- 1 Open-world evaluations for measuring frontier AI capabilities ↗
- 2 New Paper: Towards a science of AI agent reliability ↗
- 3 AI Won’t Automatically Make Legal Services Cheaper ↗
- 4 Fact checking Moravec's paradox ↗
- 5 A guide to understanding AI as normal technology ↗
- 6 Could AI slow science? ↗
- 7 AGI is not a milestone ↗
- 8 AI as Normal Technology ↗
- 9 Is AI progress slowing down? ↗
- 10 We Looked at 78 Election Deepfakes. Political Misinformation is not an AI Problem. ↗
- 11 Does the UK’s liver transplant matching algorithm systematically exclude younger patients? ↗
- 12 FAQ about the book and our writing process ↗
- 13 Can AI automate computational reproducibility? ↗
- 14 Start reading the AI Snake Oil book online ↗
- 15 AI companies are pivoting from creating gods to building products. Good. ↗
- 16 AI existential risk probabilities are too unreliable to inform policy ↗
- 17 New paper: AI agents that matter ↗
- 18 AI scaling myths ↗
- 19 Scientists should use AI as a tool, not an oracle ↗
- 20 AI leaderboards are no longer useful. It's time to switch to Pareto curves. ↗