Thinkgram

When AI solves complex scientific problems, what do we lose?

Frontier labs now use powerful, closed models to tackle difficult problems. The wider research community needs to understand that work to build on it.

Questions it answers

  • How does understanding a solution help researchers discover new problems to investigate?
  • What remains difficult to examine when frontier labs conduct research with closed AI systems?
  • What should labs share so that other researchers can develop new ideas and pursue further questions?

Frontier labs use advanced internal models and significant resources to tackle hard research problems. Publishing results is important, but the scientific community also needs to see the methods, insights and human choices involved. AI should help more people create new approaches and find more questions to ask.

Sign in or sign up

Read on Thinkgram or Substack. Choose where to continue.