2027 is representative of a median (50%) forecast, and there are of course other uncertainties aside from the timeline itself. So 30% seems a bit overpriced even if the scenario is directionally correct.

I’m conflicted between:

1. This sounds extremely outlandish and science fiction when you read all the conclusions (post-scarcity society, solving medicine/nuclear fusion/gene editing, datacenter of geniuses

2. It’s done by some of the smartest people I’ve read who make truth their number one goal, and have thought for a long time about what’s the most true possible reality

This is fantastic science fiction.

The heuristics "trust Eli Lifland" and "never trust a plan that requires more than three different things to go right" are currently battling for dominance in my head

@MingCat Eli AND Daniel (plus Scott thrown in for free) is definitely making me take this super seriously. Which is awkward because I have a lot of bets implying AGI by ~2027/2028 is significantly less than 30% likely. XD

😆 great comment.

@MingCat Do note that Eli's median timelines are a bit longer than the scenario: see the first graph in https://ai-2027.com/research/timelines-forecast. (and I agree with him fwiw).

Timelines Forecast — AI 2027

AI 2027 predicts AIs with superhuman coding ability in 2027. This supplement details our model for forecasting when the Superhuman Coder milestone will be achieved, building off METR's time horizon work, and justifies why we think it’s plausible it is achieved by 2027.

@MingCat note that January 2027 is still early/normal, lots of the contingent things happen later in the scenario.

It might be better to break down the predictions and have markets for different claims.

@ConorHartin I think most of those already exist in some form or another