Students' Perception of LLM Use in Requirements Engineering Education: An Empirical Study Across Two Universities

TL;DR

The paper evaluates guided use of Large Language Models (LLMs) in Requirements Engineering education across two universities with distinct pedagogy (UA's blended self-directed tasks vs UB's Agile, team-based project). It uses a student survey (n=179) to assess impacts on understanding, motivation, and task performance, identifying benefits in elicitation and documentation alongside concerns about integrity, accuracy, and overreliance. Findings show LLMs can boost engagement and conceptual understanding, particularly for advanced students, but efficacy depends on course structure and explicit guidelines for use. The work argues for structured AI-assisted learning with prompt-engineering training and ethics policies to maximize benefits while preserving critical thinking and collaboration in RE education.

Abstract

The integration of Large Language Models (LLMs) in Requirements Engineering (RE) education is reshaping pedagogical approaches, seeking to enhance student engagement and motivation while providing practical tools to support their professional future. This study empirically evaluates the impact of integrating LLMs in RE coursework. We examined how the guided use of LLMs influenced students' learning experiences, and what benefits and challenges they perceived in using LLMs in RE practices. The study collected survey data from 179 students across two RE courses in two universities. LLMs were integrated into coursework through different instructional formats, i.e., individual assignments versus a team-based Agile project. Our findings indicate that LLMs improved students' comprehension of RE concepts, particularly in tasks like requirements elicitation and documentation. However, students raised concerns about LLMs in education, including academic integrity, overreliance on AI, and challenges in integrating AI-generated content into assignments. Students who worked on individual assignments perceived that they benefited more than those who worked on team-based assignments, highlighting the importance of contextual AI integration. This study offers recommendations for the effective integration of LLMs in RE education. It proposes future research directions for balancing AI-assisted learning with critical thinking and collaborative practices in RE courses.

Figures (8)

• Figure 1: Impact of LLMs on understanding RE concepts
• Figure 2: Influence of LLMs on student's motivation to learn RE
• Figure 3: Agreement with the integration of LLMs being smooth and effective
• Figure 4: The most common challenges students perceived in using LLMs in the course. Labels: a. Understanding how to use LLMs effectively, b. Technical issues (access or usability), c. Overreliance on LLMs for answers, d. Difficulty integrating LLM-generated outputs, e. Concern about academic integrity.
• Figure 5: Students' perception of the support of LLMs for RE tasks