University teaching after generative AI: beyond lectures and inquiry to a certification-and-consultation paradigm in which learners become evaluators

Undergraduate teaching has rested on two paradigms, lectures and inquiry. Both treat what a student submits, an answer or the product of an inquiry, as evidence of the student's own learning, and generative AI breaks that premise. Credible judgement of ability now requires face-to-face probing, which demands more human time than the current staffing model can supply. Meanwhile AI is removing the junior work, such as grading and starter projects, through which people learned to judge, the first rung of the career ladder, just as demand for judgement grows: when a task completed with AI spans enough fields, no single person can judge all of it, and we conjecture that judging such work will increasingly depend on consulting people with depth in each field. Proposals for assessment reform call for structural change but assume that people able to certify and consult can be hired or redeployed, without asking where they will come from. We argue that the two problems can solve each other, and that the solution amounts to a third paradigm of university teaching. Teaching is organized around certification of understanding and consultation rather than the delivery or discovery of knowledge; lectures and inquiry remain as ways to learn. Students who have demonstrated competence take up certification and consultation posts under supervision. These posts supply the human time that certification requires, and the judging work itself trains those who hold them. Learners thus become evaluators, and universities rebuild the first rung, now leading toward the role of expert consultant that AI-assisted work increasingly needs. Teaching-assistant posts should move from grading to probing and consultation, so that credentials attest to what students can explain and defend. Personalized learning within a degree depends on certifying each student's own path. Probing can do this because it checks, face to face, whether a student holds the competences of a common framework, whatever work the student has done, and the same posts supply the time it takes.