Projects
Here you can find the grants that fund our research and posters about our work. The public posters explain our studies for a general audience, and some let you take part yourself by solving the same puzzles we give to AI. For the full list of papers, see the Research page.
Grants
AI seems to be gaining human-like intelligence fast, but methods for measuring human intelligence don’t carry over to AI, and vice versa. This project develops methods to compare the intelligence of AI models and humans, now and in the future, and to inform how AI can be used safely in everyday life.
Children learn to solve analogies such as “body is to feet as tree is to …?” in new domains, while AI models often fail outside familiar territory. This project studies how children learn analogical reasoning, to explain why AI falls short and to help build more robust, trustworthy AI.
A theme group with Lucia Donatelli and Martha Lewis on how analogical reasoning develops in humans and whether machines can replicate it. Claire’s fellowship project traces the shift from associative to relational reasoning in children and AI models.
Posters for the public
Can AI generate creative output? Participate in our experiment!
Can AI answer Bongard Problems correctly? Solve some classic Bongard problems yourself!
Can AI do analogical reasoning? Participate in analogy-solving tasks using letterstrings!