LessWrong (30+ Karma) “Bridge Thinking and Wall Thinking” by Jay Bailey
Mar 15, 2026
Jay Bailey, author and AI-safety commentator, presents two simple frames for thinking about AI risk: wall thinking (incremental bricks) and bridge thinking (big investments that must be completed to work). He contrasts standards-as-wall approaches with treaty-style bridge proposals. Short, clear examples and lively contrasts drive the conversation.
AI Snips
Chapters
Transcript
Episode notes
Wall Versus Bridge Framing
- AI safety debates split into two frames: wall thinking and bridge thinking.
- Jay Bailey defines wall thinking as incremental bricks where each addition is helpful, illustrated by Chris Olah's image and “eating marginal probability.”
Bridge Thinking Requires Threshold Outcomes
- Bridge thinking demands a threshold outcome before anything is useful, so partial progress is largely worthless.
- Jay Bailey uses the bridge metaphor to explain efforts that need a full solution before improvements matter, like treaties or minimum sufficient policies.
Prioritize Standards Work As Incremental Safety
- Build standards and evaluation paradigms as incremental, broadly useful contributions under a wall frame.
- Jay Bailey references his Inspect Evals standards work as pointing the field in a useful direction even without a full world-saving story.