All times in this Schedule are Pacific Time (PT)

Monday, August 31, 2026

Morning Session

08:00-08:55
Check-in/Breakfast (hosted by IPAM)
08:55-09:00
Welcome and Opening Remarks
Session Chair Marina Melia
09:00-09:45
Been Kim (Google DeepMind)
Changing Shapes: Interpretability, AI, and the Future of Human Agency
10:00-10:15
Break
10:15-11:00
Jacob Hilton (Alignment Research Center)
Miniature mathematical mysteries: uninterpreted models with fewer than 1,500 parameters
11:15-11:30
Break
11:30-12:15
Paul Christiano (Alignment Research Center)
Explaining computational phenomena

Afternoon Session

12:30-14:30
Lunch (on your own)
Session Chair Been Kim
14:30-15:15
Marina Meila (University of Waterloo)
Counterfactual Explanations in the statistical limit
15:30-16:00
Break
16:00-16:45
Josh Batson (Anthropic)
Interpretability: From Art Towards Science
17:00-17:20
Lightning Poster Round
17:20-18:30
Poster Session & Reception (Hosted by IPAM)

Tuesday, September 1, 2026

Morning Session

08:00-09:00
Check-in/Breakfast (hosted by IPAM)
Session Chair Christopher Potts
09:00-09:45
Bin Yu (University of California, Berkeley (UC Berkeley))
Not All Interpretation Methods Are Created Equal: Principled Evaluation of AI Interpretability
10:00-10:15
Break
10:15-11:00
Max Tegmark (Massachusetts Institute of Technology)
Neural network interpretability: symmetry, geometry and formal verification
11:15-11:30
Break
11:30-12:15
Jacob Steinhardt (University of California, Berkeley (UC Berkeley))
Oversight Foundation Models

Afternoon Session

12:30-14:30
Lunch (on your own)
Session Chair Bin Yu
14:30-15:15
Ryan Cotterell (ETH Zurich)
Two views of a language model
15:30-16:00
Break
16:00-16:45
Christopher Potts (Stanford University)
Creating and detecting subliminal learning effects on demand

Wednesday, September 2, 2026

Morning Session

08:00-09:00
Check-in/Breakfast (hosted by IPAM)
Session Chair Terry Tao
09:00-09:45
Sam Eisenstat (Machine Intelligence Research Institute)
Concepts, information, and objectivity
10:00-10:15
Break
10:15-11:00
Yusu Wang (University of California, San Diego (UCSD))
Neural Network Generalization through an algorithmic lens
11:15-11:30
Break
11:30-12:15
Michael Winer (The Alignment Research Center)
Mechanistic Understanding of Multilayer Perceptrons

Afternoon Session

12:30-12:40
Group Photo
12:40-14:00
Lunch (on your own)
Session Chair Gitta Kutyniok
14:00-14:45
Elias Bareinboim (Columbia University)
Towards Causal AI: From Mechanism to Understanding
15:00-15:15
Break
15:15-16:00
Gal Mishne (University of California, San Diego (UCSD))
From Explanations to Mechanisms: Interpreting Computation in Graph Neural Networks
16:15-16:30
Break
16:30-17:15
Gyorgy Turan (University of Illinois at Chicago)
Error-correcting codes: a sandbox for learning, interpretability and optimization

Thursday, September 3, 2026

Morning Session

08:00-09:00
Check-in/Breakfast (hosted by IPAM)
Session Chair Elias Bareinboim
09:00-09:45
Gitta Kutyniok (Ludwig-Maximilians-Universität München)
From Mathematical Guarantees to Computational Limits of AI Interpretability
10:00-10:15
Break
10:15-11:00
Mahdi Soltanolkotabi (University of Southern California (USC))
Interpreting Generative Models for Better Steering: From Visual Generation to Verifiable Reasoning
11:15-11:30
Break
11:30-12:15
Dmitry Vaintrob (Principles of Intelligence, PIBBSS)
Statistical theory of learning sparse structure

Afternoon Session

12:30-14:30
Lunch (on your own)
Session Chair Thurs PM TBD
14:30-15:15
Jacob Tsimerman (University of Toronto)
Some new results in Open Source Game Theory
15:30-16:00
Break
16:00-17:00
Panel Discussion - Moderator: Been Kim, Panelists: Terry Tao, Chris Potts, Thomas Icard, Eunice Jun

Friday, September 4, 2026

Morning Session

08:00-09:00
Check-in/Breakfast (hosted by IPAM)
Session Chair Thomas Icard
09:00-09:45
Eunice Jun (University of California, Los Angeles (UCLA))
Imagining a new future for science
10:00-10:15
Break
10:15-11:00
Paul Riechers (Simplex)
The shape of beliefs and abstraction in neural networks
11:15-11:30
Break
11:30-12:15
Terence Tao (University of California, Los Angeles (UCLA))
Reflections on the workshop
Printable Version