How Software Engineering Students Use LLMs to Write Research Papers: An Experience Report

2026-06-04 · Source: cs.SE updates on arXiv.org · Field: Education & Learning — Educational Technology (EdTech), Academic Research & Higher Education · Depth: Intermediate, long

Summary

An educational experience integrated reflective Large Language Model (LLM) use into a third-year software architecture course's empirical methods assignment. Students, tasked with developing a 2-3 page research paper using rapid or gray literature review methodologies, were required to disclose their LLM usage. Researchers analyzed 146 student disclosure statements, combining LLM-assisted categorization with manual verification. The analysis revealed students incorporated LLMs for presentation refinement, brainstorming, methodological clarification, organizing findings, and articulating relevance. While students found LLMs useful for grammar, idea organization, and concept comprehension, they also reported concerns about inaccuracies, hallucinations, and potential meaning distortion, emphasizing the need for critical verification.

Key takeaway

For educators designing empirical software engineering assignments, integrating reflective LLM disclosure activities is crucial. This approach encourages students to critically engage with AI tools, understanding both their benefits for writing and organization, and the inherent risks of inaccuracies and meaning distortion. You should provide explicit guidance on verifying AI-generated content and maintaining authorship to foster responsible AI-assisted academic work.

Key insights

Students use LLMs across academic writing, balancing utility for clarity and organization with critical verification of AI-generated content.

Principles

• LLMs serve as interactive writing companions.
• Critical verification of AI output is essential.
• Transparency in LLM use fosters reflection.

Method

Researchers conducted a four-phase cross-analysis of 146 student disclosures, combining ChatGPT-4.0 for initial categorization based on peer-review dimensions (Novelty, Rigor, Relevance, Transparency, Presentation) with manual verification and refinement.

In practice

• Use LLMs for grammar and sentence structure.
• Employ LLMs for brainstorming research topics.
• Clarify empirical methods with LLM assistance.

Topics

• Large Language Models
• Software Engineering Education
• Empirical Methods
• Academic Writing
• Student AI Use
• AI Ethics

Best for: Research Scientist, AI Ethicist

Related on AIssential

• Using AI for Writing like a Responsible Adult — Responsible AI writing requires human judgment to guide models and evaluate outputs, avoiding pandering.
• The Revenge of the Data Scientist — Core data science skills are essential for effective LLM evaluation and robust AI system development.
• From Earth to Algorithms: Generative AI in Geoscience — AI literacy and ethical guidelines are crucial for responsible and effective LLM integration in geoscience.
• The Crutch or the Ceiling? How Different Generations of LLMs Shape EFL Student Writings — Advanced LLMs improve EFL writing metrics but may mask true ability and hinder deep coherence.
• Attributing and situating knowledge cannot be left to language models — LLMs cannot reliably handle scholarly attribution, risking miscitation and undermining research standards.
• Is AI Making Us Dumber? — Relying on LLMs for writing tasks can diminish critical thinking and reduce brain activity when tools are removed.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.SE updates on arXiv.org.