How Game Theory Can Make AI More Reliable

Researchers are drawing on ideas from game theory to improve large language models and make them more correct, efficient, and consistent.

4 comments

Nope, not playing Card Wars with evil robot Jake.
that's so cool, llms seem like they'll be more useful in a couple years
- Indeed, if people can figure out how to get LLMs to produce stable outputs and create ways to verify them, then this tech would be a lot more powerful.
  
  We just need a checker AI to check the main AI and then maybe another checker checker AI to check the checker AI... All the way down.