenda for educational leaders, teacher-educators, and policymakers in the U.S. and internationally. Educators face the dual challenge of managing the risks while harnessing the potential benefits of generative AI in teaching and learning (Michel-Villarreal et al., 2023). The most critical are those related to academic integrity, student authorship, and the pedagogical value of writing as a central mode of learning in higher education. Recent scholarship has examined both the opportunities and challenges that ChatGPT presents for academic integrity and student learning in higher education (Sullivan et al., 2023). On AI and writing, Bozkurt et al. (2024) present three core tenets for how generative AI is reshaping the landscape of higher education and its implications for teaching and learning, that reverberate through the whole profession: that generative AI may in fact be a potential paradigm shift for HE; that writing tasks and writing-related learning outcomes will be redefined and reconceptualized; and that the infusion of GAI in higher education necessitates an ethical infrastructure to govern education. Central to these three tenets is the core concern of what role student voice can, should, and might play in higher education when writing in particular and language and literacy in general are concerned. This question of student voice acquires a special degree of acuity and urgency in the context of minority-serving institutions (MSIs). Student voice is not just an epistemological and pedagogical matter but also one of educational equity and social justice for first-generation students, multilingual writers, neurodiverse learners, and other students from the historically underrepresented populations. They have to find and affirm their voices in the academic community, not just by speaking up but also by writing up, in the context of the educational system that does not center their forms of discursive expression. The affordances of powerful new AI tools to produce texts on demand in a standardized academic register may represent a serious risk of further alienation from the educational process for students from these populations (Warschauer et al., 2023). The present study is an attempt to examine how generative AI tools might affect student writing in practice at an MSI and to explore pedagogical strategies that could keep student voice, genuine student perspective, and the subjective student response to the writing task as the central focus of academic writing assignments. Our stance is not the prohibitionist one on the one hand or the one of utter tech optimism on the other. Instead, we seek to understand and document how students and faculty members at an MSI attempt to make sense of and find their way in this rapidly changing educational landscape. This study contributes to the literature in four key ways. First, it examines AI usage patterns at a minority-serving institution, providing crucial insights into equity implications often missing from existing research. Second, it identifies a substantial perception gap between faculty assumptions and student practices regarding AI use for complete assignment drafting, challenging prohibitionist approaches. Third, it provides empirical support for distinguishing instrumental support (grammar, mechanics) from expressive support (ideas, voice) in AI-assisted writing. Finally, it centers multilingual student voices in discussions of AI and academic writing, addressing concerns about linguistic identity and homogenization. Research Questions This study was guided by the following research questions: How do students and faculty at a minority-serving institution perceive the impact of generative AI tools on writing authenticity and student voice? What patterns of AI use emerge among graduate students, and how do these patterns relate to concerns about authorship, identity, and academic integrity? What pedagogical strategies can help preserve student voice while acknowledging the reality of AI tools in academic writing contexts? Literature Review Large language models have evolved from GPT-2 through GPT-4 and beyond, transitioning from quantitative to qualitative improvements in capacity relevant to academic writing. As opposed to earlier writing assistance technologies that were limited to grammar and style, contemporary large language models are able to produce fluent, contextually coherent prose within academic genres. Lendvai (2025) presents a scientometric study on ChatGPT for academic writing, with the analysis period from January 2022 to April 2023. The author found a dramatic increase in research output that included ChatGPT, but the articles also contained rampant issues with authenticity. Can AI writing produce academic prose that is intellectually engaging, genuinely reflective of the writer's critical thinking, and authentic in voice (Amirjalili et al., 2024)? Student Voice and Identity in Academic Writing Student voice is a much more complex issue than that of simply writing in the voice of the student. Writing is a sociocultural act (Prior, 2006), a process embedded within a community and context of power relations. From this view, student voice in academic writing then is not only a matter of individuals but also of sociocultural values of whose knowledge and ways of knowing get to be valued in the academy. In its sociocultural dimensions, writing is also a tool for mediating thought and learning (Vygotsky, 1978), and the question of AI’s place in this process is one that may have significant implications for students’ development as thinkers and writers. From another perspective, the development of student voice in academic writing can be framed as the development and projection of identity and authority in the writing process. Elbow (1994) drew a distinction between “real voice” (the personal, authentic self) and “persona” (formalized voice appropriate to a context). The author also noted that both elements must be negotiated for any academic writing endeavor to be successful. Voice in academic writing is not something one simply has but rather a rhetorical construct that is dependent on genre conventions and audience (Matsuda & Tardy, 2007). The concept of student voice extends beyond the realm of writing to include notions of student agency and involvement in decision-making in education (Cook-Sather, 2020). In academic writing, specifically, student voice is both the development of expertise and authority in the discipline and also the assertion of one’s own perspective in scholarly discourse. The issue of finding one’s voice in academic writing is made even more complex for multilingual writers and students from historically marginalized communities by linguistic and cultural tensions inherent in their writing. Canagarajah (2024) proposes that writing instruction must be decolonized. Ivanič (1998) provides an overview of research on how the very act of writing is both a construction of and a reflection on the self. Research on identity development in bilingual writers has demonstrated the process of finding one’s voice in academic writing is both linguistic and disciplinary (Kibler, 2017). A recent bibliometric study confirms that academic interest in how multilingual writers construct identities through writing in English is a robust and growing body of research (Tian & Liu, 2024). Academic Integrity and AI Writing Tools One of the main issues with using AI tools in academic writing is academic integrity. For example, Cotton et al., (2024) argue that if AI writing tools are creating original text, but structurally similar to human text, then traditional plagiarism detection and prevention methods are inadequate for addressing the issue. Evangelista (2025) claims that universities must rethink assessment design, whereas other studies (Qian, 2025; Perkins & Roe, 2023) suggest that universities must develop "academic integrity literacy" to address students' ability to make decisions about ethical issues when writing. Gamage et al. (2024) indicate that there are a range of interpretations of acceptable and unacceptable uses of AI tools among both students and faculty. While some researchers have explored technical solutions such as digital watermarking for AI-generated text (Lancaster, 2023), such approaches face significant implementation challenges. Research suggests that reliably detecting AI-generated text remains technically challenging (Sadasivan et al., 2023), and empirical testing of AI detection tools has revealed significant limitations in their accuracy and reliability (Weber-Wulff et al., 2023). The emergence of generative AI requires reconceptualization of what constitutes authentic assessment in higher education contexts (Kofinas et al., 2025). AI-Assisted Writing Processes AI-assisted writing has been defined as a range of activities where the tool in some way supports the writing process. Nguyen et al. (2024) identified three main types of collaboration between humans and AI writing assistants: (1) AI as brainstorming partner, (2) AI as editor, and (3) AI as co-author. The authors claim that the majority of students used more than one type of collaboration when using AI tools. Khuder (2025) shows that doctoral students use AI feedback as a strategic tool to develop a disciplinary voice, whereas Bedington et al., (2024) state that humans must still maintain some level of control over central ideas in the writing process if authorship is to remain meaningful. These studies on collaboration between humans and AIs indicate that the relationship between AI use and voice is more nuanced than the distinction between "authentic" or "AI-generated" allows. AI tools present a particular set of challenges for multilingual writers. Warschauer et al. (2023) describe how AI can mitigate language barriers, but this can lead to a homogenization of writing toward standardized academic English, thereby erasing linguistic and cultural variation. Liang et al. (2023) state that AI detection tools are less accurate on essays written by non-native English speakers, creating additional issues of equity for multilingual writers. Multilingual scholars often face a conflict between maintaining their linguistic identity and conforming to the dominant anglophone academic culture (Langum & Sullivan, 2020), which can be amplified by AI writing tools. Research on multilingual students' writing processes reveals unique challenges in navigating linguistic expectations while developing authentic voice (Troughton, 2024). Methodology A mixed-methods survey design was chosen for this study, collecting both quantitative and qualitative data through a single comprehensive instrument. Following established mixed-methods research design principles (Creswell & Plano Clark, 2018), this approach allowed us to capture both the breadth of AI usage patterns and the depth of participants' experiences and perceptions. The integration of quantitative and qualitative data provided a more complete understanding of how AI tools affect student voice and authenticity in academic writing. The data were collected from a northeastern minority-serving institution in the United States in fall 2024. One hundred two graduate students and twenty-five faculty served as participants for this study. Students were aged 23 to 45 (average age of 33), of which 66% were women and 34% were men. Faculty participants included both tenure-track and non-tenure-track instructors, with teaching experience ranging from new teachers to those who had been teaching at the institution for over 10 years. Faculty participants came from various backgrounds, including education, business, and engineering. The survey instrument used in this study consisted of 21 items that measured the participants' AI writing tools usage patterns, their confidence using these tools, and their perceptions of the authenticity of their writing. The survey included 16 quantitative items organized into four categories: AI Use in Writing Stages (4 items), Perceived Authenticity (4 items), Ethical Concerns (4 items), and Writing Confidence (4 items). All quantitative responses were measured on five-point Likert scales (1 = Strongly Disagree to 5 = Strongly Agree). Additionally, the survey included 5 qualitative open-ended questions that allowed participants to elaborate on their experiences and perceptions. Student qualitative questions focused on describing specific instances of AI use, perceptions of ownership, effects on voice, aspects influenced by AI, and decision-making processes. Faculty qualitative questions addressed definitions of student voice, assessment of authenticity, observed changes in student writing, ethical concerns, and pedagogical adaptations. The survey design was based on previous studies that have examined students' and teachers' use of AI in education (Bhullar et al., 2024; Dempere et al., 2023). The survey was distributed electronically, and students were offered course credit in exchange for their participation.\ Quantitative data were subjected to descriptive analysis to identify patterns and beliefs. Qualitative data from the open-ended survey questions were subjected to thematic analysis using a six-phase approach suggested by Braun and Clarke (2006): familiarizing oneself with the data, initial coding, searching for themes, reviewing themes, defining and naming themes, and producing the report. Two researchers coded the qualitative responses independently, and inter-rater reliability was 87%. Research on qualitative data saturation suggests that our sample size was adequate for identifying core themes and reaching theoretical saturation (Guest et al., 2006). The Institutional Review Board (IRB) approved the study, and participants provided their informed consent. Findings AI Usage Patterns Survey results provided evidence of differential AI use, highlighting a disconnect between student actual and perceived AI use. Although 79% of students reported never or rarely using AI to completely write an assignment, 64% reported using AI regularly or often for assistance with revision and clarity. Students made a clear distinction between using AI for generation and using AI for revision: they were not having AI write their papers, they were using AI to help them with the papers they were writing. Furthermore, faculty self-report estimates of student AI use were much higher than the self-report estimates from students. Faculty overwhelmingly believed students used AI to completely draft their papers with 68% of faculty estimating that most students do this, while only 21% of students self-reported doing this regularly or often, representing a 47% faculty-student perception gap. Overall, results suggested the need to replace assumptions about student AI use with candid discussions of student actual AI use. Qualitative responses from the open-ended survey questions confirmed and elaborated on three key themes of use: (1) brainstorming/outlining (59% of student responses); (2) checking for grammar/clarity (73%); and (3) translation/linguistic support (41%: mostly multilingual students). A typical comment from a multilingual student was: "I have my ideas in my head but need help on the how to say it in English part. I never have it write for me, I check if it sounds natural in English." Another student commented that they view using AI as "a second pair of eyes to check for errors and awkward phrasing before I hit submit." Students consistently identified "instrumental" areas of support (grammar/mechanics/structure/organization) as distinct from "expressive" dimensions of support (ideas/voice/perspective/argument), viewing the former as much more permissible for assistance and the latter as more problematic, constituting "cheating" or lack of authenticity. This distinction closely mirrors classical writing theory that defines "lower order concerns" (grammar, spelling, mechanics) and "higher order concerns" (content, organization, argumentation). Students felt comfortable using AI for instrumental support but strongly disagreed that it was "ethical" to use AI to assist on the expressive dimensions that students viewed as uniquely theirs as the authors of their papers. As one student wrote in the open-ended responses: "If AI is helping me fix my commas, that's fine. If it's giving me my argument, that's not really my paper anymore." Table 1 provides more detail on use across a range of AI-assisted writing activities. Table 1 Student AI Usage Patterns by Activity Type (N = 102) Activity Never/Rarely Sometimes/Often Complete entire assignments 79% 21% Brainstorm ideas 41% 59% Revise and improve clarity 36% 64% Structure or outline 57% 43% Translation/linguistic support 59% 41% Note. Percentages represent combined responses for 'never' and 'rarely' versus 'regularly' and 'often' on 5-point Likert scales. Translation/linguistic support data reflects multilingual student subsample (n = 44). Voice and Authenticity Concerns Students and faculty also expressed notable concern in their open-ended survey responses that the use of AI tools would cause issues with voice, though their perspectives on the concern were different. Students were concerned that AI could make their writing less personal. One student wrote, "When I use AI to edit, sometimes it makes my writing sound too formal, like I'm not myself anymore. It's technically better, but it doesn't sound like me." Another student echoed: "I have a way of explaining things that's mine. When AI rewrites my sentences, they're clearer maybe, but they don't have my personality." Faculty expressed concern in their qualitative responses about their ability to identify authentic student work. One faculty member wrote, "I can tell when students have used AI heavily because their writing loses its personality. It becomes generic: grammatically perfect but bland. The interesting quirks and personal connections disappear." Another faculty member noted that they observed when the writing that a student produced suddenly improved during a semester: "A student who struggled with clarity all semester suddenly submits a paper that's polished but completely different in style. That inconsistency is a red flag." Multilingual students shared particular concern and tension around voice and authenticity in their survey responses. While they appreciated that AI tools could help them "match" the academic conventions of academic English, they also worried that AI would erase their cultural and linguistic identity. One student reflected, "My first language influences how I think and write in English. Sometimes my sentence structure is a little different because that's how we say things in my language. AI makes my English 'correct' but sometimes it feels like I'm erasing part of who I am." Another multilingual student described AI as being particularly helpful for "translating my thoughts into academic English" but worried that this translation process made all student work homogenous: "Everyone's papers start sounding the same when we all use the same AI to edit." This finding is consonant with Canagarajah's (2024) argument concerning the erasure and homogenization of linguistic identity in AI-facilitated writing, and it raises questions about whose English is privileged in academic spaces. A few multilingual students described in their open-ended responses the practice of code-meshing, or strategically blending home language or dialect features into their academic writing, as an act of resistance against linguistic standardization. However, they noted that AI tools also "corrected" these intentional code choices, meaning that AI may work counter to pedagogies that prioritize and celebrate linguistic diversity. Table 2 Faculty Perceptions vs. Student Self-Reported AI Practices Category Percentage Faculty estimate: Students who regularly use AI for complete assignment drafting 68% Student self-report: Regularly or often use AI for complete assignment drafting 21% Perception gap 47% Note. Faculty participants (n = 25) were asked to estimate percentage of students who regularly or often use AI for complete assignment drafting. Student participants (n = 102) self-reported their actual AI usage patterns. Academic Integrity Perspectives Faculty members had strong opinions about AI use and academic integrity in their quantitative survey responses, with 76% agreeing or strongly agreeing that using AI without disclosure is a violation of academic integrity. However, their qualitative responses revealed that beneath this consensus, many faculty members were unclear about the boundaries of academic integrity. When asked if they could clearly articulate when using AI would and would not be a violation of academic integrity, 52% neither agreed nor disagreed with the statement, indicating a significant amount of ambiguity about appropriate boundaries. Faculty members' written responses about AI use and academic integrity similarly suggested uncertainty about when it is appropriate to use AI: "I know I don't want students to be submitting AI-written papers, but where's the line? Is it okay to use AI for brainstorming? For editing? I honestly don't know." Student responses about AI and academic integrity varied widely, but many also expressed confusion about institutional expectations in their open-ended responses. "Every professor has different rules about AI. In one class I can use it for brainstorming, in another I can't use it at all. It's confusing and stressful," one student wrote. Another student noted that they avoid using AI entirely despite feeling like it would help them because they were not sure what was allowed: "I'm not sure what's allowed, so I just don't use it at all. But I know other students are using it and maybe getting better grades because their writing is more polished." This lack of clarity and consistency around academic integrity creates equity concerns, as students who are more willing to take risks and/or do not mind the possibility of academic sanctions may use AI liberally while students who are more conscientious or concerned about making violations may avoid potentially useful tools. This is further compounded by the fact that some students who can afford to pay for human tutoring services that can provide support and guidance similar to (some) AI writing tools have a major advantage over students who only have access to their own skills and limited institutional resources. As one student put it in the open-ended responses, "Rich students can pay for tutors who basically do what AI does, that being, help with editing, clarity, organization. Why is that okay but AI isn't?" Many faculty and students felt that institutions needed to provide more clarity about how they would and would not allow AI tools to be used in the classroom. Faculty members in particular expressed the need for guidelines that recognize the potential legitimate uses of AI tools while upholding the learning goals of their courses. "We can't pretend AI doesn't exist, but we also can't let it replace the thinking and writing process that's essential to learning. We need clear policies that help students use these tools responsibly," one faculty member wrote. Students also expressed a desire for greater clarity around AI use and academic integrity, and several students even suggested that specific policies for AI use for each individual assignment should be provided in the syllabus for a course, rather than relying on general institutional guidelines. Table 3 summarizes key themes from faculty and student perspectives on academic integrity and AI use. Table 3 Academic Integrity Perspectives on AI Use Statement Faculty Students Using AI without disclosure is academic dishonesty (Agree/Strongly Agree) 76% 36% Unsure where to draw ethical line (Neither Agree nor Disagree) 52% 54% Used AI without disclosure N/A 76% Note. Responses based on 5-point Likert scales. Faculty n = 25, Students n = 102. Discussion Students' focus on the instrumental, rather than the generative, functions of AI tools does not match faculty assumptions about the pervasiveness of AI-generated writing. This divergence is the strongest evidence that at this particular moment, prohibitionist policies are jumping the gun to address a problem in faculty consciousness more than in student reality. Insofar as voice erosion, particularly the voice erosion of multilingual students, is a real problem, however, it is worth figuring out how to take on this challenge from a pedagogical rather than a punitive perspective. In short, instead of creating rules about when and where students can or cannot turn to AI, faculty and administrators might work to ensure that AI use is additive to rather than substitutive for the student's learning and ongoing identity development. To this end, it is encouraging to see the distinction students made between instrumental and expressive support. Policies that might productively grow out of this framing would not be about denying or policing AI at all, but rather about openly acknowledging the ways in which it is helpful for students to have tools that can support them with the mechanical and linguistic challenges of writing while drawing attention to the non-negotiable value of a student's unique perspective, lived experience, and critical engagement with course materials and conversations. This framing of the problem recognizes that the work of writing is multivalent: AI tools are not necessarily a problem if students are using them for the dimensions of the task that do not actually require an original human thought (grammar, clarity, structure, perhaps coherence) while leaving the parts of writing that are still fundamentally a human task (original thinking, personal perspective, authentic voice) as their own. This model dovetails nicely with Bedington et al., (2024) recent advocacy for protecting authorship by maintaining sovereignty over ideas and arguments even while AI is used to assist with expression (Fig. 1). Beyond concerns about academic integrity, the broader implications of AI substitution for human interaction in learning contexts may affect student success, retention, and sense of connection to the educational community (Crawford et al., 2024). Faculty reported a much greater use of AI for drafting entire papers than the students in our survey (68% versus 21%). This perception gap may drive punitive measures, fueling distrust between students and faculty. If instructors assume AI use is widespread and unchecked, they might view any polished paper with suspicion. Conversely, if students sense a "gotcha" attitude, they may mistrust their instructors and reject constructive feedback. Building shared understanding requires open dialogue about AI use and its potential risks and benefits. A spirit of inquiry might also shift policies away from zero-tolerance and toward a more holistic, thoughtful approach that prioritizes pedagogical outcomes and student learning. The survey results show a disparity between the attitudes of students and faculty on allowing students to use AI tools for course work. While many students would use AI if permitted, some faculty oppose its use for assignments. These divergent perspectives reflect a tension between the desire to maintain academic standards and the need to prepare students for a rapidly evolving technological landscape. Instructors have a responsibility to help students become critical and ethical users of technology. Developing students' critical digital literacy requires distinguishing between technical skills and critical consciousness about technology use (Pangrazio & Sefton-Green, 2021). While it is essential to ensure students are developing the skills necessary to succeed academically and professionally, it is also important to recognize that AI tools are likely to become an increasingly important part of their lives. Rather than banishing AI from the classroom, we should instead work to understand how it can best be integrated into our pedagogy and curriculum. The concern about standardizing language is particularly relevant for AI use by multilingual writers and students from minoritized communities. AI tools like GPT-4 may enable students to more easily access standard academic conventions, which could be a positive outcome. However, if every draft of a paper is funneled through an algorithm to standardize language, we risk erasing the language diversity that makes our academic communities so rich. Moreover, by privileging standardized academic English, we risk reinforcing a narrow definition of whose voices matter in academic spaces. Canagarajah (2024) advocates for the decolonization of writing pedagogies by resisting the homogenizing influence of global academic standards. We should be encouraging students to use their diverse linguistic repertoires as tools for meaning-making rather than trying to "correct" them toward standardized English. AI tools that automatically "correct" students' writing to align with standardized academic English may perpetuate linguistic hierarchies and alienate students who speak languages or dialects that differ from the norm. AI use by multilingual students raises equity concerns, but other student populations may also benefit from using AI writing tools. Students with disabilities that impact writing ability may see AI tools as genuinely leveling the playing field, providing support that approximates human accommodations but at a much larger scale and with less stigma. Students from educational backgrounds where writing conventions were never explicitly taught may use AI to learn "correct" academic writing. While these uses of AI may be appropriate for some students and for some instructors, blanket policies that treat all use of AI as misconduct make it difficult for these students to get the support they need. Students and faculty expressed uncertainty about the boundaries of academic integrity in our survey responses, pointing to a need for institutional guidance and clearer communication from instructors. The current policy of vague warnings about "unauthorized use" of AI is inadequate. Instead of blanket bans and vague pronouncements, it is incumbent upon instructors to set clear parameters about when and how students can use AI to complete their coursework. Instructors should also consider co-creating policies with students and including student voice in drafting policies at the institutional level. In addition to clear policies, students also need critical literacy education about AI that includes a focus on developing a critical consciousness about their relationship to these tools. Implications for Practice This research suggests several practical implications for educators and institutions. First, writing assignments should be redesigned to emphasize elements AI cannot effectively replicate, particularly personal reflection, lived experience, and context-specific analysis. Systematic reviews of authentic assessment practices emphasize assignments that require students to apply knowledge to real-world contexts and demonstrate higher-order thinking (Vlachopoulos & Makri, 2024). Assignments asking students to analyze their own experiences, compare theoretical frameworks to their observations, or develop arguments based on primary research are inherently resistant to AI generation while remaining pedagogically valuable. Second, institutions should develop voice-centered assessment rubrics that explicitly evaluate authenticity, personal perspective, and critical engagement alongside traditional criteria like organization and clarity. Research on rubric use in higher education demonstrates their effectiveness in clarifying expectations and providing feedback (Reddy & Andrade, 2010), suggesting that well-designed rubrics could help students navigate appropriate AI use. While some scholars have critiqued traditional rubric use as potentially constraining creativity and reducing complex performance to checklists (Panadero & Jonsson, 2020), voice-centered rubrics that explicitly value authenticity, personal perspective, and intellectual engagement may address these concerns while helping students understand what institutions truly value. Such rubrics make transparent that what institutions value is not merely polished prose but genuine intellectual engagement and authentic student perspective. Third, faculty development should address both pedagogical strategies and ethical frameworks for AI integration. Scholars have called for strategies that promote responsible implementation of AI tools rather than outright prohibition (Halaweh, 2023). Faculty need support in designing AI-resistant assignments, developing rubrics that reward authenticity, and facilitating classroom discussions about ethical AI use. Additionally, faculty themselves should model thoughtful AI engagement, using these tools transparently in their own work while maintaining scholarly integrity. Finally, institutions should adopt transparent, educative policies that acknowledge AI's legitimate uses while protecting learning objectives. Policies should include clear disclosure requirements, distinguish between appropriate and inappropriate uses, and incorporate student voices in policy development. Rather than emphasizing detection and punishment, policies should promote student agency and metacognitive awareness about their writing processes. Limitations The present study was not without limitations. First, the data were taken from a convenience sample of graduate students attending a single MSI and, as a result, the results of this study cannot be generalized to the population as a whole. Next, the study was based on self-reported information about use of AI and the data reported might not reflect true use of such tools due to various forms of bias, including social desirability bias related to academic integrity and cheating. Third, this study was limited to the use of text-based AI tools, and the findings of this study may not be generalizable to use of multimodal AI tools. And, while the study did not assess for disciplinary differences in use of AI tools, the generalizability of the study is limited to the broader academic trends as a whole. Finally, it would be interesting for future research to assess similar variables (e.g., use patterns of AI, perceptions about AI’s educational value) in undergraduate students at multiple institutions and longitudinally to assess if such trends are changing or if differences between groups emerge. Moreover, while the open-ended survey questions provided an important qualitative window into these topics, a more in-depth method of follow-up interviews or focus groups may provide a richer understanding of the nuances of the complex dynamics of AI use and student voice. Conclusion The use of generative AI in academic writing is a rapidly developing and already somewhat controversial topic in higher education. While they can be genuinely helpful to students, multilingual writers and students from historically marginalized groups among them, their use can also trivialize the process of writing and, ultimately, the cognitive and emotional engagement and learning that writing assignments are supposed to support. The results of the present study show that while students are using AI, they do so mostly for instrumental and not generative purposes. This might indicate that there is no need for a complete and total ban on the use of these tools. However, higher education providers need to think through new approaches and specific aspects of their use. Students will use generative AI in the future, and institutions will need to develop policies that are both realistic and that will help ensure that their use does not erode the learning that writing assignments are supposed to support. Focusing on lived experience, point of view, and critical thinking as student writing aspects difficult for an LLM to reproduce will help with the development of new approaches to writing assignments that are rigorous but also authentic. This, in turn, will also help with the development of specific policies and voice-centered rubrics to help students make sense of their own uses of AI in their writing. Declarations Compliance with Ethical Standards Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Conflict of Interest The authors declare that they have no conflict of interest. Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee [name of institution/committee] and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the IRB (approval number: 70). Informed Consent Informed consent was obtained from all individual participants included in the study. References Amirjalili, F., Neysani, M., & Nikbakht, A. (2024). Exploring the boundaries of authorship: a comparative analysis of AI-generated text and human academic writing in English literature. Frontiers in Education . https://doi.org/10.3389/feduc.2024.1347421 Bedington, A., Halcomb, E. F., McKee, H. A., Sargent, T., & Smith, A. (2024). Writing with generative AI and human-machine teaming: Insights and recommendations from faculty and students. Computers and Composition, 71, Article 102833. https://doi.org/10.1016/j.compcom.2024.102833 Bhullar, P. S., Joshi, M., & Chugh, R. (2024). ChatGPT in higher education: A synthesis of the literature and a future research agenda. Educational Information Technology, 29, 21501–21522. https://doi.org/10.1007/s10639-024-12723-x Bozkurt, A., et al. (2024). Speculative futures on ChatGPT and generative artificial intelligence: A collective reflection. Asian Journal of Distance Education, 19 (1), 1–24. https://doi.org/10.5281/zenodo.7636568 Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3 (2), 77–101. https://doi.org/10.1191/1478088706qp063oa Canagarajah, S. (2024). Decolonizing academic writing pedagogies for multilingual students. TESOL Quarterly, 58 (1), 12–40. https://doi.org/10.1002/tesq.3231 Cook-Sather, A. (2020). Student voice across contexts: Fostering student agency in today's schools. Theory into Practice, 59 (2), 182–191. https://doi.org/10.1080/00405841.2019.1705091 Cotton, D. R., Cotton, P. A., & Shipway, J. R. (2024). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International, 61 (2), 228–239. https://doi.org/10.1080/14703297.2023.2190148 Crawford, J., Cowling, M., & Allen, K. A. (2024). When artificial intelligence substitutes humans in higher education: The cost of loneliness, student success, and retention. Studies in Higher Education , 1–15. https://doi.org/10.1080/03075079.2024.2326956 Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd ed.). SAGE Publications. Dempere, J., Modugu, K., Hesham, A., & Ramasamy, L. K. (2023). The impact of ChatGPT on higher education. Frontiers in Education, 8 , 1206936. https://doi.org/10.3389/feduc.2023.1206936 Elbow, P. (1994). Introduction. In P. Elbow (Ed.), Landmark essays on voice and writing (pp. xi-xlvii). Hermagoras Press. Evangelista, E. D. L. (2025). Ensuring academic integrity in the age of ChatGPT: Rethinking exam design, assessment strategies, and ethical AI policies in higher education. Contemporary Educational Technology, 17 (1), ep559. https://doi.org/10.30935/cedtech/15775 Gamage, K. A. A., et al. (2024). Ethical use of ChatGPT in education - Best practices to combat AI-induced plagiarism. Frontiers in Education, 9 , 1465703. https://doi.org/10.3389/feduc.2024.1465703 Guest, G., Bunce, A., & Johnson, L. (2006). How many interviews are enough? An experiment with data saturation and variability. Field Methods, 18 (1), 59–82. Halaweh, M. (2023). ChatGPT in education: Strategies for responsible implementation. Contemporary Educational Technology, 15 (2), ep421. https://doi.org/10.30935/cedtech/13036 Ivanič, R. (1998). Writing and identity: The discoursal construction of identity in academic writing . John Benjamins Publishing. https://doi.org/10.1075/swll.5 Khuder, B. (2025). Enhancing disciplinary voice through feedback-seeking in AI-assisted doctoral writing for publication. Applied Linguistics . https://doi.org/10.1093/applin/amaf022 Kibler, A. K. (2017). Becoming a "Mexican feminist": A minoritized bilingual's development of disciplinary identities through writing. Journal of Second Language Writing, 38 , 26–41. https://doi.org/10.1016/j.jslw.2017.10.011 Kofinas, A., et al. (2025). The impact of generative AI on academic integrity of authentic assessments. British Journal of Educational Technology . https://doi.org/10.1111/bjet.13585 Lancaster, T. (2023). Artificial intelligence, text generation tools and ChatGPT – does digital watermarking offer a solution? International Journal for Educational Integrity, 19 , Article 10. https://doi.org/10.1007/s40979-023-00131-6 Langum, V., & Sullivan, K. P. H. (2020). Academic writing, scholarly identity, voice and the benefits and challenges of multilingualism. Linguistics and Education, 60 , 100883. https://doi.org/10.1016/j.linged.2020.100883 Lendvai, G. F. (2025). ChatGPT in academic writing: A scientometric analysis of literature published between 2022 and 2023. Journal of Scholarly Publishing . https://doi.org/10.1177/15562646251350203 Liang W., Yuksekgonul M., Mao Y., Wu E., Zou J. (2023). GPT detectors are biased against non-native English writers. Patterns (N Y).4(7):100779. https://doi.org/10.1016/j.patter.2023.100779 Matsuda, P. K., & Tardy, C. M. (2007). Voice in academic writing: The rhetorical construction of author identity in blind manuscript review. Journal of Second Language Writing, 16 (2), 75–93. https://doi.org/10.1016/j.esp.2006.10.001 Michel-Villarreal, R., et al. (2023). Challenges and opportunities of generative AI for higher education as explained by ChatGPT. Education Sciences, 13 (9), 856. https://doi.org/10.3390/educsci13090856 Nguyen, A., Hong, Y., Dang, B., & Huang, X. (2024). Human-AI collaboration patterns in AI-assisted academic writing. Studies in Higher Education , 1–18. https://doi.org/10.1080/03075079.2024.2323593 Panadero, E., & Jonsson, A. (2020). A critical review of the arguments against the use of rubrics. Educational Research Review, 30 , 100329. https://doi.org/10.1016/j.edurev.2020.100329 Pangrazio, L., & Sefton-Green, J. (2021). Digital rights, digital citizenship and digital literacy: What's the difference? Journal of New Approaches in Educational Research, 10 (1), 15–27. https://doi.org/10.7821/naer.2021.1.616 Perkins, M., & Roe, J. (2023). Academic integrity considerations of AI large language models in the post-pandemic era: ChatGPT and beyond. Journal of University Teaching and Learning Practice, 20 (2). https://doi.org/10.53761/1.20.02.07 Prior, P. (2006). A sociocultural theory of writing. In C. A. MacArthur, S. Graham, & J. Fitzgerald (Eds.), Handbook of writing research (pp. 54–66). Guilford Press. Qian, Y. (2025). Pedagogical Applications of Generative AI in Higher Education: A Systematic Review of the Field. TechTrends 69 , 1105–1120. https://doi.org/10.1007/s11528-025-01100-1 Reddy, Y. M., & Andrade, H. (2010). A review of rubric use in higher education. Assessment & Evaluation in Higher Education, 35 (4), 435–448. https://doi.org/10.1080/02602930902862859 Sadasivan, V. S., et al. (2023). Can AI-generated text be reliably detected? arXiv preprint . https://arxiv.org/abs/2303.11156 Sullivan, M., Kelly, A., & McLaughlan, P. (2023). ChatGPT in higher education: Considerations for academic integrity and student learning. Journal of Applied Learning & Teaching, 6 (1), 1–10. https://doi.org/10.37074/jalt.2023.6.1.17 Tian, Y., & Liu, D. (2024). A bibliometric study of identity construction in English writing for academic purposes. Frontiers in Psychology, 15 , 1499917. https://doi.org/10.3389/fpsyg.2024.1499917 Troughton, T. E. (2024). Exploring the writing process of multilingual postsecondary students. Discourse and Writing, 34 , 24–42. https://doi.org/10.31468/dwr.1045 Vlachopoulos, D., & Makri, A. (2024). A systematic literature review on authentic assessment in higher education: Best practices for the development of 21st century skills, and policy considerations. Studies in Educational Evaluation, 83, 101425. https://doi.org/10.1016/j.stueduc.2024.101425 Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes . Harvard University Press. Warschauer, M., et al. (2023). The affordances and contradictions of AI-generated text for writers of English as a second or foreign language. Journal of Second Language Writing, 62 , 101071. https://doi.org/10.1016/j.jslw.2023.101071 Weber-Wulff, D., et al. (2023). Testing of detection tools for AI-generated text. International Journal for Educational Integrity, 19 (1), 26. https://doi.org/10.1007/s40979-023-00146-z 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-8427622","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":564293028,"identity":"f6e4f617-9396-4731-ad2f-29fabdc74c73","order_by":0,"name":"John C. Chick","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA60lEQVRIiWNgGAWjYBACxgYehgMMB2wSGJgZDOCiEkRoSUNo4SGkBaLkwOEEIEmkFub2swcP/DhzPs+8nXnzi487bPLtGZgP3ubB57CevISDPTduF8scZiuznHkmzbKHgS3ZGq+WGTwGhxk+3E6cwcxjZszbdtiAh4HHTJoILecgWv6CtfB/I0LLjQMgLcaPGSG2sOHX0pNjcLDnTHKxBDNbGWPvmTQDnsNsxpZz8GgxbD9j/OHHMbs8Cf7Dmz/83GFjwN7e/PDGG3xaGhBsNglGEI8Zj3IQkEdiM39gbMClbhSMglEwCkYyAABl/k2brvk+TgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0009-0003-2237-6279","institution":"University of Bridgeport","correspondingAuthor":true,"prefix":"","firstName":"John","middleName":"C.","lastName":"Chick","suffix":""}],"badges":[],"createdAt":"2025-12-22 18:09:19","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8427622/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8427622/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98937960,"identity":"2c51a501-91a4-4022-ab85-c1e2f518e7b3","added_by":"auto","created_at":"2025-12-24 10:36:38","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43125,"visible":true,"origin":"","legend":"","description":"","filename":"ResearchSquareAIVoiceManuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/7b752173da17a1ac05d744e7.docx"},{"id":98937963,"identity":"8d1023e0-415e-4abd-a9c4-d987688f80d3","added_by":"auto","created_at":"2025-12-24 10:36:38","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs8427622.json","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/82de4c4c228c6913cf0d990a.json"},{"id":98937961,"identity":"6d3e821f-3146-406b-8373-30a59e4c4b92","added_by":"auto","created_at":"2025-12-24 10:36:38","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":78438,"visible":true,"origin":"","legend":"","description":"","filename":"rs84276220enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/040484e8c5fddfae6ddc5f18.xml"},{"id":98937962,"identity":"adb158cb-73d7-4ad6-a352-de333e06a982","added_by":"auto","created_at":"2025-12-24 10:36:38","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":75631,"visible":true,"origin":"","legend":"","description":"","filename":"rs84276220structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/92fd5434388c78023eadc115.xml"},{"id":98937964,"identity":"6797c495-df45-47a6-b3b7-21294024193f","added_by":"auto","created_at":"2025-12-24 10:36:38","extension":"html","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":84113,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/600b9fbb4fb04ae4a3c2f713.html"},{"id":98937959,"identity":"2ff24839-ee2a-4d05-925d-d2372c148dcd","added_by":"auto","created_at":"2025-12-24 10:36:38","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":81411,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eConceptual Framework: Instrumental vs. Expressive Dimensions of AI Writing Support\u003c/em\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/08354bf500d69d97ed3c17cb.png"},{"id":99787956,"identity":"e8329bb7-e77b-4690-8e01-3427b36e67c3","added_by":"auto","created_at":"2026-01-08 12:42:20","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":571782,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8427622/v1/7dc466bc-44cd-4daf-9b50-11aebba9a65a.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eWriting with AI at the Margins: Student Voice and Authenticity at a Minority-Serving Institution\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe introduction of generative artificial intelligence tools in education settings at an accelerated pace is a phenomenon that has no historical parallel with respect to writing instruction and assessment. Large language models such as ChatGPT, Claude, Gemini, and other AI writing assistants are now able to produce lengthy academic prose on par with or higher than the quality of a typical undergraduate student's essay on multiple levels, from the overall structure and organization to coherence, argument quality, and stylistic sophistication. The unprecedented speed at which generative AI has entered educational technology and the prospect of its imminent ubiquity in education settings has led to several issues rising to the top of the agenda for educational leaders, teacher-educators, and policymakers in the U.S. and internationally. Educators face the dual challenge of managing the risks while harnessing the potential benefits of generative AI in teaching and learning (Michel-Villarreal et al., 2023). The most critical are those related to academic integrity, student authorship, and the pedagogical value of writing as a central mode of learning in higher education.\u003c/p\u003e \u003cp\u003eRecent scholarship has examined both the opportunities and challenges that ChatGPT presents for academic integrity and student learning in higher education (Sullivan et al., 2023). On AI and writing, Bozkurt et al. (2024) present three core tenets for how generative AI is reshaping the landscape of higher education and its implications for teaching and learning, that reverberate through the whole profession: that generative AI may in fact be a potential paradigm shift for HE; that writing tasks and writing-related learning outcomes will be redefined and reconceptualized; and that the infusion of GAI in higher education necessitates an ethical infrastructure to govern education.\u003c/p\u003e \u003cp\u003eCentral to these three tenets is the core concern of what role student voice can, should, and might play in higher education when writing in particular and language and literacy in general are concerned. This question of student voice acquires a special degree of acuity and urgency in the context of minority-serving institutions (MSIs). Student voice is not just an epistemological and pedagogical matter but also one of educational equity and social justice for first-generation students, multilingual writers, neurodiverse learners, and other students from the historically underrepresented populations. They have to find and affirm their voices in the academic community, not just by speaking up but also by writing up, in the context of the educational system that does not center their forms of discursive expression. The affordances of powerful new AI tools to produce texts on demand in a standardized academic register may represent a serious risk of further alienation from the educational process for students from these populations (Warschauer et al., 2023).\u003c/p\u003e \u003cp\u003eThe present study is an attempt to examine how generative AI tools might affect student writing in practice at an MSI and to explore pedagogical strategies that could keep student voice, genuine student perspective, and the subjective student response to the writing task as the central focus of academic writing assignments. Our stance is not the prohibitionist one on the one hand or the one of utter tech optimism on the other. Instead, we seek to understand and document how students and faculty members at an MSI attempt to make sense of and find their way in this rapidly changing educational landscape.\u003c/p\u003e \u003cp\u003eThis study contributes to the literature in four key ways. First, it examines AI usage patterns at a minority-serving institution, providing crucial insights into equity implications often missing from existing research. Second, it identifies a substantial perception gap between faculty assumptions and student practices regarding AI use for complete assignment drafting, challenging prohibitionist approaches. Third, it provides empirical support for distinguishing instrumental support (grammar, mechanics) from expressive support (ideas, voice) in AI-assisted writing. Finally, it centers multilingual student voices in discussions of AI and academic writing, addressing concerns about linguistic identity and homogenization.\u003c/p\u003e\n\u003ch3\u003eResearch Questions\u003c/h3\u003e\n\u003cp\u003eThis study was guided by the following research questions:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHow do students and faculty at a minority-serving institution perceive the impact of generative AI tools on writing authenticity and student voice?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWhat patterns of AI use emerge among graduate students, and how do these patterns relate to concerns about authorship, identity, and academic integrity?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWhat pedagogical strategies can help preserve student voice while acknowledging the reality of AI tools in academic writing contexts?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eLiterature Review\u003c/h2\u003e \u003cp\u003eLarge language models have evolved from GPT-2 through GPT-4 and beyond, transitioning from quantitative to qualitative improvements in capacity relevant to academic writing. As opposed to earlier writing assistance technologies that were limited to grammar and style, contemporary large language models are able to produce fluent, contextually coherent prose within academic genres. Lendvai (2025) presents a scientometric study on ChatGPT for academic writing, with the analysis period from January 2022 to April 2023. The author found a dramatic increase in research output that included ChatGPT, but the articles also contained rampant issues with authenticity. Can AI writing produce academic prose that is intellectually engaging, genuinely reflective of the writer's critical thinking, and authentic in voice (Amirjalili et al., 2024)?\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudent Voice and Identity in Academic Writing\u003c/h3\u003e\n\u003cp\u003eStudent voice is a much more complex issue than that of simply writing in the voice of the student. Writing is a sociocultural act (Prior, 2006), a process embedded within a community and context of power relations. From this view, student voice in academic writing then is not only a matter of individuals but also of sociocultural values of whose knowledge and ways of knowing get to be valued in the academy. In its sociocultural dimensions, writing is also a tool for mediating thought and learning (Vygotsky, 1978), and the question of AI\u0026rsquo;s place in this process is one that may have significant implications for students\u0026rsquo; development as thinkers and writers. From another perspective, the development of student voice in academic writing can be framed as the development and projection of identity and authority in the writing process. Elbow (1994) drew a distinction between \u0026ldquo;real voice\u0026rdquo; (the personal, authentic self) and \u0026ldquo;persona\u0026rdquo; (formalized voice appropriate to a context). The author also noted that both elements must be negotiated for any academic writing endeavor to be successful. Voice in academic writing is not something one simply has but rather a rhetorical construct that is dependent on genre conventions and audience (Matsuda \u0026amp; Tardy, 2007).\u003c/p\u003e \u003cp\u003eThe concept of student voice extends beyond the realm of writing to include notions of student agency and involvement in decision-making in education (Cook-Sather, 2020). In academic writing, specifically, student voice is both the development of expertise and authority in the discipline and also the assertion of one\u0026rsquo;s own perspective in scholarly discourse. The issue of finding one\u0026rsquo;s voice in academic writing is made even more complex for multilingual writers and students from historically marginalized communities by linguistic and cultural tensions inherent in their writing. Canagarajah (2024) proposes that writing instruction must be decolonized. Ivanič (1998) provides an overview of research on how the very act of writing is both a construction of and a reflection on the self. Research on identity development in bilingual writers has demonstrated the process of finding one\u0026rsquo;s voice in academic writing is both linguistic and disciplinary (Kibler, 2017). A recent bibliometric study confirms that academic interest in how multilingual writers construct identities through writing in English is a robust and growing body of research (Tian \u0026amp; Liu, 2024).\u003c/p\u003e\n\u003ch3\u003eAcademic Integrity and AI Writing Tools\u003c/h3\u003e\n\u003cp\u003eOne of the main issues with using AI tools in academic writing is academic integrity. For example, Cotton et al., (2024) argue that if AI writing tools are creating original text, but structurally similar to human text, then traditional plagiarism detection and prevention methods are inadequate for addressing the issue. Evangelista (2025) claims that universities must rethink assessment design, whereas other studies (Qian, 2025; Perkins \u0026amp; Roe, 2023) suggest that universities must develop \"academic integrity literacy\" to address students' ability to make decisions about ethical issues when writing. Gamage et al. (2024) indicate that there are a range of interpretations of acceptable and unacceptable uses of AI tools among both students and faculty.\u003c/p\u003e \u003cp\u003eWhile some researchers have explored technical solutions such as digital watermarking for AI-generated text (Lancaster, 2023), such approaches face significant implementation challenges. Research suggests that reliably detecting AI-generated text remains technically challenging (Sadasivan et al., 2023), and empirical testing of AI detection tools has revealed significant limitations in their accuracy and reliability (Weber-Wulff et al., 2023). The emergence of generative AI requires reconceptualization of what constitutes authentic assessment in higher education contexts (Kofinas et al., 2025).\u003c/p\u003e\n\u003ch3\u003eAI-Assisted Writing Processes\u003c/h3\u003e\n\u003cp\u003eAI-assisted writing has been defined as a range of activities where the tool in some way supports the writing process. Nguyen et al. (2024) identified three main types of collaboration between humans and AI writing assistants: (1) AI as brainstorming partner, (2) AI as editor, and (3) AI as co-author. The authors claim that the majority of students used more than one type of collaboration when using AI tools. Khuder (2025) shows that doctoral students use AI feedback as a strategic tool to develop a disciplinary voice, whereas Bedington et al., (2024) state that humans must still maintain some level of control over central ideas in the writing process if authorship is to remain meaningful. These studies on collaboration between humans and AIs indicate that the relationship between AI use and voice is more nuanced than the distinction between \"authentic\" or \"AI-generated\" allows.\u003c/p\u003e \u003cp\u003eAI tools present a particular set of challenges for multilingual writers. Warschauer et al. (2023) describe how AI can mitigate language barriers, but this can lead to a homogenization of writing toward standardized academic English, thereby erasing linguistic and cultural variation. Liang et al. (2023) state that AI detection tools are less accurate on essays written by non-native English speakers, creating additional issues of equity for multilingual writers. Multilingual scholars often face a conflict between maintaining their linguistic identity and conforming to the dominant anglophone academic culture (Langum \u0026amp; Sullivan, 2020), which can be amplified by AI writing tools. Research on multilingual students' writing processes reveals unique challenges in navigating linguistic expectations while developing authentic voice (Troughton, 2024).\u003c/p\u003e"},{"header":"Methodology","content":"\u003cp\u003eA mixed-methods survey design was chosen for this study, collecting both quantitative and qualitative data through a single comprehensive instrument. Following established mixed-methods research design principles (Creswell \u0026amp; Plano Clark, 2018), this approach allowed us to capture both the breadth of AI usage patterns and the depth of participants' experiences and perceptions. The integration of quantitative and qualitative data provided a more complete understanding of how AI tools affect student voice and authenticity in academic writing.\u003c/p\u003e \u003cp\u003eThe data were collected from a northeastern minority-serving institution in the United States in fall 2024. One hundred two graduate students and twenty-five faculty served as participants for this study. Students were aged 23 to 45 (average age of 33), of which 66% were women and 34% were men. Faculty participants included both tenure-track and non-tenure-track instructors, with teaching experience ranging from new teachers to those who had been teaching at the institution for over 10 years. Faculty participants came from various backgrounds, including education, business, and engineering.\u003c/p\u003e \u003cp\u003eThe survey instrument used in this study consisted of 21 items that measured the participants' AI writing tools usage patterns, their confidence using these tools, and their perceptions of the authenticity of their writing. The survey included 16 quantitative items organized into four categories: AI Use in Writing Stages (4 items), Perceived Authenticity (4 items), Ethical Concerns (4 items), and Writing Confidence (4 items). All quantitative responses were measured on five-point Likert scales (1 = Strongly Disagree to 5 = Strongly Agree). Additionally, the survey included 5 qualitative open-ended questions that allowed participants to elaborate on their experiences and perceptions. Student qualitative questions focused on describing specific instances of AI use, perceptions of ownership, effects on voice, aspects influenced by AI, and decision-making processes. Faculty qualitative questions addressed definitions of student voice, assessment of authenticity, observed changes in student writing, ethical concerns, and pedagogical adaptations. The survey design was based on previous studies that have examined students' and teachers' use of AI in education (Bhullar et al., 2024; Dempere et al., 2023). The survey was distributed electronically, and students were offered course credit in exchange for their participation.\\\u003c/p\u003e \u003cp\u003eQuantitative data were subjected to descriptive analysis to identify patterns and beliefs. Qualitative data from the open-ended survey questions were subjected to thematic analysis using a six-phase approach suggested by Braun and Clarke (2006): familiarizing oneself with the data, initial coding, searching for themes, reviewing themes, defining and naming themes, and producing the report. Two researchers coded the qualitative responses independently, and inter-rater reliability was 87%. Research on qualitative data saturation suggests that our sample size was adequate for identifying core themes and reaching theoretical saturation (Guest et al., 2006). The Institutional Review Board (IRB) approved the study, and participants provided their informed consent.\u003c/p\u003e"},{"header":"Findings","content":"\u003ch2\u003eAI Usage Patterns\u003c/h2\u003e\u003cp\u003eSurvey results provided evidence of differential AI use, highlighting a disconnect between student actual and perceived AI use. Although 79% of students reported never or rarely using AI to completely write an assignment, 64% reported using AI regularly or often for assistance with revision and clarity. Students made a clear distinction between using AI for generation and using AI for revision: they were not having AI write their papers, they were using AI to help them with the papers they were writing. Furthermore, faculty self-report estimates of student AI use were much higher than the self-report estimates from students. Faculty overwhelmingly believed students used AI to completely draft their papers with 68% of faculty estimating that most students do this, while only 21% of students self-reported doing this regularly or often, representing a 47% faculty-student perception gap. Overall, results suggested the need to replace assumptions about student AI use with candid discussions of student actual AI use.\u003c/p\u003e\u003cp\u003eQualitative responses from the open-ended survey questions confirmed and elaborated on three key themes of use: (1) brainstorming/outlining (59% of student responses); (2) checking for grammar/clarity (73%); and (3) translation/linguistic support (41%: mostly multilingual students). A typical comment from a multilingual student was: \"I have my ideas in my head but need help on the how to say it in English part. I never have it write for me, I check if it sounds natural in English.\" Another student commented that they view using AI as \"a second pair of eyes to check for errors and awkward phrasing before I hit submit.\" Students consistently identified \"instrumental\" areas of support (grammar/mechanics/structure/organization) as distinct from \"expressive\" dimensions of support (ideas/voice/perspective/argument), viewing the former as much more permissible for assistance and the latter as more problematic, constituting \"cheating\" or lack of authenticity. This distinction closely mirrors classical writing theory that defines \"lower order concerns\" (grammar, spelling, mechanics) and \"higher order concerns\" (content, organization, argumentation). Students felt comfortable using AI for instrumental support but strongly disagreed that it was \"ethical\" to use AI to assist on the expressive dimensions that students viewed as uniquely theirs as the authors of their papers. As one student wrote in the open-ended responses: \"If AI is helping me fix my commas, that's fine. If it's giving me my argument, that's not really my paper anymore.\" Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e provides more detail on use across a range of AI-assisted writing activities.\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eStudent AI Usage Patterns by Activity Type (N = 102)\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eActivity\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNever/Rarely\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSometimes/Often\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eComplete entire assignments\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e79%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e21%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBrainstorm ideas\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e41%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e59%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRevise and improve clarity\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e36%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e64%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStructure or outline\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e57%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e43%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTranslation/linguistic support\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e59%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e41%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"3\"\u003e\u003cem\u003eNote. Percentages represent combined responses for 'never' and 'rarely' versus 'regularly' and 'often' on 5-point Likert scales. Translation/linguistic support data reflects multilingual student subsample (n = 44).\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003ch3\u003eVoice and Authenticity Concerns\u003c/h3\u003e\u003cp\u003eStudents and faculty also expressed notable concern in their open-ended survey responses that the use of AI tools would cause issues with voice, though their perspectives on the concern were different. Students were concerned that AI could make their writing less personal. One student wrote, \"When I use AI to edit, sometimes it makes my writing sound too formal, like I'm not myself anymore. It's technically better, but it doesn't sound like me.\" Another student echoed: \"I have a way of explaining things that's mine. When AI rewrites my sentences, they're clearer maybe, but they don't have my personality.\"\u003c/p\u003e\u003cp\u003eFaculty expressed concern in their qualitative responses about their ability to identify authentic student work. One faculty member wrote, \"I can tell when students have used AI heavily because their writing loses its personality. It becomes generic: grammatically perfect but bland. The interesting quirks and personal connections disappear.\" Another faculty member noted that they observed when the writing that a student produced suddenly improved during a semester: \"A student who struggled with clarity all semester suddenly submits a paper that's polished but completely different in style. That inconsistency is a red flag.\"\u003c/p\u003e\u003cp\u003eMultilingual students shared particular concern and tension around voice and authenticity in their survey responses. While they appreciated that AI tools could help them \"match\" the academic conventions of academic English, they also worried that AI would erase their cultural and linguistic identity. One student reflected, \"My first language influences how I think and write in English. Sometimes my sentence structure is a little different because that's how we say things in my language. AI makes my English 'correct' but sometimes it feels like I'm erasing part of who I am.\" Another multilingual student described AI as being particularly helpful for \"translating my thoughts into academic English\" but worried that this translation process made all student work homogenous: \"Everyone's papers start sounding the same when we all use the same AI to edit.\"\u003c/p\u003e\u003cp\u003eThis finding is consonant with Canagarajah's (2024) argument concerning the erasure and homogenization of linguistic identity in AI-facilitated writing, and it raises questions about whose English is privileged in academic spaces. A few multilingual students described in their open-ended responses the practice of code-meshing, or strategically blending home language or dialect features into their academic writing, as an act of resistance against linguistic standardization. However, they noted that AI tools also \"corrected\" these intentional code choices, meaning that AI may work counter to pedagogies that prioritize and celebrate linguistic diversity.\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eFaculty Perceptions vs. Student Self-Reported AI Practices\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePercentage\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFaculty estimate: Students who regularly use AI for complete assignment drafting\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e68%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudent self-report: Regularly or often use AI for complete assignment drafting\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePerception gap\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e47%\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"2\"\u003e\u003cem\u003eNote. Faculty participants (n = 25) were asked to estimate percentage of students who regularly or often use AI for complete assignment drafting. Student participants (n = 102) self-reported their actual AI usage patterns.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003ch2\u003eAcademic Integrity Perspectives\u003c/h2\u003e\u003cp\u003eFaculty members had strong opinions about AI use and academic integrity in their quantitative survey responses, with 76% agreeing or strongly agreeing that using AI without disclosure is a violation of academic integrity. However, their qualitative responses revealed that beneath this consensus, many faculty members were unclear about the boundaries of academic integrity. When asked if they could clearly articulate when using AI would and would not be a violation of academic integrity, 52% neither agreed nor disagreed with the statement, indicating a significant amount of ambiguity about appropriate boundaries. Faculty members' written responses about AI use and academic integrity similarly suggested uncertainty about when it is appropriate to use AI: \"I know I don't want students to be submitting AI-written papers, but where's the line? Is it okay to use AI for brainstorming? For editing? I honestly don't know.\"\u003c/p\u003e\u003cp\u003eStudent responses about AI and academic integrity varied widely, but many also expressed confusion about institutional expectations in their open-ended responses. \"Every professor has different rules about AI. In one class I can use it for brainstorming, in another I can't use it at all. It's confusing and stressful,\" one student wrote. Another student noted that they avoid using AI entirely despite feeling like it would help them because they were not sure what was allowed: \"I'm not sure what's allowed, so I just don't use it at all. But I know other students are using it and maybe getting better grades because their writing is more polished.\" This lack of clarity and consistency around academic integrity creates equity concerns, as students who are more willing to take risks and/or do not mind the possibility of academic sanctions may use AI liberally while students who are more conscientious or concerned about making violations may avoid potentially useful tools.\u003c/p\u003e\u003cp\u003eThis is further compounded by the fact that some students who can afford to pay for human tutoring services that can provide support and guidance similar to (some) AI writing tools have a major advantage over students who only have access to their own skills and limited institutional resources. As one student put it in the open-ended responses, \"Rich students can pay for tutors who basically do what AI does, that being, help with editing, clarity, organization. Why is that okay but AI isn't?\" Many faculty and students felt that institutions needed to provide more clarity about how they would and would not allow AI tools to be used in the classroom. Faculty members in particular expressed the need for guidelines that recognize the potential legitimate uses of AI tools while upholding the learning goals of their courses. \"We can't pretend AI doesn't exist, but we also can't let it replace the thinking and writing process that's essential to learning. We need clear policies that help students use these tools responsibly,\" one faculty member wrote. Students also expressed a desire for greater clarity around AI use and academic integrity, and several students even suggested that specific policies for AI use for each individual assignment should be provided in the syllabus for a course, rather than relying on general institutional guidelines. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e summarizes key themes from faculty and student perspectives on academic integrity and AI use.\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eAcademic Integrity Perspectives on AI Use\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStatement\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFaculty\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStudents\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsing AI without disclosure is academic dishonesty (Agree/Strongly Agree)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e36%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnsure where to draw ethical line (Neither Agree nor Disagree)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e52%\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e54%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsed AI without disclosure\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e76%\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"3\"\u003e\u003cem\u003eNote. Responses based on 5-point Likert scales. Faculty n = 25, Students n = 102.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eStudents' focus on the instrumental, rather than the generative, functions of AI tools does not match faculty assumptions about the pervasiveness of AI-generated writing. This divergence is the strongest evidence that at this particular moment, prohibitionist policies are jumping the gun to address a problem in faculty consciousness more than in student reality. Insofar as voice erosion, particularly the voice erosion of multilingual students, is a real problem, however, it is worth figuring out how to take on this challenge from a pedagogical rather than a punitive perspective. In short, instead of creating rules about when and where students can or cannot turn to AI, faculty and administrators might work to ensure that AI use is additive to rather than substitutive for the student's learning and ongoing identity development. To this end, it is encouraging to see the distinction students made between instrumental and expressive support.\u003c/p\u003e \u003cp\u003ePolicies that might productively grow out of this framing would not be about denying or policing AI at all, but rather about openly acknowledging the ways in which it is helpful for students to have tools that can support them with the mechanical and linguistic challenges of writing while drawing attention to the non-negotiable value of a student's unique perspective, lived experience, and critical engagement with course materials and conversations. This framing of the problem recognizes that the work of writing is multivalent: AI tools are not necessarily a problem if students are using them for the dimensions of the task that do not actually require an original human thought (grammar, clarity, structure, perhaps coherence) while leaving the parts of writing that are still fundamentally a human task (original thinking, personal perspective, authentic voice) as their own. This model dovetails nicely with Bedington et al., (2024) recent advocacy for protecting authorship by maintaining sovereignty over ideas and arguments even while AI is used to assist with expression (Fig.\u0026nbsp;1).\u003c/p\u003e \u003cp\u003eBeyond concerns about academic integrity, the broader implications of AI substitution for human interaction in learning contexts may affect student success, retention, and sense of connection to the educational community (Crawford et al., 2024). Faculty reported a much greater use of AI for drafting entire papers than the students in our survey (68% versus 21%). This perception gap may drive punitive measures, fueling distrust between students and faculty. If instructors assume AI use is widespread and unchecked, they might view any polished paper with suspicion. Conversely, if students sense a \"gotcha\" attitude, they may mistrust their instructors and reject constructive feedback. Building shared understanding requires open dialogue about AI use and its potential risks and benefits. A spirit of inquiry might also shift policies away from zero-tolerance and toward a more holistic, thoughtful approach that prioritizes pedagogical outcomes and student learning.\u003c/p\u003e \u003cp\u003eThe survey results show a disparity between the attitudes of students and faculty on allowing students to use AI tools for course work. While many students would use AI if permitted, some faculty oppose its use for assignments. These divergent perspectives reflect a tension between the desire to maintain academic standards and the need to prepare students for a rapidly evolving technological landscape. Instructors have a responsibility to help students become critical and ethical users of technology. Developing students' critical digital literacy requires distinguishing between technical skills and critical consciousness about technology use (Pangrazio \u0026amp; Sefton-Green, 2021). While it is essential to ensure students are developing the skills necessary to succeed academically and professionally, it is also important to recognize that AI tools are likely to become an increasingly important part of their lives. Rather than banishing AI from the classroom, we should instead work to understand how it can best be integrated into our pedagogy and curriculum.\u003c/p\u003e \u003cp\u003eThe concern about standardizing language is particularly relevant for AI use by multilingual writers and students from minoritized communities. AI tools like GPT-4 may enable students to more easily access standard academic conventions, which could be a positive outcome. However, if every draft of a paper is funneled through an algorithm to standardize language, we risk erasing the language diversity that makes our academic communities so rich. Moreover, by privileging standardized academic English, we risk reinforcing a narrow definition of whose voices matter in academic spaces. Canagarajah (2024) advocates for the decolonization of writing pedagogies by resisting the homogenizing influence of global academic standards. We should be encouraging students to use their diverse linguistic repertoires as tools for meaning-making rather than trying to \"correct\" them toward standardized English. AI tools that automatically \"correct\" students' writing to align with standardized academic English may perpetuate linguistic hierarchies and alienate students who speak languages or dialects that differ from the norm.\u003c/p\u003e \u003cp\u003eAI use by multilingual students raises equity concerns, but other student populations may also benefit from using AI writing tools. Students with disabilities that impact writing ability may see AI tools as genuinely leveling the playing field, providing support that approximates human accommodations but at a much larger scale and with less stigma. Students from educational backgrounds where writing conventions were never explicitly taught may use AI to learn \"correct\" academic writing. While these uses of AI may be appropriate for some students and for some instructors, blanket policies that treat all use of AI as misconduct make it difficult for these students to get the support they need.\u003c/p\u003e \u003cp\u003eStudents and faculty expressed uncertainty about the boundaries of academic integrity in our survey responses, pointing to a need for institutional guidance and clearer communication from instructors. The current policy of vague warnings about \"unauthorized use\" of AI is inadequate. Instead of blanket bans and vague pronouncements, it is incumbent upon instructors to set clear parameters about when and how students can use AI to complete their coursework. Instructors should also consider co-creating policies with students and including student voice in drafting policies at the institutional level. In addition to clear policies, students also need critical literacy education about AI that includes a focus on developing a critical consciousness about their relationship to these tools.\u003c/p\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eImplications for Practice\u003c/h2\u003e \u003cp\u003eThis research suggests several practical implications for educators and institutions. First, writing assignments should be redesigned to emphasize elements AI cannot effectively replicate, particularly personal reflection, lived experience, and context-specific analysis. Systematic reviews of authentic assessment practices emphasize assignments that require students to apply knowledge to real-world contexts and demonstrate higher-order thinking (Vlachopoulos \u0026amp; Makri, 2024). Assignments asking students to analyze their own experiences, compare theoretical frameworks to their observations, or develop arguments based on primary research are inherently resistant to AI generation while remaining pedagogically valuable.\u003c/p\u003e \u003cp\u003eSecond, institutions should develop voice-centered assessment rubrics that explicitly evaluate authenticity, personal perspective, and critical engagement alongside traditional criteria like organization and clarity. Research on rubric use in higher education demonstrates their effectiveness in clarifying expectations and providing feedback (Reddy \u0026amp; Andrade, 2010), suggesting that well-designed rubrics could help students navigate appropriate AI use. While some scholars have critiqued traditional rubric use as potentially constraining creativity and reducing complex performance to checklists (Panadero \u0026amp; Jonsson, 2020), voice-centered rubrics that explicitly value authenticity, personal perspective, and intellectual engagement may address these concerns while helping students understand what institutions truly value. Such rubrics make transparent that what institutions value is not merely polished prose but genuine intellectual engagement and authentic student perspective.\u003c/p\u003e \u003cp\u003eThird, faculty development should address both pedagogical strategies and ethical frameworks for AI integration. Scholars have called for strategies that promote responsible implementation of AI tools rather than outright prohibition (Halaweh, 2023). Faculty need support in designing AI-resistant assignments, developing rubrics that reward authenticity, and facilitating classroom discussions about ethical AI use. Additionally, faculty themselves should model thoughtful AI engagement, using these tools transparently in their own work while maintaining scholarly integrity.\u003c/p\u003e \u003cp\u003eFinally, institutions should adopt transparent, educative policies that acknowledge AI's legitimate uses while protecting learning objectives. Policies should include clear disclosure requirements, distinguish between appropriate and inappropriate uses, and incorporate student voices in policy development. Rather than emphasizing detection and punishment, policies should promote student agency and metacognitive awareness about their writing processes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eThe present study was not without limitations. First, the data were taken from a convenience sample of graduate students attending a single MSI and, as a result, the results of this study cannot be generalized to the population as a whole.\u003c/p\u003e \u003cp\u003eNext, the study was based on self-reported information about use of AI and the data reported might not reflect true use of such tools due to various forms of bias, including social desirability bias related to academic integrity and cheating. Third, this study was limited to the use of text-based AI tools, and the findings of this study may not be generalizable to use of multimodal AI tools. And, while the study did not assess for disciplinary differences in use of AI tools, the generalizability of the study is limited to the broader academic trends as a whole.\u003c/p\u003e \u003cp\u003eFinally, it would be interesting for future research to assess similar variables (e.g., use patterns of AI, perceptions about AI\u0026rsquo;s educational value) in undergraduate students at multiple institutions and longitudinally to assess if such trends are changing or if differences between groups emerge. Moreover, while the open-ended survey questions provided an important qualitative window into these topics, a more in-depth method of follow-up interviews or focus groups may provide a richer understanding of the nuances of the complex dynamics of AI use and student voice.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe use of generative AI in academic writing is a rapidly developing and already somewhat controversial topic in higher education. While they can be genuinely helpful to students, multilingual writers and students from historically marginalized groups among them, their use can also trivialize the process of writing and, ultimately, the cognitive and emotional engagement and learning that writing assignments are supposed to support. The results of the present study show that while students are using AI, they do so mostly for instrumental and not generative purposes. This might indicate that there is no need for a complete and total ban on the use of these tools. However, higher education providers need to think through new approaches and specific aspects of their use. Students will use generative AI in the future, and institutions will need to develop policies that are both realistic and that will help ensure that their use does not erode the learning that writing assignments are supposed to support. Focusing on lived experience, point of view, and critical thinking as student writing aspects difficult for an LLM to reproduce will help with the development of new approaches to writing assignments that are rigorous but also authentic. This, in turn, will also help with the development of specific policies and voice-centered rubrics to help students make sense of their own uses of AI in their writing.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCompliance with Ethical Standards\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFunding\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConflict of Interest\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eEthical Approval\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee [name of institution/committee] and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the IRB (approval number: 70).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eInformed Consent\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent was obtained from all individual participants included in the study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAmirjalili, F., Neysani, M., \u0026amp; Nikbakht, A. (2024). Exploring the boundaries of authorship: a comparative analysis of AI-generated text and human academic writing in English literature. \u003cem\u003eFrontiers in Education\u003c/em\u003e. https://doi.org/10.3389/feduc.2024.1347421\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBedington, A., Halcomb, E. F., McKee, H. A., Sargent, T., \u0026amp; Smith, A. (2024). Writing with generative AI and human-machine teaming: Insights and recommendations from faculty and students. Computers and Composition, 71, Article 102833. https://doi.org/10.1016/j.compcom.2024.102833\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhullar, P. S., Joshi, M., \u0026amp; Chugh, R. (2024). ChatGPT in higher education: A synthesis of the literature and a future research agenda. Educational Information Technology, 29, 21501\u0026ndash;21522. https://doi.org/10.1007/s10639-024-12723-x\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBozkurt, A., et al. (2024). Speculative futures on ChatGPT and generative artificial intelligence: A collective reflection. \u003cem\u003eAsian Journal of Distance Education, 19\u003c/em\u003e(1), 1\u0026ndash;24. https://doi.org/10.5281/zenodo.7636568\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraun, V., \u0026amp; Clarke, V. (2006). Using thematic analysis in psychology. \u003cem\u003eQualitative Research in Psychology, 3\u003c/em\u003e(2), 77\u0026ndash;101. https://doi.org/10.1191/1478088706qp063oa\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCanagarajah, S. (2024). Decolonizing academic writing pedagogies for multilingual students. \u003cem\u003eTESOL Quarterly, 58\u003c/em\u003e(1), 12\u0026ndash;40. https://doi.org/10.1002/tesq.3231\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCook-Sather, A. (2020). Student voice across contexts: Fostering student agency in today's schools. \u003cem\u003eTheory into Practice, 59\u003c/em\u003e(2), 182\u0026ndash;191. https://doi.org/10.1080/00405841.2019.1705091\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCotton, D. R., Cotton, P. A., \u0026amp; Shipway, J. R. (2024). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. \u003cem\u003eInnovations in Education and Teaching International, 61\u003c/em\u003e(2), 228\u0026ndash;239. https://doi.org/10.1080/14703297.2023.2190148\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCrawford, J., Cowling, M., \u0026amp; Allen, K. A. (2024). When artificial intelligence substitutes humans in higher education: The cost of loneliness, student success, and retention. \u003cem\u003eStudies in Higher Education\u003c/em\u003e, 1\u0026ndash;15. https://doi.org/10.1080/03075079.2024.2326956\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCreswell, J. W., \u0026amp; Plano Clark, V. L. (2018). \u003cem\u003eDesigning and conducting mixed methods research\u003c/em\u003e (3rd ed.). SAGE Publications.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDempere, J., Modugu, K., Hesham, A., \u0026amp; Ramasamy, L. K. (2023). The impact of ChatGPT on higher education. \u003cem\u003eFrontiers in Education, 8\u003c/em\u003e, 1206936. https://doi.org/10.3389/feduc.2023.1206936\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElbow, P. (1994). Introduction. In P. Elbow (Ed.), \u003cem\u003eLandmark essays on voice and writing\u003c/em\u003e (pp. xi-xlvii). Hermagoras Press.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEvangelista, E. D. L. (2025). Ensuring academic integrity in the age of ChatGPT: Rethinking exam design, assessment strategies, and ethical AI policies in higher education. \u003cem\u003eContemporary Educational Technology, 17\u003c/em\u003e(1), ep559. \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ehttps://doi.org/10.30935/cedtech/15775\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGamage, K. A. A., et al. (2024). Ethical use of ChatGPT in education - Best practices to combat AI-induced plagiarism. \u003cem\u003eFrontiers in Education, 9\u003c/em\u003e, 1465703. https://doi.org/10.3389/feduc.2024.1465703\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuest, G., Bunce, A., \u0026amp; Johnson, L. (2006). How many interviews are enough? An experiment with data saturation and variability. \u003cem\u003eField Methods, 18\u003c/em\u003e(1), 59\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHalaweh, M. (2023). ChatGPT in education: Strategies for responsible implementation. \u003cem\u003eContemporary Educational Technology, 15\u003c/em\u003e(2), ep421. https://doi.org/10.30935/cedtech/13036\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIvanič, R. (1998). \u003cem\u003eWriting and identity: The discoursal construction of identity in academic writing\u003c/em\u003e. John Benjamins Publishing. https://doi.org/10.1075/swll.5\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhuder, B. (2025). Enhancing disciplinary voice through feedback-seeking in AI-assisted doctoral writing for publication. \u003cem\u003eApplied Linguistics\u003c/em\u003e. https://doi.org/10.1093/applin/amaf022\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKibler, A. K. (2017). Becoming a \"Mexican feminist\": A minoritized bilingual's development of disciplinary identities through writing. \u003cem\u003eJournal of Second Language Writing, 38\u003c/em\u003e, 26\u0026ndash;41. https://doi.org/10.1016/j.jslw.2017.10.011\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKofinas, A., et al. (2025). The impact of generative AI on academic integrity of authentic assessments. \u003cem\u003eBritish Journal of Educational Technology\u003c/em\u003e. https://doi.org/10.1111/bjet.13585\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLancaster, T. (2023). Artificial intelligence, text generation tools and ChatGPT \u0026ndash; does digital watermarking offer a solution? \u003cem\u003eInternational Journal for Educational Integrity, 19\u003c/em\u003e, Article 10. https://doi.org/10.1007/s40979-023-00131-6\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangum, V., \u0026amp; Sullivan, K. P. H. (2020). Academic writing, scholarly identity, voice and the benefits and challenges of multilingualism. \u003cem\u003eLinguistics and Education, 60\u003c/em\u003e, 100883. https://doi.org/10.1016/j.linged.2020.100883\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLendvai, G. F. (2025). ChatGPT in academic writing: A scientometric analysis of literature published between 2022 and 2023. \u003cem\u003eJournal of Scholarly Publishing\u003c/em\u003e. https://doi.org/10.1177/15562646251350203\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiang W., Yuksekgonul M., Mao Y., Wu E., Zou J. (2023). GPT detectors are biased against non-native English writers. \u003cem\u003ePatterns\u003c/em\u003e (N Y).4(7):100779. https://doi.org/10.1016/j.patter.2023.100779\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatsuda, P. K., \u0026amp; Tardy, C. M. (2007). Voice in academic writing: The rhetorical construction of author identity in blind manuscript review. \u003cem\u003eJournal of Second Language Writing, 16\u003c/em\u003e(2), 75\u0026ndash;93. https://doi.org/10.1016/j.esp.2006.10.001\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMichel-Villarreal, R., et al. (2023). Challenges and opportunities of generative AI for higher education as explained by ChatGPT. \u003cem\u003eEducation Sciences, 13\u003c/em\u003e(9), 856. https://doi.org/10.3390/educsci13090856\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNguyen, A., Hong, Y., Dang, B., \u0026amp; Huang, X. (2024). Human-AI collaboration patterns in AI-assisted academic writing. \u003cem\u003eStudies in Higher Education\u003c/em\u003e, 1\u0026ndash;18. https://doi.org/10.1080/03075079.2024.2323593\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePanadero, E., \u0026amp; Jonsson, A. (2020). A critical review of the arguments against the use of rubrics. \u003cem\u003eEducational Research Review, 30\u003c/em\u003e, 100329. https://doi.org/10.1016/j.edurev.2020.100329\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePangrazio, L., \u0026amp; Sefton-Green, J. (2021). Digital rights, digital citizenship and digital literacy: What's the difference? \u003cem\u003eJournal of New Approaches in Educational Research, 10\u003c/em\u003e(1), 15\u0026ndash;27. https://doi.org/10.7821/naer.2021.1.616\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerkins, M., \u0026amp; Roe, J. (2023). Academic integrity considerations of AI large language models in the post-pandemic era: ChatGPT and beyond. \u003cem\u003eJournal of University Teaching and Learning Practice, 20\u003c/em\u003e(2). https://doi.org/10.53761/1.20.02.07\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrior, P. (2006). A sociocultural theory of writing. In C. A. MacArthur, S. Graham, \u0026amp; J. Fitzgerald (Eds.), \u003cem\u003eHandbook of writing research\u003c/em\u003e (pp. 54\u0026ndash;66). Guilford Press.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQian, Y. (2025). Pedagogical Applications of Generative AI in Higher Education: A Systematic Review of the Field. \u003cem\u003eTechTrends\u003c/em\u003e \u003cb\u003e69\u003c/b\u003e, 1105\u0026ndash;1120. https://doi.org/10.1007/s11528-025-01100-1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReddy, Y. M., \u0026amp; Andrade, H. (2010). A review of rubric use in higher education. \u003cem\u003eAssessment \u0026amp; Evaluation in Higher Education, 35\u003c/em\u003e(4), 435\u0026ndash;448. https://doi.org/10.1080/02602930902862859\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSadasivan, V. S., et al. (2023). Can AI-generated text be reliably detected? \u003cem\u003earXiv preprint\u003c/em\u003e. https://arxiv.org/abs/2303.11156\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSullivan, M., Kelly, A., \u0026amp; McLaughlan, P. (2023). ChatGPT in higher education: Considerations for academic integrity and student learning. \u003cem\u003eJournal of Applied Learning \u0026amp; Teaching, 6\u003c/em\u003e(1), 1\u0026ndash;10. https://doi.org/10.37074/jalt.2023.6.1.17\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTian, Y., \u0026amp; Liu, D. (2024). A bibliometric study of identity construction in English writing for academic purposes. \u003cem\u003eFrontiers in Psychology, 15\u003c/em\u003e, 1499917. https://doi.org/10.3389/fpsyg.2024.1499917\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTroughton, T. E. (2024). Exploring the writing process of multilingual postsecondary students. \u003cem\u003eDiscourse and Writing, 34\u003c/em\u003e, 24\u0026ndash;42. https://doi.org/10.31468/dwr.1045\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVlachopoulos, D., \u0026amp; Makri, A. (2024). A systematic literature review on authentic assessment in higher education: Best practices for the development of 21st century skills, and policy considerations. Studies in Educational Evaluation, 83, 101425. https://doi.org/10.1016/j.stueduc.2024.101425\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVygotsky, L. S. (1978). \u003cem\u003eMind in society: The development of higher psychological processes\u003c/em\u003e. Harvard University Press.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWarschauer, M., et al. (2023). The affordances and contradictions of AI-generated text for writers of English as a second or foreign language. \u003cem\u003eJournal of Second Language Writing, 62\u003c/em\u003e, 101071. https://doi.org/10.1016/j.jslw.2023.101071\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeber-Wulff, D., et al. (2023). Testing of detection tools for AI-generated text. \u003cem\u003eInternational Journal for Educational Integrity, 19\u003c/em\u003e(1), 26. https://doi.org/10.1007/s40979-023-00146-z\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"University of Bridgeport","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, student voice, writing authenticity, educational technology, minority-serving institutions, academic integrity, large language models, writing assessment","lastPublishedDoi":"10.21203/rs.3.rs-8427622/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8427622/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe rapid rise in generative artificial intelligence programs like ChatGPT and Claude has prompted important questions around authorship, student voice, and academic integrity. This mixed-methods study surveyed 102 students and 25 faculty at a minority-serving institution to explore perceptions of AI writing tools and their impacts on writing authenticity. The surveys included 16\u0026ndash;18 quantitative items on five-point Likert scales and 5 qualitative open-ended questions, gathering information on patterns of AI use, confidence, ethics, and institutional supports. Few students (21%) used AI to complete assignments, but 64% used it for revisions and 43% for clarity support. Faculty and students viewed grammar support as AI's most positive use, though students expressed concerns about originality. A significant majority (76%) used AI without disclosure, constituting an academic integrity violation. Responses about ethics were split between \"neither agree nor disagree\" (54%) and those acknowledging violations (36%). Multilingual students valued AI assistance with Standard Academic English grammar, viewing it as a positive learning addition. However, students worried these tools could diminish student voice and homogenize written perspectives across cultures. The 47% perception gap between faculty estimates (68%) and student self-reports (21%) of AI use for complete drafting suggests prohibitionist policies may address faculty concerns more than student realities. Findings support distinguishing instrumental support (grammar, mechanics) from expressive support (ideas, voice) in developing AI policies that preserve authentic student perspective while acknowledging AI's legitimate uses.\u003c/p\u003e","manuscriptTitle":"Writing with AI at the Margins: Student Voice and Authenticity at a Minority-Serving Institution","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-24 10:36:31","doi":"10.21203/rs.3.rs-8427622/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":"a8dd70e2-5d86-4722-8e78-598c634e66c8","owner":[],"postedDate":"December 24th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":60084791,"name":"Educational Philosophy and Theory"}],"tags":[],"updatedAt":"2025-12-24T10:36:31+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-24 10:36:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8427622","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8427622","identity":"rs-8427622","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}