Reading paths

Start here

Four guided routes through the archive, depending on what you want to learn.

Updated

Pick the path that matches your background. Each takes about an hour or less.

Path 1: New to AI

  1. About Dario Amodei for who he is.
  2. Anthropic explained for the company.
  3. Machines of Loving Grace for the optimistic vision.
  4. Glossary whenever a term is unfamiliar.

Path 2: Interested in AI safety

  1. Key ideas for the overview.
  2. Constitutional AI and the Responsible Scaling Policy.
  3. The Urgency of Interpretability.
  4. The original papers on the reading list, beginning with Concrete Problems in AI Safety.

Path 3: Technical readers

  1. Scaling laws and the underlying paper.
  2. Deep RL from human preferences, then Constitutional AI.
  3. Anthropic's interpretability work on Transformer Circuits.

Path 4: Policy and strategy

  1. On DeepSeek and Export Controls.
  2. Responsible Scaling Policy and its criticisms.
  3. Timeline to place events in order.

Short on time? Read the FAQ.