John Platt (Google): AI for Climate and Science
Oct 9, 2026
The Information Bottleneck
Why AI Still Can't See with Andrew Dai
About
Andrew Dai joins The Information Bottleneck to discuss language-model pre-training, next-token prediction, training-data quality and visual reasoning. The conversation examines counting and image understanding, world models and JEPA. Hosted by Ravid Shwartz-Ziv and Allen Roush.
Topics
language-model pre-trainingnext-token predictiontraining-data qualityvisual reasoningimage understandingcountingworld modelsJEPAAndrew Dai
More from this series
- Sara Hooker on the End of Static AISep 16, 2026 · 1 h 36 min
- AI Agents and The Golden Age of Asking Questions with Dimitris Papailiopoulos (MSR/UW-Madison)Jul 9, 2026 · 1 h 13 min
- Infrastructure for AI at Scale - With Benny Chen (Fireworks AI)Jun 24, 2026 · 1 h 6 min
- Nathan Lambert: Inside Post-Training and the Open Model FightAug 8, 2026 · 1 h 15 min
- World Models | John Langford (Microsoft AI Labs)Sep 5, 2026 · 1 h 6 min
More from The Information Bottleneck
All episodesYuandong Tian on Recursive Self-Improvement
Oct 5, 2026The Information BottleneckChris Manning: Language Is the Real Unlock for Intelligence
Sep 27, 2026The Information BottleneckCan AI Replace Doctors? Zachary Lipton on the Future of Healthcare and AI
Sep 21, 2026The Information Bottleneck
New episodes in Machine Learning
All episodesWhy All Models Learn the Same Thing with Phillip Isola (MIT)
Jul 2, 2026The Information BottleneckContinual Learning Is the Next Bottleneck | Rohan Anil (Core Automation )
Sep 10, 2026The Information BottleneckWhy Deep Learning Finally Works on Tables | Frank Hutter (Prior Labs)
Aug 24, 2026The Information BottleneckKaggle Grandmasters, Agent Skills, and Why Everyone Is Overfitting with Jean-Francois Puget (Nvidia)
Jul 13, 2026The Information Bottleneck