Lately I’ve been most concerned about coordination for governance, and having the ability to nicely merge values from lots of people to determine what policies a democratic government should take. I think current methods of voting leave a lot to be desired, and it should be possible to use AI to help people build personal models of their preferences that can be diffed/merged with other preferences, enabling easy creation of coalitions or identifying points of agreement. I think Andrey Tang’s post on Pluralityis pretty useful for getting an idea of how similar things could work.
There are a lot of other dimensions for coordination though, getting accurate information from distorted sources, figuring out how to align representatives, better voting, etc. Mechanism design and social choice theory are good places to look for academic research on this stuff.
Lately I’ve been most concerned about coordination for governance, and having the ability to nicely merge values from lots of people to determine what policies a democratic government should take. I think current methods of voting leave a lot to be desired, and it should be possible to use AI to help people build personal models of their preferences that can be diffed/merged with other preferences, enabling easy creation of coalitions or identifying points of agreement. I think Andrey Tang’s post on Plurality is pretty useful for getting an idea of how similar things could work.
There are a lot of other dimensions for coordination though, getting accurate information from distorted sources, figuring out how to align representatives, better voting, etc. Mechanism design and social choice theory are good places to look for academic research on this stuff.