Best AI papers explained

525 Episodes

1. Why in-context learning models are good few-shot learners?

   Published: 17/06/2025
2. Take Caution in Using LLMs as Human Surrogates: Scylla Ex Machina∗

   Published: 14/06/2025
3. The Logic of Machines: The AI Reasoning Debate

   Published: 12/06/2025
4. Layer by Layer: Uncovering Hidden Representations in Language Models

   Published: 12/06/2025
5. Causal Attribution Analysis for Continuous Outcomes

   Published: 12/06/2025
6. Training a Generally Curious Agent

   Published: 12/06/2025
7. Estimation of Treatment Effects Under Nonstationarity via Truncated Difference-in-Q’s

   Published: 12/06/2025
8. Strategy Coopetition Explains the Emergence and Transience of In-Context Learning

   Published: 12/06/2025
9. Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

   Published: 11/06/2025
10. Agentic Supernet for Multi-agent Architecture Search

    Published: 11/06/2025
11. Sample Complexity and Representation Ability of Test-time Scaling Paradigms

    Published: 11/06/2025
12. Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators

    Published: 10/06/2025
13. LLMs Get Lost In Multi-Turn Conversation

    Published: 9/06/2025
14. PromptPex: Automatic Test Generation for Prompts

    Published: 8/06/2025
15. General Agents Need World Models

    Published: 8/06/2025
16. The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models

    Published: 7/06/2025
17. Decisions With Algorithms

    Published: 7/06/2025
18. Adapting, fast and slow: Causal Approach to Few-Shot Sequence Learning

    Published: 6/06/2025
19. Conformal Arbitrage for LLM Objective Balancing

    Published: 6/06/2025
20. Simulation-Based Inference for Adaptive Experiments

    Published: 6/06/2025