AI Progress Is Smooth Until It Crosses a Threshold @LennysPodcast
AI Progress Is Smooth Until It Crosses a Threshold  @LennysPodcast
Uploaded October 2026 | Updated November 2026, 8 hours ago
AI progress looks like sudden jumps from the outside. From the inside, it's a smooth line that crosses a threshold.

Emmanuel Ameisen, Interpretability Researcher at Anthropic, on why 80% accuracy is useless, 99% is useful, and the model that got Claude writing most of his code.
AI Progress Is Smooth Until It Crosses a ThresholdThe best product leaders arent the ones with the most ideas.The AI jobpocalypse isnt realNetflixs CPTO on whether teams need specialistsWhy companies need to change how they think about hiringHire for Unlearning Rate, Not Learning RateAutonomy Only Works If You Own the MistakesThe three memes of product developmentGreat products tell a storyAI still needs humansWhy Self-Sufficiency Is Costing YouThe rise of taste, human authenticity and judgment in an AI world | Adam Mosseri (Head of IG)
Lennys Podcast |

AI Progress Is Smooth Until It Crosses a Threshold

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER