Yao (2026) — Independent / university policy audit. arXiv preprint (cs.CY).
Generative AI undermines a basic premise of educational assessment: that submitted work reliably evidences the human capacities a credential certifies. This paper proposes cognitive stewardship, a framework linking four elements \u2014 the learning claim, the delegation boundary, the evidence standard, and safeguards \u2014 to reason about what remains inferable about learning once cognitive work is delegated to AI. It then audits verified public GenAI assessment guidance from 30 universities using a pre-specified scoring codebook applied by four open-weight LLMs as structured coders, with scores averaged to dampen single-model bias.\n\nThe audit's headline finding is a governance gap already visible in genai-assessment-governance and genai-policies-higher-ed-computing: institutional policies are getting better at classifying AI use but not at explaining what evidence of learning remains valid under each class. The framework sharpens the wiki's assessment-validity thread \u2014 shifting the question from detection and academic-integrity enforcement toward specifying delegation boundaries per learning claim \u2014 and complements ai-assessment-scale-reform and universities-ai-era-rethinking on redesigning credentials for AI-mediated education.
Related Pages
- assessment-validity — related wiki thread
- academic-integrity — related wiki thread
- ai-assessment-scale-reform — related wiki thread
- universities-ai-era-rethinking — related wiki thread
- genai-assessment-governance — related wiki thread