Will continuing progress in AI capabilities result in every human dying by Jan 1st, 2030?
Due to M$ not being worth anything if the world ends, people have an incentive to bet "NO" on questions about AI killing everyone even if they believe the correct answer is probably YES. Pleading with people to just bet their beliefs on this important question doesn't seem like the best possible solution.
This market resolves N/A on Jan 1st, 2027. All trades on this market will be rolled back on Jan 1st, 2027. However, up until that point, any profit or loss you make on this market will be reflected in your current wealth; which means that purely profit-interested traders can make temporary profits on this market, and use them to fund other permanent bets that may be profitable; via correctly anticipating future shifts in prices among people who do bet their beliefs on this important question, buying low from them and selling high to them.
In principle there's still a distortionary effect if mana is worth less to you in 2026 within worlds that end by 2030. But since M$ are in any event only useful for charity (including charity that could try for last-ditch efforts to save us in doomed timelines) and not for personal partying in any timeline, the distortion should be less. I'm sure people will come up with galaxy-brained reasons not to bet their true beliefs in this market too, including trolls who come up with elaborate arguments for that in the comments just to be trolls; but the actual incentive distortion should be less.
The intent of this market is that the semantic question for people to bet their beliefs on, is about AI progress causally resulting in literally every human dying by 2030. I don't think it's particularly likely that this would happen by accident, or by the deliberate choice of humans staying in control; but eg terrorists purposefully using LLAMA 4 to build a supervirus would count if the last survivors died by Jan 1st, 2030.
Related market:
Related questions
@IsaacKing Clearly, since tips yields have gone a bit below -1% at the peak of the bond bubble
@AlexanderLeCampbell yes he does. look at other markets that are similar but didnt do that and they have very different predictions.
@BAUEREsaie You can hold any position to expiry to guarantee a zero payoff when it N/A’s. It is just a Keynesian beauty contest.
@aashiq Don't you think people winning money 2 hours before resolution won't close out their positions?
@BAUEREsaie N/A undoes all trades. Doesn't matter if you close out your position, it all gets rolled back.
@jack not all trades. I can get in profit M100, pull out of this market, and then turn this M100 in M300 in a few years before resolution. N/A resolution will leave me these M200
@BAUEREsaie They may. But then again, there will likely also be insane trading dynamics in the last 2 hours, and some users may not be able to keep their hands out of the cookie jar. This will then drive other users to prey on them and so forth and so on.
@Lavander That is correct, you can get temporary profits for the duration of the market, which is long. However, in reply to the comments about trading 2 hours before resolution - there is no point to that as those profits vanish on resolution.
@jack You may be able to open and close a position at a profit 2 hours before resolution, especially if Manifold has become a much larger and more liquid market by then. This regularly happens in ie physically settled commodity futures where the vast majority of the trading volume does not want to hold a position to expiry.
@aashiq Anyone who has a profit will want to close, so as to keep their profit. This creates an incentive to trade ahead of them, which may persist right up until market expiry.
@Joshua If I open a position and close it at a profit, do I not keep the profit even in an N/A event?
@Joshua So what happens if you close your position and then lose all your mana? Does it go negative?
@DavidBolin I think they're both unreliable, if in different ways. But it's useful to try different ways around the anthropic-bias-blindspot, and sort of triangulate.
My attempt: drawing a "curve" from 100 deaths in an AI disaster, up to 1bn deaths, but deliberately stopping there. We then infer the 8bn percent must be slightly lower than what we see for 1bn.
https://manifold.markets/ScroogeMcDuck/does-an-ai-disaster-kill-at-least-1-73bcb6a788ab
