AI assessment needs proof of what students can do alone
Samar Ansari reviewed 41 research and sector sources and 55 policy documents, mostly tied to the UK. The paper argues that assessment needs clear AI-use rules and separate proof of what students can do without AI. It does not measure whether those changes help students or reflect Texas schools.
What the source reports
Samar Ansari tested four ideas about assessment in the age of generative AI. One person coded 41 research and sector sources and 55 school or college policy documents. The review puts evidence about learning beside rules institutions have written. Its argument is that schools cannot rely on AI detectors or student disclosures alone. They also need to state which uses are allowed and gather direct proof of skills without AI help. That is a design claim, not a result from a classroom trial. The sources lean toward the UK, and their counts do not tell us how Texas schools teach or assess. A local policy would still need its own test of whether students understand the rules and can show their work.
Original source
- Title
- From Policing Provenance to Designing for Evidence: Testing Four Propositions on Generative AI Against Research and Institutional Policy
- Author
- Samar Ansari
- Publication
- arXiv cs.CY
- Date
- Tuesday, November 3, 2026