Generative AI as Teacher-Guided Feedback: Exploring Independent Academic Writing among First-Generation Undergraduates in Tamil Nadu

Abstract

First-generation undergraduates in Tamil Nadu often arrive at college with strong oral reasoning but little prior exposure to the conventions of academic English. Many have studied in Tamil-medium schools, and few have a parent or older sibling who can read a draft and say what is missing. Generative AI tools now offer them a reader at any hour. Yet unmediated use can slide into outsourcing, where the tool writes and the student submits. This article examines a middle path: generative AI positioned as teacher-guided feedback rather than as a text producer. Drawing on a qualitative multiple-case design in [two] arts and science colleges in the Kongu region, we followed [n] first-generation undergraduates across a [12]-week writing module. Teachers set explicit boundaries, modelled feedback-seeking prompts, and required students to keep a revision log showing what they accepted, rejected or reworked. Data came from semi-structured interviews, draft histories, revision logs and teacher reflective notes, analysed through reflexive thematic analysis. Four themes emerged: (1) from answer-seeking to question-asking, (2) the teacher as a trust anchor, (3) language as both bridge and barrier, and (4) the slow growth of evaluative judgement. Students who used the tool within a teacher-framed routine reported greater confidence in planning and revising their own arguments, while teacher mediation appeared central to preventing dependency. The study argues that the pedagogical value of generative AI for first-generation writers lies less in the tool than in the feedback culture built around it. Implications are offered for writing instruction, institutional AI policy and bilingual support in regional Indian higher education.

Introduction

The text presents a study on teacher-guided generative AI feedback for improving academic writing among first-generation undergraduate students in Tamil Nadu. It focuses on students in small-town arts and science colleges, particularly those who may understand concepts well but struggle to express their ideas in English academic writing.

Background and Problem

First-generation students often have limited access to writing support outside college. They may face difficulties with:

• Writing clearly in English.
• Organizing arguments and paragraphs.
• Understanding academic writing expectations.
• Receiving detailed feedback from teachers because of large class sizes.
• Knowing how to identify and correct weaknesses in their own writing.

Generative AI tools such as ChatGPT, Gemini, and Copilot can provide rapid feedback on content, organization, grammar, and language. However, unguided AI use can become a problem if students ask AI to write or rewrite their assignments instead of learning to improve their own writing.

The study therefore proposes using AI as a feedback assistant rather than a writing replacement, with the teacher controlling how and when students use it.

Research Objectives

The study investigates three main questions:

1. How do first-generation students use generative AI for writing feedback when teachers provide guidance?
2. How does teacher mediation determine whether AI supports students' writing decisions or replaces them?
3. How do students perceive changes in their independence and judgment as academic writers?

Literature Review

Previous research suggests that AI can provide useful writing feedback, particularly for:

• Grammar and language improvement.
• Content development.
• Organization and structure.
• Generating questions about an argument.
• Providing immediate feedback when teacher support is limited.

However, research also identifies important risks. Students with weaker writing skills may focus mainly on grammar corrections and struggle to understand higher-level feedback about argument and organization. Some students may simply copy AI-generated revisions instead of thinking about the feedback.

The literature therefore increasingly supports teacher-guided or human–AI collaboration, where students remain responsible for making writing decisions.

Theoretical Framework

The study uses two major concepts:

1. Feedback Literacy

Students need to learn not simply to receive feedback but to understand, evaluate, and act on it. Teachers play an important role in developing this ability.

2. Student Engagement

Student interaction with AI feedback is examined through three dimensions:

• Behavioural: What students do with AI suggestions.
• Cognitive: How deeply they understand and evaluate the suggestions.
• Affective: Their confidence, trust, frustration, and attitudes toward AI.

Teacher-Guided AI Feedback Model

The proposed model has four key features:

Students can ask AI for comments, questions, explanations, and suggestions, but they are not allowed to ask it to write or rewrite entire passages.

Methodology

The study uses a qualitative multiple-case research design. It is conducted in two arts and science colleges in the Erode–Tiruppur region of western Tamil Nadu.

The participants include:

• 55 focal students who were followed in detail.
• 45 additional students whose drafts and revision logs were analyzed.
• 5 teachers who taught the writing module.

Participants are first-generation college students, meaning neither parent had completed a college degree.

AI Feedback Module

The module lasts approximately 12 weeks and focuses on two major assignments:

• A 600-word argumentative essay.
• A 1,000-word research report.

The process involves three stages:

1. Introduction and rules: Teachers explain AI capabilities, limitations, errors, and acceptable uses.
2. Teacher modelling: Teachers demonstrate useful prompts in English and Tamil.
3. Guided practice: Students create multiple drafts, obtain AI feedback, record the suggestions, and explain whether they accepted, rejected, or modified each suggestion.

Teachers also review students' revision logs and conduct short individual conferences.

Main Idea

The central argument of the study is that AI should supplement rather than replace student writing. The goal is not to make students independent of all assistance, but to make them independent decision-makers when using assistance.

A student who evaluates AI feedback, rejects unsuitable suggestions, and rewrites ideas in their own words demonstrates greater writing independence than a student who simply copies AI-generated text.

Conclusion

First-generation undergraduates in Tamil Nadu bring real intellectual resources to college, yet they often meet academic writing alone. Generative AI can offer them a patient, always-available reader, and that is no small thing. Left unguided, though, the same tool can just as easily write for them, leaving the underlying skill undeveloped and their confidence no stronger.This study suggests that the outcome depends largely on the teacher. When teachers set clear limits, modelled good questions, required students to account for their choices and talked with them about those choices, AI feedback became a way of learning to judge one\'s own writing. Students moved, slowly and unevenly, from asking for correct answers to asking better questions, and from trusting the machine by default to weighing what it said.The lesson for regional Indian higher education is not that every classroom needs better technology. It is that a modest, well-designed feedback routine can turn a widely available tool into genuine support for students who have had too little of it. Independence, for these writers, does not mean doing without help. It means learning to decide what help is worth taking.

References

[1] Baek, C., Tate, T., &Warschauer, M. (2024). \"ChatGPT seems too good to be true\": College students\' use and perceptions of generative AI. Computers and Education: Artificial Intelligence, 7, Article 100294. https://doi.org/10.1016/j.caeai.2024.100294 [2] Barrot, J. S. (2023). Using ChatGPT for second language writing: Pitfalls and potentials. Assessing Writing, 57, Article 100745. https://doi.org/10.1016/j.asw.2023.100745 [3] Beck, S. W., & Levine, S. (2024). The next word: A framework for imagining the benefits and harms of generative AI as a resource for learning to write. Reading Research Quarterly, 59(4), 706–715. https://doi.org/10.1002/rrq.567 [4] 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 [5] Carless, D., &Winstone, N. (2023). Teacher feedback literacy and its interplay with student feedback literacy. Teaching in Higher Education, 28(1), 150–163. https://doi.org/10.1080/13562517.2020.1782372 [6] Chan, C. K. Y., & Hu, W. (2023). Students\' voices on generative AI: Perceptions, benefits, and challenges in higher education. International Journal of Educational Technology in Higher Education, 20, Article 43. https://doi.org/10.1186/s41239-023-00411-8 [7] Cummings, R. E., Monroe, S. M., & Watkins, M. (2024). Generative AI in first-year writing: An early analysis of affordances, limitations, and a framework for the future. Computers and Composition, 71, Article 102827. https://doi.org/10.1016/j.compcom.2024.102827 [8] Escalante, J., Pack, A., & Barrett, A. (2023). AI-generated feedback on writing: Insights into efficacy and ENL student preference. International Journal of Educational Technology in Higher Education, 20, Article 57. https://doi.org/10.1186/s41239-023-00425-2 [9] Guo, K., & Wang, D. (2024). To resist it or to embrace it? Examining ChatGPT\'s potential to support teacher feedback in EFL writing. Education and Information Technologies, 29(7), 8435–8463. https://doi.org/10.1007/s10639-023-12146-0 [10] Koltovskaia, S., Rahmati, P., &Saeli, H. (2024). Graduate students\' use of ChatGPT for academic text revision: Behavioral, cognitive, and affective engagement. Journal of Second Language Writing, 65, Article 101130. https://doi.org/10.1016/j.jslw.2024.101130 [11] Mahapatra, S. (2024). Impact of ChatGPT on ESL students\' academic writing skills: A mixed methods intervention study. Smart Learning Environments, 11, Article 9. https://doi.org/10.1186/s40561-024-00295-9 [12] Steiss, J., Tate, T., Graham, S., Cruz, J., Hebert, M., Wang, J., Moon, Y., Tseng, W., Warschauer, M., & Olson, C. B. (2024). Comparing the quality of human and ChatGPT feedback of students\' writing. Learning and Instruction, 91, Article 101894. https://doi.org/10.1016/j.learninstruc.2024.101894 [13] Tate, T. P., Harnick-Shapiro, B., Ritchie, D. R., Tseng, W., Dennin, M., &Warschauer, M. (2025). Incorporating generative AI into a writing-intensive undergraduate course without off-loading learning. Discover Computing, 28, Article 72. https://doi.org/10.1007/s10791-025-09563-9 [14] Warschauer, M., Tseng, W., Yim, S., Webster, T., Jacob, S., Du, Q., & Tate, T. (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, Article 101071. https://doi.org/10.1016/j.jslw.2023.101071 [15] Yeung, S. (2025). University students\' engagement with generative AI-supported automated writing evaluation (AWE) feedback. Journal of Second Language Writing, 68, Article 101203. https://doi.org/10.1016/j.jslw.2025.101203

Copyright

Copyright © 2026 Ms. K. Kayathri, Ms. P. Poonkodi, Ms. S. Mythili, Ms. K. Kavipriya. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.