Hi titotal. This website âpresents the AI Futures Model (Dec 2025 version), following up on the timelines and takeoff models we [AI Futures Project] published alongside AI 2027â. I am sure many would be interested in a post with your thoughts, myself included! As expected based on your analysis, the parameter defining âHow much easier/âharder each coding time horizon doubling getsâ is crucial. If I set it to 1 to make the time horizon increase exponentially with the effective training compute, as it arguably has trended recently, the automated coder only comes in 2042.
Iâm not super excited about revisiting the model, to be honest, but Iâll probably take a look at some point.
What Iâd really like to see, and what I havenât noticed from a quick look through the update, is some attempt to prove the validity of the models with reference to actual data. For example, I think METR comes off looking pretty good right now with their exponential model of horizon growth, which has held up for nearly a year post-publication now. The AI2027 modelâs prediction of superexponential growth has not. So I think they have to make a pretty strong case for why I should trust the new model.
It seems to me like the really important thing is interpreting what âMETR 80% time horizon goes to a yearâ, or whatever endpoint you have in mind actually means. Itâs important if that takes longer than AI2027 predicts, obviously, but it seems more crux-y to me whether getting to that point means transformative AI is near or not, since the difference between â3 years and 7 seven yearsâ say, while important seems less important to me than between âdefinitely in 7 yearsâ and âwho knows, could still be 20+ years awayâ.
Agreed, David. The post Whereâs my ten minute AGI? by Anson Ho discusses why METRâs task time horizon does not translate into as much automation as one may naively expect.
[...] if AIs are actually able to perform most tasks on 1-hour task horizons, why donât we see more real-world task automation? For example, most emails take less than an hour to write, but crafting emails remains an important part of the lives of billions of people every day.
Some of this could be due to people underusing AI systems,2 but in this post I want to focus on reasons that are more fundamental to the capabilities of AI systems. In particular, I think there are three such reasons that are the most important:
Tasks are very bundled together and hard to separate out.
While itâs hard to be quantitative about just how much each of these reasons matter, theyâre all strong enough to explain why many tasks with 1-hour or even 10-minute horizons remain unautomated.
Yeah, I am inclined to agree-for what my opinion is worth which on this topic is probably not that much-that there will be many things AIs canât do even once they have a METR 80% time-horizon of say 2 days. But I am less sure of that than I am of the meta-level point about this being an important crux.
Hi titotal. This website âpresents the AI Futures Model (Dec 2025 version), following up on the timelines and takeoff models we [AI Futures Project] published alongside AI 2027â. I am sure many would be interested in a post with your thoughts, myself included! As expected based on your analysis, the parameter defining âHow much easier/âharder each coding time horizon doubling getsâ is crucial. If I set it to 1 to make the time horizon increase exponentially with the effective training compute, as it arguably has trended recently, the automated coder only comes in 2042.
Iâm not super excited about revisiting the model, to be honest, but Iâll probably take a look at some point.
What Iâd really like to see, and what I havenât noticed from a quick look through the update, is some attempt to prove the validity of the models with reference to actual data. For example, I think METR comes off looking pretty good right now with their exponential model of horizon growth, which has held up for nearly a year post-publication now. The AI2027 modelâs prediction of superexponential growth has not. So I think they have to make a pretty strong case for why I should trust the new model.
It seems to me like the really important thing is interpreting what âMETR 80% time horizon goes to a yearâ, or whatever endpoint you have in mind actually means. Itâs important if that takes longer than AI2027 predicts, obviously, but it seems more crux-y to me whether getting to that point means transformative AI is near or not, since the difference between â3 years and 7 seven yearsâ say, while important seems less important to me than between âdefinitely in 7 yearsâ and âwho knows, could still be 20+ years awayâ.
Agreed, David. The post Whereâs my ten minute AGI? by Anson Ho discusses why METRâs task time horizon does not translate into as much automation as one may naively expect.
Yeah, I am inclined to agree-for what my opinion is worth which on this topic is probably not that much-that there will be many things AIs canât do even once they have a METR 80% time-horizon of say 2 days. But I am less sure of that than I am of the meta-level point about this being an important crux.