Integrating AI into Requirements Quality Learning in Software Engineering Education: A TPACK-Guided Empirical Study

Structured assignment design matters for responsible AI use: When educators carefully designed the requirements engineering assignment with specific learning goals in mind, students used the AI tool selectively for analysis and evaluation rather than relying on it for full automation—demonstrating that thoughtful pedagogical structure can guide students toward more critical, rather than passive, engagement with AI tools.

AI tools work better for concrete, measurable learning outcomes: The study found that students showed the most improvement in understanding concrete requirements quality dimensions like "value articulation" and "testability," while more abstract concepts like "negotiability" showed mixed results, suggesting that AI integration is most effective when paired with clearly defined, measurable learning objectives.

Students develop "conditional trust" rather than blind reliance: Rather than accepting AI outputs uncritically, students reported actively refining AI suggestions and maintaining awareness of the tool's limitations—indicating that proper educational framing can foster healthy skepticism and critical evaluation of AI-generated content.

TPACK framework provides practical guidance for responsible AI in education: The study demonstrates that integrating pedagogical content knowledge (TPACK) into AI tool design helps align AI capabilities with actual learning goals, offering educators a concrete methodology for implementing AI responsibly rather than simply adopting tools without educational purpose.

Usability challenges remain a barrier to equitable AI adoption: Students reported moderate usability difficulties with the AI tool, which could create disparities in who benefits from AI-assisted learning—highlighting that responsible AI integration requires attention to accessibility and user experience, not just pedagogical alignment.