8:45-9:00am: Welcome
9:00-9:40am: Spotlight talks
On the Emergence of "Useless" Features in Next Token Predictors (Mark Rofin, Jalal Naghiyev, Michael Hahn)
Leveraging the Sequential Nature of Language for Interpretability (Usha Bhalla, Alex Oesterling, Claudio Mayrink Verdun, Flavio Calmon, Himabindu Lakkaraju)
World Models and Consistent Mistakes in LLMs (Christopher Wolfram, Aaron Schein)
Tracking World States with Language Models: State-Based Evaluation Using Chess (Romain Harang, Jason Naradowsky, Yaswitha Gujju, Yusuke Miyao)
Measuring Belief Updates in Curious Agents (Joschka Strüber, Ilze Amanda Auzina, Shashwat Goel, Susanne Keller, Jonas Geiping, Ameya Prabhu, Matthias Bethge)
9:40-10:00am: Coffee break
10:00-10:40am: Invited talk: Naomi Saphra (Harvard)
And Nothing Between: Using Categorical Differences to Understand and Predict Model Behavior
10:40-11:20am: Invited talk: Shiry Ginosar (TTIC)
What Do Vision and Vision-Language Models Really Know About the World?
11:20am-12:00pm: Invited talk: Jacob Andreas (MIT)
Language Models as World Models?
12:00-1pm: Lunch break
1:00-1:40pm: Invited talk: Shirley Ho (NYU/Flatiron)
Polymathic AI: Building Scientific Foundation Models
1:40-2:20pm: Invited talk: Sendhil Mullainathan (MIT)
Testing for Understanding Requires First Defining It
2:20-3:20pm: Panel Discussion
Jacob Andreas, Jon Kleinberg, Mengye Ren, Alane Suhr (moderated by Keyon Vafa)
3:20-3:45pm: Coffee break
3:45-5:00pm: Poster session
5:00-5:15pm: Concluding remarks
6:00-9:00pm: Social Event (Optional; Please Register Here)