Tom Kouwenhoven: Can AI ever truly understand us?

In this episode postdoc in Hybrid Intelligence Centre, Tom Kouwenhoven tells all about his research which focusses on how humans and machines learn to communicate. From chatbot misunderstandings to experiments where people and AI invent entirely new languages, Tom shows how collaboration can work even when both sides interpret information differently.

The conversation also dives into visual intuition, like the Bouba–Kiki effect, and why shared human preferences matter when designing future AI systems. It provides an honest view of how far human–machine communication has come and the challenges that continue to shape its future.

Idea by Dimitra Kouimtzidou

Research, planning & coordination by Marcel Tichelaar & Dimitra Kouimtzidou

Interview, production & editing by Michiel van Poelgeest