Shashwat Goel(@ShashwatGoel7):I have to say :) the events were discrete (but yes, noisy!) I don't think the inference from FutureSim should be LLMs suck. The environment is long horizon and relatively open-ended, GPT 5.5 still does surprisingly well. And we know RL makes them better

Shashwat Goel

@ShashwatGoel7

Training AI for Decision Making Past work: Training AI Co-scientists, ΔBelief-RL, Measuring Long Horizon Execution

参加 June 2020

2.3K フォロー中 4.1K ファン

Shashwat Goel@ShashwatGoel7

2026.05.18 07:43

I have to say :) the events were discrete (but yes, noisy!) I don't think the inference from FutureSim should be LLMs suck. The environment is long horizon and relatively open-ended, GPT 5.5 still does surprisingly well. And we know RL makes them better