From workload to verified learning: reframing ECTS for AI-rich assessment

Generative artificial intelligence (GenAI) weakens a practical assumption in the European Credit Transfer and Accumulation System (ECTS): that estimated student workload remains a credible proxy for learning when assessment evidence is interpretable. This conceptual article examines what follows when academic products are routinely produced through hybrid human–AI cognition. Drawing on Kane’s interpretation/use argument, assessment-security scholarship and recent work on GenAI in higher education, it reconstructs ECTS as an inferential system in which scores, generalisation, extrapolation and credit-awarding uses must all be warranted. The analysis shows that opaque take-home production can weaken each link: scoring may misread distributed authorship, generalisation becomes unstable, extrapolation to learning outcomes is less secure, and credit awards risk retaining administrative form without evidential force. The article proposes verified learning as a credit-level validity condition and specifies four safeguards: programme-level alignment, authentic transfer tasks, process visibility with declared AI use, and strategic controlled verification. It also identifies equity and governance conditions required for these safeguards to support, rather than undermine, comparability across ECTS-based institutions in AI-rich assessment environments. The contribution is to connect task-level assessment validity to credit legitimacy and to translate that connection into practical design and quality-assurance implications for European higher education policy.