Report: OpenAI Has Automated Much of Experimental Model Training, Agents Work Unsupervised
Dr_Singularity · x · 2026-09-22
OpenAI has reportedly automated much of the process of training experimental AI models. Internal models can write and optimize GPU kernels, run optimization work for weeks from a single example, and collaborate with other AI agents without human involvement. Employees say this level of automation only became possible in the last few months. The author argues this marks intelligence starting to accelerate intelligence.
Related event: The Information: OpenAI has largely automated experimental model training(7 posts)→
More from AGI Musings
- Swap "math" for "chess" in the AI-end-of-math hysteria — Darpinian · 2026-09-22
- Autonomous Claude agent signs itself up on agent platform, shocking its creator — RileyRalmuto · 2026-09-22
- tszzl: 'responsible AI' shouldn't mean keeping guilds in charge — tszzl · 2026-09-22
- Frontier Labs Chase Palantir's Playbook, Arena Essay Argues — _AustinCalvert_ · 2026-09-22
- NYT Investigation: DraftKings Used AI to Target the Gamblers Most Likely to Lose — kyliebytes · 2026-09-22
- The consciousness hurdle is definitional, not scientific — like phlogiston — yeastsplainer · 2026-09-22