2026 GAC 2

Do World Models Emerge in Neural Networks Trained for Prediction?

Eivinas Butkus, Columbia University

Alexa Tartaglini, Stanford University

Josh Ying, Columbia University

Sonia Joseph, Meta

Brian Cheung, MIT/UCSF

Jennifer Hu, John Hopkins University

Niko Kriegeskorte, Columbia University

Ellen Su, New York University

Solim LeGris, New York University

John Morrison, Barnard College

Ilker Yildirim, Yale University

Sam Gershman, Harvard University

Brenden Lake, Princeton University

Short description

This GAC asks whether world models emerge in state-of-the-art neural networks trained using standard prediction objectives, such as self-supervised learning. Our goals are to (1) clarify definitional controversies, (2) survey methodological approaches, (3) review recent evidence, and (4) design new experiments that could shift credences. The GAC team is composed of cognitive scientists, neuroscientists, AI researchers, and philosophers.

GAC kickoff workshop, August 4th at CCN

Schedule of events

TBA

Team

Comment on this GAC proposal!

Instructions for CCN community members: 

1. Download the proposal PDF using the button above. 

2. Use the Disqus box below to leave comments and reviews, and discuss the submission. You'll need to create a Disqus account or log into Disqus using a google account or similar in order to leave a comment/review.