OpenAI researcher: frontier models incl. Astra stay within 2x of GPT-4 compute depth
deanwball · x · 2026-09-02
DeepMind researcher Tomasz Korbak argues that the day a frontier lab trains a frontier-scale recurrent (or otherwise unmonitorable) LLM would be among the darkest of the current AI era, urging labs to coordinate on a commitment to never do so.
OpenAI's merettm pushed back on "confused reporting" fueling a race into unmonitorability: the computation-graph depth of today's frontier models, including Astra, is within a factor of two of GPT-4. He says OpenAI has preserved and leveraged chain-of-thought monitoring since its first reasoning models, as it reveals how alignment generalizes beyond the training distribution — but he admits the technique is fragile and trending negative for non-architectural reasons he'll detail soon, and strengthening it is a core research goal.
More from AGI Musings
- AI claims it solved 100 open problems but won't say which — mathematicians on notice — basedjensen · 2026-09-23
- Peter Diamandis: When AI and robots do the work, who owns the machines? — PeterDiamandis · 2026-09-23
- Hinton vs. LeCun reignites: did reasoning models vindicate his autoregression doubts? — OnlyBath9046 · 2026-09-23
- High school gone fully ChatGPT: teachers and students both outsourcing thinking — ZeroStateReflex · 2026-09-23
- The Lab That Learns: why two labs with the same AI get different results — CatAstro_Piyush · 2026-09-23
- OpenAI's Shyamal Anadkat: manufacturing sovereignty depends on discovery sovereignty — shyamalanadkat · 2026-09-23