OpenAI's AGI plan: build an automated AI researcher, then iterate on alignment with it
trevposts · x · 2026-09-22
Per OpenAI's own post on standards, its plan for AGI is to "build an automated AI researcher" and "iterate with it on the alignment problem," with the human researcher's role TBD. Developer Trevor notes this has been OpenAI's plan for years — and, as far as he knows, also Anthropic's, xAI's and Google's. Effectively: recursive self-improvement, ask the AIs to make it safe, figure out where humans fit. He calls on labs to publish a better, more detailed plan.
Related event: OpenAI Plans Automated AI Researcher to Solve Alignment, Human Role TBD(2 posts)→
More from AGI Musings
- Agents are becoming the internet's primary consumers, reshaping 20 years of Google-mediated traffic — illscience · 2026-09-22
- Ethan Mollick: AI firms can't just replace doctors and lawyers, must negotiate their role — emollick · 2026-09-22
- Agents are becoming the internet's primary consumers, reshaping traffic economics — illscience · 2026-09-22
- Hilton gets 12,000 applications for 72 internships as AI fuels application surge — Polymarket · 2026-09-22
- Taking AI risk seriously doesn't require being an EA, argues rSanti97 — nwilliams030 · 2026-09-22
- AI is ending the Enlightenment Age by displacing humans as the sole source of discovery — notmisha · 2026-09-22