Anthropic Researcher: LLMs Are Grown, Not Built, and Plan Ahead

Anthropic interpretability researcher Emmanuel Ameisen told Lenny's Podcast that LLMs are "grown rather than built" and plan many tokens ahead, such as deciding rhyme words before writing a poem.

2026-10-18 ~ 2026-10-19 · 3 related posts

• Anthropic interpretability researcher: LLMs are 'grown' not built, and they plan many tokens ahead — lennysan · 2026-10-18
• Anthropic interpretability researcher: models plan rhymes before writing a line — lennysan · 2026-10-19

1 near-duplicate retellings: lennysan