HackerTube 中文?

Dario Amodei: Anthropic CEO on Claude, AGI & the Future of AI & Humanity | Lex Fridman Podcast #452

Lex FridmanYouTube ↗

Further reading 1

How much does a successful demonstration establish? →A program gives the expected output. Between “it did” and “it will” lie the test distribution, formal assumptions and deployment context. Those boundaries are easy to miss under pressure to ship.This video helps answer · Does surviving an occlusion demonstrate a 3-D model?

Caption notes

The interview discusses retaining earlier evaluations and adding new ones alongside direct interaction. A fixed suite helps compare releases, while an acting system encounters situations shaped by its own earlier actions.

Video or transcript source ↗
Original passages · 2
there’s no substitute for a human interacting with it
Dario Amodei: Anthropic CEO on Claude, AGI & the Future of AI & Humanity | Lex Fridman Podcast #452 · ▶ 52:34–52:56 · passage interval
we’re still very early in our ability to study evaluations, particularly for dynamic systems acting in the world
Dario Amodei: Anthropic CEO on Claude, AGI & the Future of AI & Humanity | Lex Fridman Podcast #452 · ▶ 106:07–107:14 · passage interval

Hacker News comments

View original on HN ↗