Best AI papers explained

550 Episodes

1. Provably Learning from Language Feedback

   Published: 7/9/2025
2. Markets with Heterogeneous Agents: Dynamics and Survival of Bayesian vs. No-Regret Learners

   Published: 7/5/2025
3. Why Neural Network Can Discover Symbolic Structures with Gradient-based Training: An Algebraic and Geometric Foundation

   Published: 7/5/2025
4. Causal Abstraction with Lossy Representations

   Published: 7/4/2025
5. The Winner's Curse in Data-Driven Decisions

   Published: 7/4/2025
6. Embodied AI Agents: Modeling the World

   Published: 7/4/2025
7. Beyond Statistical Learning: Exact Learning Is Essential for General Intelligence

   Published: 7/4/2025
8. What Has a Foundation Model Found? Inductive Bias Reveals World Models

   Published: 7/4/2025
9. Language Bottleneck Models: A Framework for Interpretable Knowledge Tracing and Beyond

   Published: 7/3/2025
10. Learning to Explore: An In-Context Learning Approach for Pure Exploration

    Published: 7/3/2025
11. Human-AI Matching: The Limits of Algorithmic Search

    Published: 6/25/2025
12. Uncertainty Quantification Needs Reassessment for Large-language Model Agents

    Published: 6/25/2025
13. Bayesian Meta-Reasoning for Robust LLM Generalization

    Published: 6/25/2025
14. General Intelligence Requires Reward-based Pretraining

    Published: 6/25/2025
15. Deep Learning is Not So Mysterious or Different

    Published: 6/25/2025
16. AI Agents Need Authenticated Delegation

    Published: 6/25/2025
17. Probabilistic Modelling is Sufficient for Causal Inference

    Published: 6/25/2025
18. Not All Explanations for Deep Learning Phenomena Are Equally Valuable

    Published: 6/25/2025
19. e3: Learning to Explore Enables Extrapolation of Test-Time Compute for LLMs

    Published: 6/17/2025
20. Extrapolation by Association: Length Generalization Transfer in Transformers

    Published: 6/17/2025