Abstract
Generative AI has prompted renewed scrutiny of how universities assess student learning and interpret assessment outcomes. This study examined the implications of generative AI for assessment validity by investigating how AI-generated coursework was evaluated within an authentic university assessment process. Using a real-world testing approach, a research team generated two American Psychological Association-style empirical research reports using ChatGPT and submitted them for marking within an undergraduate psychology degree programme. Neither submission was identified as unusual by the markers, and both received passing grades under routine marking conditions. These findings demonstrate that, in these instances, AI-generated coursework was evaluated as authentic student work and received passing grades under routine marking conditions, raising important questions about the extent to which written coursework can continue to function as evidence of student learning when generative AI tools are readily available. The study contributes to ongoing SoTL discussions regarding assessment validity, the interpretation of assessment outcomes, and the design of assessment practices in AI-rich educational contexts.
| Original language | English |
|---|---|
| Journal | Teaching and Learning Inquiry |
| Publication status | Accepted/In press - 25 Sept 2026 |
Keywords
- generative AI
- assessment validity
- university assessment
- student learning
Fingerprint
Dive into the research topics of 'Generative AI, written coursework, and assessment validity in higher education'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver