"To make them is not to understand them." Emmanuel Ameisen ( @mlpowered ) has spent the last 3 years doing interpretability research at @AnthropicAI , looking inside LLMs like Claude to figure out what they're actually doing. As he puts it, models are "grown," not built. We discuss: 🔸 His discovery that models plan many tokens ahead 🔸 Why anthropomorphizing AI can actually be the right approach 🔸 What the "J-space" reveals about how models think 🔸 How interpretability connects to AI alignment and safety Listen no youtu.be/JHo-aiePjoM fpkab
← 返回归档
查看 X 原文帖子详情
查看 Markdown 内容
--- tweet_id: "2111814390521413660" author: "lennysan" published_at: "2026-10-18T13:39:20+00:00" source_url: "https://x.com/lennysan/status/2111814390521413660" --- "To make them is not to understand them." Emmanuel Ameisen ([@mlpowered](https://x.com/mlpowered)) has spent the last 3 years doing interpretability research at [@AnthropicAI](https://x.com/AnthropicAI), looking inside LLMs like Claude to figure out what they're actually doing. As he puts it, models are "grown," not built. We discuss: 🔸 His discovery that models plan many tokens ahead 🔸 Why anthropomorphizing AI can actually be the right approach 🔸 What the "J-space" reveals about how models think 🔸 How interpretability connects to AI alignment and safety Listen no[youtu.be/JHo-aiePjoM](https://youtu.be/JHo-aiePjoM)fpkab