Workshop on Information Theory, AI and Language

Workshop on Information Theory, AI and Language 

February 8 - 12, 2027
 

Language is one of the central functions of cognition and was long thought to be uniquely human. Recent advances in artificial intelligence, however, have challenged this view in a profound way. Systems trained to predict words now exhibit emergent capabilities often associated with intelligence, including logical inference, abstraction and mathematical reasoning. For the first time, we possess fully functional statistical models of natural language, yet still lack a principled understanding of why they work, what structures they capture and how intelligent functions emerge from them.

This moment presents a unique scientific opportunity. By achieving a mechanistic understanding of language and language models, we may uncover new principles that govern natural language, the emergence of intelligent functions and perhaps even cognition itself.  Such an effort has the potential to bridge multiple levels of description, from neural computation and learning dynamics to linguistic structure and human behavior.

This workshop seeks to bring together physicists, neuroscientists, mathematicians, computer scientists, cognitive scientists and linguists in a concerted effort to develop first-principles and phenomenological models of language and language models. Success in this endeavor will not only deepen our understanding of natural language but also advance the frontiers of artificial intelligence, offering insights into the nature of intelligence and cognition.

Program Organizers:
Uri Hasson, Princeton University
Sebastian Seung, Princeton University and Institute for Advanced Study
Misha Tsodyks, Institute for Advanced Study
Weishun Zhong, Institute for Advanced Study
 

Confirmed Participants:
Yasaman Bahri, Google DeepMind
David Beniaguev, The Hebrew University
Robert Berwick, MIT
Jonathan Cohen, Princeton University
Lukasz Debowski, Polish Academy of Sciences
Michael DeWeese, University of California, Berkeley
Surya Ganguli, Stanford University
Stratis Gavves, University of Amsterdam
Gregory Falkovich, Weizmann Institute of Science
George Karniadakis, Brown University
Roger Levy, MIT
Gary Lupyan, University of Wisconsin
Matilde Marcolli, Caltech
Jay McClelland, Stanford University
Cory Shain, Stanford University
Haim Sompolinsky, Harvard University
Alessandro Treves, International School for Advanced Studies, Trieste
Matthieu Wyart, Johns Hopkins University
Noga Zaslavsky, New York University
 

This workshop will be held in person only.  

Contact Audrey Smerkanich, [email protected], with any inquiries.