Dario Amodei — Anthropic CEO on Claude, AGI & the Future of Civilization

Dwarkesh Patel's marathon interview with Dario Amodei on Anthropic's scaling roadmap, mechanistic interpretability, Claude's economic impact and the geopolitics of frontier AI.

The CEO of Anthropic explains why AI is getting so much smarter so quickly, what could go wrong, and why he thinks the next few years matter more than any before.

1. 1Scaling laws still hold — bigger, better data and compute keep unlocking capability jumps.
2. 2Interpretability is Anthropic's core safety bet: understand the model, don't just test it.
3. 3Powerful AI could compress decades of biomedical progress into years.
4. 4Export controls on advanced chips are one of the highest-leverage policy tools.
5. 5Alignment and commercial success are complementary, not opposed.

• company
  Anthropic
• model
  Claude