Castuner

Interconnects

Nathan Lambert

Audio essays about the latest developments in AI and interviews with leading scientists in the field. Breaking the hype, understanding what's under the hood, and telling stories. www.interconnects.ai

Poddens webbplats

Lyssna i din app

Avsnitt

169 avsnitt

  1. Was I wrong about open models & cyber risks? 9:47
  2. Artificial Analysis v5 and the benchmarker's dilemma 10:38
  3. Podcast: Kimi K3.1 Pro, the new cyber reality, and Muse musings 46:42
  4. Meta shakes up the frontiers, open and closed 11:44
  5. The Cyber Risk Discourse is Broken 9:12
  6. Debating RSI, the US-China Gap, and Jaggedness with JS Denain of Epoch AI 1:05:29
  7. The current balance of power in open models 17:59
  8. Why I still haven’t bought into true RSI 9:32
  9. One resignation turned the embers of AI fear into a wildfire 10:59
  10. When will average people feel AI’s impact? 8:31
  11. I wrote an AI textbook — how long until AI can do it better? 12:32
  12. Open models recap: more on Kimi K3, Qwen 3.8, Xi's WAIC speech, distillation, the open-closed gap, and what's next 49:05
  13. Kimi K3: The open-weights escalation 20:03
  14. GLM-5.2 is the step change for open agents 9:27
  15. Banning Open Source AI Would Be A Mistake 6:50
  16. State of the blog, mid-2026 6:19
  17. Frontier post-training recipe review with Finbarr Timbers 56:36
  18. Claude Fable 5 and new AI safety fables 12:11
  19. Farewell Ai2 15:51
  20. Open and closed models are on different exponentials 7:21
  21. Some ideas for what comes next, May 2026 9:37
  22. Notes from inside China's AI labs 16:35
  23. The distillation panic 8:52
  24. My bets on open models, mid-2026 6:57
  25. The inevitable need for an open model consortium 5:45
  26. Claude Mythos and misguided open-weight fearmongering 8:36
  27. Gemma 4 and what makes an open model succeed 8:55
  28. Lossy self-improvement 13:23
  29. GPT 5.4 is a big step for Codex 6:49
  30. What comes next with open models 18:08
  31. Dean Ball on open models and government control 35:36
  32. Olmo Hybrid and future LLM architectures 11:21
  33. How much does distillation really matter for Chinese LLMs? 11:20
  34. Opus 4.6, Codex 5.3, and the post-benchmark era 8:09
  35. Why Nvidia builds open models with Bryan Catanzaro 1:07:42
  36. Thoughts on the job market in the age of LLMs 10:41
  37. Arcee AI goes all-in on open models built in the U.S. 1:12:15
  38. Get Good at Agents 5:05
  39. Use multiple models 7:12
  40. Claude Code Hits Different 4:58
  41. Open models: Hot or Not with Nathan Lambert & Florian Brand 37:36
  42. New Talk: Building Olmo 3 Think 1:02:22
  43. Olmo 3: America’s truly open reasoning models 10:57
  44. Why AI writing is mid 8:28
  45. Interview: Ant Group's open model ambitions 1:17:49
  46. 5 Thoughts on Kimi K2 Thinking 7:37
  47. Burning out 10:09
  48. How to scale RL 13:01
  49. The State of Open Models 47:04
  50. Thoughts on The Curve 11:58