Private university (Spain)

Context

A university with 8,000 students found that its assessment mechanisms — final assignments and remote multiple-choice exams — had stopped being reliable evidence after the irruption of generative AI.

Macro–meso–micro reading

• Macro: the educational model declared competency-based learning, but assessment measured reproduction.
• Meso: every degree programme assessed differently; there were no institutional criteria.
• Micro: faculty lacked instruments to observe process.

Intervention

An 18-month process: diagnosis, institutional framework for authentic assessment, co-design with 60 faculty leaders, pilots in 14 courses, staged training and a process-evidence system.

The university had initially responded like so many others: oral defences to verify authorship. It worked as a containment measure, but it did not solve the underlying problem — it was still assessing the final product, just with an extra check — and it did not scale. The redesign shifted the centre of gravity: from verifying who did the work to making how it was learned observable.

Results and lessons

• Assessment framework approved by the academic council.
• 14 degree programmes with redesigned assessment maps.
• Process evidence — documented iterations, justified decisions, judicious AI use as part of the competency — became the basis for grading; the oral defence remained a complementary instrument, not the sole safeguard.
• Key lesson: faculty adoption was achieved when process assessment reduced grading workload instead of increasing it.