PERSON DIRECTORY
Sholto Douglas
Sholto Douglas appears in 2 indexed conversations across Dwarkesh Podcast. This directory brings every appearance, source, TL;DR, digest, and transcript into one searchable feed.
Is RL + LLMs enough for AGI? — Sholto Douglas & Trenton Bricken
Dwarkesh PatelSholto DouglasTrenton Bricken
Sholto Douglas says RL in language models has finally worked, demonstrating expert-human reliability in competitive programming and math.Against white-collar salary TAM, he says a drop-in worker is very likely in two years, while 10M H100-equivalents today versus 100M by 2028 could make inference the bottleneck, with fab ramp and Taiwan key.
AMA: career advice given AGI, how I research ft. Sholto & Trenton
Dwarkesh PatelTrenton BrickenSholto Douglas
LLMs’ failure to make known cross-domain discoveries points to missing scaled RL and primitive memory, not a lack of stored knowledge.With engineers reporting 2-5x speedups, the advice is to stay near the frontier and compound leverage by managing increasingly capable AI teams, while hiring still depends on proactive referrals.