A problem with long-term markets like https://manifold.markets/ScottAlexander/in-2028-will-an-ai-be-able-to-gener?r=SGVkU2hvY2s is that many people bet to maximize their utility given market outcomes (e.g. in a YES-resolution-world, Mana will probably be much less valuable), instead of betting their true beliefs.

If we get a friendly ASI, will it commensurately compensate people for honest, accurate, selfless, and prosocial betting behavior?