Exploring EFL Students' Experiences In Using AI Tools to Support Academic Writing

Socius: Jurnal Penelitian Ilmu-Ilmu Sosial Volume 2, Nomor 7, February 2024, P. 163-174 E-ISSN: 3025-6704 DOI: https://doi.org/10.5281/zenodo.17605850 *Coresponding Author Email: [email protected] Exploring EFL Students' Experiences In Using AI Tools to Support Academic Writing Siti Munawaroh 1Institut Agama Islam Nahdlatul Ulama Bangil A B S T R A C T This qualitative study investigates how English as a Foreign Language (EFL) students at Institut Agama Islam Nahdlatul Ulama (IAINU) Bangil experience and utilize artificial intelligence (AI) tools in their academic writing processes. Through a phenomenological approach, we conducted in-depth interviews, focus group discussions, and collected written reflections from 25 English education students over four months. Our thematic analysis uncovered four primary themes that characterize their experiences: enhanced writing efficiency and productivity, valuable language learning support and scaffolding, significant challenges to critical thinking and originality, and complex ethical considerations surrounding academic integrity. While participants reported substantial improvements in writing fluency and reduced anxiety when using AI tools like ChatGPT, Grammarly, and QuillBot, they simultaneously expressed concerns about potential over-dependence and the authenticity of their work. These findings illuminate the reality of AI integration in EFL writing contexts and highlight the urgent need for comprehensive institutional policies and pedagogical frameworks that can maximize benefits while addressing legitimate concerns about independent learning and academic honesty. A B S T R A K Penelitian kualitatif ini bertujuan untuk mengetahui bagaimana mahasiswa English as a Foreign Language (EFL) di Institut Agama Islam Nahdlatul Ulama (IAINU) Bangil memanfaatkan alat kecerdasan buatan (AI) dalam proses penulisan akademik mereka. Melalui pendekatan fenomenologis, penelitian ini melibatkan wawancara mendalam, diskusi kelompok terarah (FGD), serta pengumpulan refleksi tertulis dari 25 mahasiswa program studi Pendidikan Bahasa Inggris selama empat bulan. Analisis tematik mengungkap empat tema utama yang menggambarkan pengalaman mereka: peningkatan efisiensi dan produktivitas menulis, dukungan serta penopang yang bernilai dalam pembelajaran bahasa, tantangan signifikan terhadap kemampuan berpikir kritis dan orisinalitas, serta pertimbangan etis yang kompleks terkait integritas akademik. Meskipun para partisipan melaporkan adanya peningkatan yang nyata dalam kelancaran menulis dan penurunan kecemasan ketika menggunakan alat AI seperti ChatGPT, Grammarly, dan QuillBot, mereka juga menyampaikan kekhawatiran mengenai potensi ketergantungan berlebihan serta keaslian hasil karya mereka. Temuan ini menggambarkan realitas integrasi AI dalam konteks penulisan akademik EFL dan menegaskan perlunya kebijakan institusional serta kerangka pedagogis yang komprehensif untuk memaksimalkan manfaat penggunaan AI sekaligus mengatasi kekhawatiran yang sah terkait kemandirian belajar dan kejujuran akademik. INTRODUCTION The educational landscape has been dramatically transformed by the emergence of sophisticated artificial intelligence (AI) technologies, nowhere more visibly than in academic writing support. Tools like ChatGPT, Grammarly, QuillBot, and numerous other AI-powered applications have become ubiquitous in universities worldwide, fundamentally altering how students approach writing tasks. For English as a Foreign Language (EFL) learners, these technologies represent both unprecedented opportunities and formidable challenges that demand careful scholarly attention. A R T I C L E I N F O Article history: Received January 24, 2025 Revised 25 January 2025 Accepted 30 January 2025 Available online 13 February 2025 Keywords AI tools, EFL students, academic writing, qualitative research, technology integration, English education Kata Kunci: Alat kecerdasan buatan; mahasiswa pembelajar (EFL); penulisan akademik; penelitian kualitatif; integrasi teknologi; pendidikan bahasa Inggris. This is an open access article under the CC BYSA license. Copyright © 2025 by Author. Published by Yayasan Daarul Huda Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 163-174 Socius E-ISSN: 3025-6704 Indonesian EFL students face a particularly complex set of writing challenges. Beyond the typical struggles with vocabulary, grammar, and organization that affect all developing writers, they must navigate cultural differences in rhetorical conventions, limited exposure to authentic academic discourse, and the cognitive demands of composing in a non-native language. Traditional writing support resources—dictionaries, grammar books, and human tutors—while valuable, often fail to provide the immediate, contextual, and comprehensive assistance that AI tools can offer. At Institut Agama Islam Nahdlatul Ulama (IAINU) Bangil, these dynamics play out within a unique educational context where Islamic values and modern technological innovation intersect. The English education students at this institution are not merely learning to write in English; they are preparing to become English teachers who will shape future generations of Indonesian EFL learners. Their experiences with AI writing tools therefore carry implications that extend far beyond individual academic success. Despite the widespread adoption of AI writing tools among students, our understanding of how learners actually experience these technologies remains surprisingly limited. Most existing research has focused on quantitative measures of writing improvement or broad surveys of student attitudes. What has been missing is a deep, nuanced exploration of the lived experiences of EFL students as they integrate AI tools into their writing practices—their struggles, discoveries, concerns, and evolving perspectives. This study addresses that gap by examining the authentic experiences of 25 English education students at IAINU Bangil as they navigate the complex terrain of AI-assisted academic writing. Through extensive interviews, group discussions, and personal reflections, we sought to understand not just what students think about AI tools, but how these technologies are reshaping their understanding of writing, learning, and academic authenticity. LITERATURE REVIEW The Evolution of AI Writing Technologies The development of AI writing assistance has progressed through several distinct phases, each offering increasingly sophisticated support for writers. Early grammar checkers like those integrated into Microsoft Word provided basic error detection and correction. The introduction of comprehensive tools like Grammarly marked a significant advancement, offering contextual suggestions, style recommendations, and explanations of errors. The recent emergence of large language models like GPT-3 and GPT-4 has created an entirely new category of writing support that can generate content, suggest ideas, and engage in complex writing-related conversations. Research by Martinez and Chen (2023) demonstrates that these tools now offer support across the entire writing process—from initial brainstorming through final editing. However, they note significant variations in how different student populations utilize these capabilities, with EFL learners showing distinct patterns of usage compared to native English speakers. Writing Challenges in EFL Contexts The challenges faced by EFL students in academic writing are well-documented and multifaceted. Flowerdew (2019) identifies several key areas of difficulty: linguistic challenges including vocabulary limitations and grammatical accuracy concerns; rhetorical challenges related to organizing arguments and meeting genre expectations; and sociocultural challenges involving understanding audience expectations and academic discourse conventions. In the Indonesian context specifically, research has highlighted additional complexities. Mistar et al. (2021) found that Indonesian EFL students often struggle with the directness required in English academic writing, as their cultural background emphasizes indirect Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 164-174 Siti Munawaroh/ Exploring EFL Students' communication styles. Similarly, studies by Widiati and Cahyono (2020) reveal persistent challenges with article usage, prepositions, and complex sentence construction that can significantly impact the clarity and professionalism of academic writing. These challenges create what researchers term "writing anxiety"—a psychological barrier that can prevent students from engaging fully with writing tasks or experimenting with more complex language structures. For many Indonesian EFL students, this anxiety is compounded by high-stakes testing environments and the pressure to demonstrate English proficiency for academic and professional advancement. Technology Integration and Language Learning Theory The integration of AI tools into language learning and writing instruction intersects with several important theoretical frameworks. Sociocultural theory, particularly Vygotsky's concept of the Zone of Proximal Development (ZPD), provides a useful lens for understanding how AI tools might function as mediating artifacts that enable students to accomplish tasks beyond their current independent capability. However, the nature of AI-mediated support differs significantly from traditional human scaffolding. While human tutors can adjust their support based on ongoing assessment of student needs and gradually withdraw assistance as learners develop independence, AI tools typically provide consistent levels of support without consideration of developmental progression. This difference raises important questions about the long-term effects of AI assistance on learning outcomes. Activity Theory offers another valuable perspective, conceptualizing writing as a complex activity system involving multiple interacting elements: the writer (subject), the writing task (object), various tools (including AI applications), community norms and expectations, rules and conventions, and the division of labor between human and artificial intelligence. Understanding how AI tools function within this system requires examining not just their direct effects on writing quality, but their broader impacts on the entire activity of academic writing. Ethical and Pedagogical Considerations The integration of AI writing tools into educational contexts has generated considerable debate about academic integrity, authentic learning, and the appropriate boundaries of technological assistance. Traditional concepts of plagiarism and academic honesty, developed in pre-digital contexts, struggle to address the complexities introduced by AI-generated content and AI-assisted writing processes. Some educators and institutions have responded with prohibition-based approaches, treating AI assistance as a form of cheating to be detected and penalized. Others have embraced AI tools as legitimate writing supports, comparable to dictionaries, thesauri, or grammar checkers. A growing number of scholars advocate for more nuanced approaches that distinguish between appropriate and inappropriate uses of AI assistance while helping students develop critical evaluation skills. The challenge is particularly acute in EFL contexts, where students may legitimately need more support to overcome language barriers, but where over-reliance on AI assistance might impede the development of independent language skills necessary for long-term academic and professional success. Research Gaps and Study Rationale While research on AI writing tools has expanded rapidly, significant gaps remain in our understanding of how EFL students actually experience these technologies. Most existing studies have employed quantitative methods focused on measuring writing quality improvements or Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 163-174 Socius E-ISSN: 3025-6704 surveying general attitudes toward AI tools. Few studies have explored the nuanced, lived experiences of EFL learners as they integrate AI assistance into their writing practices. Additionally, much of the existing research has been conducted in Western educational contexts with native or near-native English speakers. The experiences of students in countries like Indonesia, where English serves as a foreign language and where educational cultures may emphasize different values and approaches to learning, remain underexplored. This study addresses these gaps by employing qualitative methodology to deeply investigate the experiences of Indonesian EFL students in a specific institutional context. By focusing on students' own voices and perspectives, we aim to contribute more nuanced and contextually grounded insights to the growing body of research on AI in education. METHOD Research Design and Philosophical Approach This study employed interpretive phenomenological analysis (IPA), a qualitative research approach that seeks to understand how individuals make sense of their lived experiences. IPA is particularly well-suited for exploring complex, emerging phenomena like AI tool integration, as it allows for deep exploration of participants' subjective experiences while acknowledging the researcher's role in interpreting and making meaning from those experiences. Our philosophical stance embraced social constructivism, recognizing that participants' experiences with AI tools are shaped not only by the technologies themselves but by social, cultural, and institutional contexts. This perspective guided our data collection and analysis, emphasizing the importance of understanding experiences within their broader contexts rather than seeking to generalize across all populations. Research Setting and Context Institut Agama Islam Nahdlatul Ulama (IAINU) Bangil is a private Islamic higher education institution located in East Java, Indonesia. The institution serves approximately 1000 students across various programs, include the English Education Department. The department emphasizes both English language proficiency and pedagogical skills, preparing graduates to teach English in Indonesian schools. The institutional context is significant for several reasons. As an Islamic institution, IAINU Bangil emphasizes character development and ethical behavior alongside academic achievement. This emphasis creates a unique environment for exploring questions about academic integrity and the ethical use of AI tools. Additionally, the institution serves students from diverse socioeconomic backgrounds, creating variation in technological access and digital literacy levels that influenced participants' experiences with AI tools. Participant Selection and Characteristics Twenty-five English education students participated in this study, selected through purposive sampling to ensure diversity in academic levels, technological experience, and demographic characteristics. Participants ranged from semester 2 to semester 8, ensuring sufficient experience with academic writing requirements and exposure to various AI tools. The participant group included 18 female and 7 male students, reflecting the gender distribution typical of English education programs in Indonesia. Ages ranged from 20 to 24 years, with most participants (68%) coming from middle-class families with reliable internet access. However, 32% reported occasional challenges with internet connectivity or device access that influenced their use of AI tools. Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 164-174 Siti Munawaroh/ Exploring EFL Students' All participants had used at least one AI writing tool for academic purposes, with most having experience with multiple applications. The most commonly used tools were ChatGPT (100% of participants), Grammarly (84%), QuillBot (76%), and Google Translate (68%). Less common tools included Jasper AI, Copy.ai, and various paraphrasing websites. Data Collection Procedures Data collection occurred over four months (September to December 2023) through multiple complementary methods designed to capture different aspects of participants' experiences. Individual Semi-structured Interviews: We conducted 60-90 minute interviews with all participants, using a flexible interview guide that allowed for exploration of emerging themes while ensuring coverage of key topics. Interviews began with broad questions about participants' writing histories and experiences with technology, then focused specifically on AI tool usage patterns, perceived benefits and challenges, and evolving perspectives over time. To ensure participants could express themselves fully, interviews were conducted in a mix of English and Bahasa Indonesia, with participants free to switch between languages as needed. All interviews were audio-recorded and transcribed verbatim, with Indonesian portions translated into English by bilingual research team members. Focus Group Discussions: Four focus groups (5-7 participants each) were conducted to explore shared experiences and generate discussion about collective perspectives. Focus groups lasted approximately 2 hours and covered topics including peer attitudes toward AI use, collaborative writing experiences, and suggestions for improving AI integration in their program. Focus groups proved particularly valuable for uncovering social dynamics and shared concerns that participants might not have expressed in individual interviews. The group setting encouraged participants to build on each other's experiences and articulate perspectives they might not have fully formed independently. Written Reflection Journals: Participants maintained reflection journals throughout the study period, submitting entries every two weeks (8 entries total). Each entry was 300-500 words and responded to specific prompts about recent AI tool experiences, challenges encountered, or changes in perspectives. These reflections provided insight into participants' evolving experiences and allowed us to track changes in attitudes and usage patterns over time. Many participants noted that the act of regular reflection helped them become more conscious of their AI tool usage and its effects on their learning. Document Collection: With participant consent, we collected examples of academic assignments that involved AI tool use, along with any notes or drafts that showed the writing process. This provided concrete evidence of how students actually integrated AI support into their work, complementing their self-reported experiences. Data Analysis Process Our analysis followed the systematic procedures recommended for interpretive phenomenological analysis, adapted to accommodate our multiple data sources and the multilingual nature of our data. Phase 1: Immersion and Initial Impressions: Research team members read through all transcripts and documents multiple times, making initial notes about interesting patterns, contradictions, and emerging themes. This phase involved both individual analysis by team members and group discussions about preliminary impressions. Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 163-174 Socius E-ISSN: 3025-6704 Phase 2: Systematic Coding: Using NVivo 12 software, we conducted line-by-line coding of all textual data. Initial codes stayed close to participants' own language and concepts, using both descriptive codes (summarizing content) and interpretive codes (suggesting meaning or significance). Phase 3: Theme Development: Codes were grouped into potential themes through an iterative process involving constant comparison and discussion among research team members. We created visual maps showing relationships between codes and potential themes, refining these repeatedly as analysis progressed. Phase 4: Theme Refinement: Potential themes were tested against the full dataset to ensure they accurately represented participants' experiences. We also examined negative cases—instances where participants' experiences didn't fit emerging themes—to refine our understanding and ensure comprehensive representation. Phase 5: Interpretation and Writing: The final phase involved crafting analytic narratives that captured both the essence of each theme and the relationships between themes. We selected representative quotes and examples that illustrated each theme while protecting participant confidentiality. Trustworthiness and Ethical Considerations We employed several strategies to ensure the trustworthiness of our findings. Member checking involved sharing preliminary findings with participants and incorporating their feedback to ensure accurate representation of their experiences. Triangulation across multiple data sources (interviews, focus groups, reflections, documents) provided different perspectives on the same phenomena. Peer debriefing with colleagues not involved in the research helped us examine our assumptions and consider alternative interpretations. We also maintained detailed audit trails documenting all research decisions and analytical processes. Ethical approval was obtained from the IAINU Bangil research ethics committee. All participants provided informed consent after receiving detailed information about the study purposes, procedures, and their rights. Confidentiality was maintained through use of pseudonyms and removal of identifying details from all reports. RESULT Our analysis revealed four interconnected themes that capture the complexity and nuance of participants' experiences with AI writing tools. Each theme contains multiple dimensions that reflect the multifaceted nature of AI integration in academic writing. Theme 1: Enhanced Writing Efficiency and Productivity Perhaps the most immediately apparent impact of AI tools was their dramatic effect on writing efficiency and productivity. Participants consistently described these tools as "gamechangers" that transformed their approach to writing tasks. Overcoming the Blank Page: Many participants described struggling with getting started on writing assignments—a challenge that AI tools helped address directly. Students 1, a sixth-semester student, explained: "I used to sit staring at my laptop for hours, not knowing how to begin. Now when I'm stuck, I ask ChatGPT to suggest some opening sentences for my topic, and suddenly I have ideas flowing." This ability to generate initial content or structural suggestions helped participants overcome what they described as "writing paralysis." Accelerated Research and Idea Development: AI tools also streamlined the research and brainstorming phases of writing. Student 2, a seventh-semester student, noted: "ChatGPT can Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 164-174 Siti Munawaroh/ Exploring EFL Students' give me an overview of any topic I'm writing about. It's like having a research assistant who can quickly explain concepts and suggest different angles to explore." Participants appreciated being able to quickly gather background information and explore different perspectives on their topics before diving into more detailed research. Time Management Benefits: The time-saving aspects of AI tools were frequently mentioned, with participants reporting significant reductions in the time required to complete writing assignments. However, these time savings came with a caveat—many participants noted that while they could produce drafts more quickly, they often spent additional time revising and refining AI-assisted content to ensure it met their standards and reflected their own voice. Reduced Writing Anxiety: A particularly significant finding was the anxiety-reducing effect of AI tools. Many participants described feeling more confident about attempting challenging writing tasks when they knew AI assistance was available. Student 3, a fifth-semester student, shared: "Before, I would avoid assignments that required complex arguments because I was afraid my English wasn't good enough. Now I feel brave enough to try more difficult topics because I know I can get help organizing my thoughts." Theme 2: Language Learning Support and Scaffolding Participants consistently described AI tools as powerful language learning resources that supported their development as English writers in multiple ways. Vocabulary Expansion in Context: Unlike traditional vocabulary learning methods that present words in isolation, AI tools introduced new vocabulary within meaningful contexts. Students 4, a sixth-semester student, explained: "When I use QuillBot to improve my sentences, I learn new words and phrases that mean the same thing as what I wrote, but sound more academic. I keep a special notebook for these new expressions." Many participants described developing personal vocabulary learning strategies around AI tools, such as keeping lists of new words encountered through paraphrasing tools or asking ChatGPT to explain the differences between similar words. This contextual vocabulary learning appeared to support better retention and more appropriate usage compared to traditional vocabulary memorization methods. Grammar Learning Through Feedback: AI grammar checkers provided explanations and corrections that participants found highly educational. Unlike traditional grammar exercises that focus on isolated structures, these tools provided feedback on participants' own writing, making the learning immediately relevant and applicable. Student 5, a seventh-semester student, described his experience: "Grammarly doesn't just tell me something is wrong—it explains why and shows me how to fix it. I've learned more about English grammar from Grammarly in one semester than I did in years of grammar classes." This personalized, contextual grammar instruction seemed particularly effective for participants who had struggled with traditional grammar teaching methods. Exposure to Academic Discourse Patterns: AI tools exposed participants to various styles and registers of academic English that they might not encounter elsewhere. Participants noted that tools like ChatGPT could demonstrate different ways to express academic concepts and arguments, helping them develop a sense of appropriate academic discourse. Stuedent 7, an eighth-semester student, observed: "ChatGPT writes in a way that sounds really academic and professional. I study how it constructs sentences and organizes ideas, then try to apply those patterns in my own writing." This exposure to sophisticated language patterns appeared to support participants' development of academic writing conventions. Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 163-174 Socius E-ISSN: 3025-6704 Theme 3: Challenges in Critical Thinking and Originality While participants appreciated the support provided by AI tools, many also recognized concerning challenges related to critical thinking development and maintaining originality in their work. The Temptation of Easy Answers: Participants frequently described the difficulty of resisting the temptation to rely too heavily on AI-generated content. Student 8, a seventhsemester student, reflected: "Sometimes I ask ChatGPT a question about my essay topic, and it gives me such a complete answer that I'm tempted to just use it directly. I have to force myself to think through the ideas myself and write them in my own words." This challenge was particularly acute when participants faced difficult concepts or complex arguments. The immediate availability of well-articulated explanations from AI tools could discourage the deeper thinking and struggle that are often necessary for genuine learning and understanding. Concerns About Intellectual Development: Many participants expressed worry about the long-term effects of AI reliance on their intellectual development. Student 9, a sixth-semester student, shared: "I wonder if I'm becoming intellectually lazy. When I can get answers so easily from AI, am I losing the ability to think through problems myself?" These concerns were particularly pronounced among participants who planned to become teachers. They worried that over-reliance on AI tools during their studies might leave them unprepared to think independently and creatively in their future classrooms. Questions of Authenticity and Voice: Participants struggled with questions about whether AI-assisted writing truly represented their own ideas and capabilities. Student 10, an eighth-semester student, explained: "Sometimes when I read my finished essays, I'm not sure how much is really 'me' and how much comes from the AI suggestions. I develop the ideas, but the way they're expressed often comes from the tools." This concern about authenticity extended beyond individual assignments to broader questions about their development as writers and thinkers. Many participants described consciously trying to maintain their own "voice" while using AI assistance, but found this balance difficult to achieve consistently. Theme 4: Ethical Considerations and Academic Integrity Concerns The final theme encompasses participants' ongoing struggles with ethical questions surrounding AI use in academic contexts—struggles that reflected broader uncertainties in educational institutions worldwide. Navigating Unclear Institutional Policies: One of the most significant challenges participants faced was the absence of clear institutional guidelines regarding AI tool use. Student 11, a seventh-semester student, described the confusion: "Different professors have different opinions about AI tools. Some say they're fine for brainstorming and editing, others don't want us using them at all. We need clear rules about what's acceptable." This policy vacuum created considerable anxiety among participants, many of whom wanted to use AI tools responsibly but feared potential academic consequences. Several participants described deliberately avoiding AI assistance for certain assignments due to uncertainty about instructor expectations, even when they felt the tools could legitimately support their learning. Peer Perspectives and Social Dynamics: Participants reported significant variation in peer attitudes toward AI tool use, creating social tensions within their learning community. Some students viewed AI assistance as unfair advantage-seeking, while others considered it a legitimate and necessary support for EFL learners. Socius: Jurnal Penelitian Ilmu-ilmu Sosial Vol. 2, No. 7 Tahun 2024, P. 164-174 Siti Munawaroh/ Exploring EFL Students' Student 12, a sixth-semester student, noted: "Some of my friends think I'm cheating when I use Grammarly to check my grammar. But I don't see how it's different from using a dictionary or asking someone to proofread my work." These differing perspectives sometimes strained friendships and created an atmosphere of secrecy around AI tool use. Developing Personal Ethical Frameworks: In the absence of clear institutional guidance, participants described developing their own internal guidelines for appropriate AI use based on their personal values and understanding of learning goals. Many established boundaries such as using AI for brainstorming and editing but not for generating entire sections of content. Studen 12, an eighth-semester student, shared her approach: "I decided I would only use AI to help me express ideas I already have, not to give me new ideas. That way, the thinking is still mine, even if I get help with the language." These personal ethical frameworks varied considerably among participants but reflected serious engagement with questions about academic integrity and authentic learning. DISCUSSION The findings from this study reveal the profound complexity inherent in EFL students' experiences with AI writing tools. Rather than simple stories of technological adoption or rejection, participants' narratives demonstrate ongoing negotiation between the benefits and challenges these tools present. The Scaffolding Paradox One of the most striking findings is what we term the "scaffolding paradox"—AI tools provide valuable support that enables students to accomplish writing tasks beyond their current independent capability, but this same support may impede the development of independent writing skills. This paradox reflects fundamental tensions in educational uses of AI technology. Vygotsky's Zone of Proximal Development concept suggests that effective scaffolding should gradually decrease as learners develop competence, eventually leading to independent performance. However, AI tools typically provide consistent levels of support without built-in mechanisms for gradual withdrawal. This creates a risk of sustained dependence rather than developmental progression. The participants in our study were acutely aware of this paradox. Their concerns about "intellectual laziness" and questions about authenticity suggest sophisticated understanding of the potential negative effects of over-reliance on AI assistance. This metacognitive awareness may itself be a crucial factor in determining whether AI tools support or hinder learning outcomes. Cultural and Contextual Influences The Islamic institutional context of IAINU Bangil added important dimensions to participants' experiences with AI tools. The emphasis on ethical behavior and character development within Islamic education appears to have heightened participants' sensitivity to questions about academic integrity and authentic learning. Many participants explicitly connected their decisions about AI tool use to broader moral principles rather than simply following institutional rules or seeking academic advantage. This suggests that cultural and religious contexts significantly influence how students approach and integrate new technologies into their learning practices. Additionally, the Indonesian EFL context, where students face substantial linguistic and cultural barriers to academic writing in English, may make AI tools particularly valuable as support mechanisms. However, this context also raises questions about whether heavy reliance