The Untraining AI research group imagines, practices, and discusses alternative approaches (whether new or long overlooked) to what is commonly termed Artificial Intelligence.

We seek to untrain AI through:

• Rethinking how AI is developed, including its data sets and algorithms.
• Building situated AI alternatives, from installations to tools that let people use and understand algorithms based on alternative values and assumptions.
• Teaching AI differently, developing critical AI literacies with practitioners and the public.

Our research is experimental and practice-based. We untangle complex systems from the inside, dismantling and reverse-engineering them in order to build something else.

We work across critical platform and data studies, artistic research, institutional critique, and critical making.

Untraining AI is a collaboration between Aalto University (Finland) and the University of Amsterdam (Netherlands). In our first two-year plan (2026–2028) we are hosting a programme of events including a public series for works in progress.