The module was developend by Marc Cleiren (FSW) and Ingrid Samset (FGGA), at the request of the Asian Development Bank Institute and is now widely available.

About the course

The first seminar, Human Ways of Relating to AI, addresses the seductive aspect of general-purpose models which promise the easy completion of complex tasks. Fortunately, we humans are also inherently curious; a curiosity that comes with a more productive learning attitude grounded in our need for autonomy, competence, and healthy relationships. AI can support this ‘learning mindset’ in unprecedented ways.

The second seminar, A New Pedagogy for the Age of AI, builds on this by presenting teacher and student experiences that show how working with curated AI agents enhances learning processes, instead of undermining them.

Taking AI into our own hands: Rebuilding education

Over the past years, growing student use of general-purpose AI has led to widespread concern among educators. What this may overlook is that working with AI can improve human learning processes. Marc Cleiren (Associate Professor, FSW): 'AI has confronted us with the fact that we have for too long focused on the testing of knowledge, often erroneously assuming that output assessment stimulates knowledge acquisition.'

Chances and Risks

Since 2016, Cleiren has actively explored and taught about the potential and risks of AI in academia and society. In recent years, he has seen the attitude to AI-use change and also sour (see also this article - in Dutch). ‘Many teachers feel frustrated in the face of the growing capability of AI to produce high-level outputs: the kind of work we always expected our students to make. But pressure to deliver ‘learning products’ instead of instilling a learning mindset has pushed many students into taking shortcuts instead of acquiring knowledge: “I have four essays to deliver by the end of the week. Why not use AI to help with three of them?”’

New approach

About the teacher side, he adds: 'We have for too long been holding the wrong end of the stick. Our students want to learn: that is why they joined the university in the first place. It is the learning behaviours that need stimulation, reward, and testing, instead of the learning products. Acclaimed pedagogical research gives us clear insights into what is needed for that.'

Successful hands-on implementation

In recent years, Marc Cleiren and Ingrid Samset (FGGA) have worked on didactical revision of curriculums, using knowledge from well-established pedagogical approaches to shape AI into various helpful learning assistants, rather than general-purpose output-generators. Cleiren coined the shorthand PAIA for this type of AI agent: Pedagogical AI Assistant. Instead of AI solving the intellectual puzzles a student faces, a PAIA scaffolds and stimulates the student’s interest, engagement, and skills to help them integrate the knowledge rather than simply reproduce it.

Seasoned tutor

PAIAs act as agents that respond to students’ questions and ideas in a similar way as a seasoned human tutor or teacher would. They are designed to apply the appropriate level of friction that is needed for learning in different phases of the process. Many of the PAIAs work from our own, safe university LUCA platform, that protects privacy and supports more sustainable AI use.

Initiate and sustain

Ingrid Samset, Assistant Professor at LUC and a new fellow of the Leiden Teachers Academy, has designed and tested various PAIAs in her teaching. She experimented by employing them in regular undergraduate courses, eliciting student feedback along the way: 'We are applying well established research on the preconditions for students to initiate and sustain learning. That is what I build on when creating my agents.'