Esitelmän otsikko:
“When answers are cheap: teaching, learning, and assessment with Project Ethelh”
Abstrakti:
AI can now solve many of the problems we set our students. That does not make learning obsolete — but it does challenge us to reconsider what we teach, how we assess it, and where AI belongs in the process. The question is no longer simply what these models can do, but how universities can put their capabilities to educationally meaningful and accountable use. Drawing on empirical studies and practical deployments at ETH Zurich, I introduce Project Ethel, an open-source ecosystem supporting course-specific chatbots, feedback on handwritten work, accessible learning materials, and the scoring of open-ended exams. Recent experience with approximately 2,000 exams across six courses illustrates both the potential and the practical demands of AI-supported assessment: transparent scoring, student vetoes, independent human rescoring, and instructor responsibility for final grades.
Throughout, I draw a clear distinction between using AI to support learning and using it to judge learners. Students need an educational “backstage": a space to ask naïve questions, explore unfinished ideas, and make mistakes without every interaction becoming assessment evidence. The aim is not to replace teachers, isolate students with artificial tutors, or automate their thinking. It is to strengthen research-based teaching, make meaningful feedback and open-ended assessment feasible at scale, and preserve the human relationships and student agency on which learning depends.
Tietoa luennoitsijasta:
Gerd Kortemeyer on ETH Zürichin rehtoraatin jäsen ja ETH AI Centerin asiantuntijajäsen. Hän on myös Michigan State Universityn emeritustutkimusprofessori. Hänen tutkimuksensa keskittyy teknologiavälitteiseen STEM-aineiden oppimiseen; tällä hetkellä hän edistää tekoälypohjaisten työkalujen ja työnkulkujen tutkimusta ja kehitystä opetusta, oppimista ja arviointia varten.