Structure and Guarantees

What kinds of AI systems should we build? A perspective mixing mathematical guarantees, scalable engineering, and evolutionary dynamics.

By Adam Chlipala

· Launched 5 months ago

Discussion Points

• Why Software Requirements Get Easier in an AI Economy

• Abstraction boundaries pay off

  Date: Jun 23
  Author: Adam Chlipala

• Simpler User Interfaces in an AI Future

• Why AI will have an easier time keeping itself happy

  Date: Jun 16
  Author: Adam Chlipala

• Subversion-Resistance for Free from Formal Verification

• Why we don't need to worry about enumerating possible attacks

  Date: Jun 9
  Author: Adam Chlipala

• Abstraction Boundaries and Bubbles of Legibility

• Encapsulating intelligence away from avoidable complexity

  Date: Jun 2
  Author: Adam Chlipala

• Simplifying Alignment by Expanding Scope

• A counterintuitive advantage for the good guys

  Date: May 26
  Author: Adam Chlipala

• Intelligence Depends on Organizing Computation Correctly and Efficiently

• A unifying outline of topics covered so far and where the blog is headed

  Date: May 19
  Author: Adam Chlipala

• Signaling and Perverse Adoption of Expensive AI

• Watch out for hard problems and expensive solutions

  Date: May 12
  Author: Adam Chlipala

• Codesign for Legibility (to AI and Everyone Else)

• Sometimes all it takes is changing the world.

  Date: May 5
  Author: Adam Chlipala