“Our current cognitive concepts are simultaneously indispensable, yet inadequate; indispensable, because we need the richness of cognitive vocabulary to guide trust and decision-making; inadequate, because they encode anthropomorphic assumptions.”

That’s the guiding idea of “How We Need to Think about Artificial Cognition,” a project led by Matthieu Queloz, a philosopher at the University of Bern, that was just awarded a grant from the Swiss National Science Foundation of 1,619,334 Swiss francs, or just over $2 million.

Here’s a description of the project:

As artificial intelligence becomes increasingly integrated into critical domains, we face a fundamental question: to what extent should we conceptualize these systems in cognitive terms? This five-year project starts from the observation that our current cognitive concepts are simultaneously indispensable, yet inadequate; indispensable, because we need the richness of cognitive vocabulary to guide trust and decision-making; inadequate, because they encode anthropomorphic assumptions—e.g., of phenomenal experience and rational autonomy—that don’t neatly map onto machines. This creates a conceptual impasse.

The project’s guiding hypothesis is that resolving this tension requires conceptual adaptation. We must decouple the indispensable from the inadequate, guided both by a sense of what these concepts do for us in the first place and by a technical understanding of the internal mechanisms of these models. By working through three foundational case studies—concept formation, understanding, and reasoning—the project aims to reverse-engineer what we need these concepts to do for us and show how they can be adapted to artificial cognition. This will clarify what we gain and risk by extending cognitive language to AI; it will help bridge the definitional disconnect between philosophy and machine learning; and it will establish foundations for conceptualizing artificial cognition, proposing ways to talk about AI that are rigorous, yet realistic and helpful in navigating the complexities of this new era.

The project runs from 2027-2032. Some of the funding will support new postdoctoral fellows and PhD students.

Queloz’s project is the only one led by a philosopher that was funded in this recent round of grant awards. You can learn about the 40 other funded projects here.