Interconnects

96 Épisodes

1. State of play of AI progress (and related brakes on an intelligence explosion)

   Publié: 30/04/2025
2. Transparency and (shifting) priority stacks

   Publié: 28/04/2025
3. OpenAI's o3: Over-optimization is back and weirder than ever

   Publié: 19/04/2025
4. OpenAI's GPT-4.1 and separating the API from ChatGPT

   Publié: 14/04/2025
5. Llama 4: Did Meta just push the panic button?

   Publié: 07/04/2025
6. RL backlog: OpenAI's many RLs, clarifying distillation, and latent reasoning

   Publié: 05/04/2025
7. Gemini 2.5 Pro and Google's second chance with AI

   Publié: 26/03/2025
8. Managing frontier model training organizations (or teams)

   Publié: 19/03/2025
9. Gemma 3, OLMo 2 32B, and the growing potential of open-source AI

   Publié: 13/03/2025
10. Interviewing Eugene Vinitsky on self-play for self-driving and what else people do with RL

    Publié: 12/03/2025
11. Elicitation, the simplest way to understand post-training

    Publié: 10/03/2025
12. Where inference-time scaling pushes the market for AI companies

    Publié: 05/03/2025
13. GPT-4.5: "Not a frontier model"?

    Publié: 28/02/2025
14. Character training: Understanding and crafting a language model's personality

    Publié: 26/02/2025
15. Claude 3.7 thonks and what's next for inference-time scaling

    Publié: 24/02/2025
16. Grok 3 and an accelerating AI roadmap

    Publié: 18/02/2025
17. An unexpected RL Renaissance

    Publié: 13/02/2025
18. Deep Research, information vs. insight, and the nature of science

    Publié: 12/02/2025
19. Making the U.S. the home for open-source AI

    Publié: 05/02/2025
20. Why reasoning models will generalize

    Publié: 28/01/2025