OpenAI Updates Its AI Research Acceleration Report: Agents Complete Ever-Longer Tasks

daveholtz · x · 2026-11-19

OpenAI released updated, expanded data on how AI models accelerate AI research internally, following its first September report. Key findings: (1) AI agents successfully complete increasingly long tasks, but humans remain central — success rates vary hugely across EpochAI's ONET AI R&D task taxonomy; (2) despite that heterogeneity, researcher adoption of agents keeps growing, and AI researchers assign high value to agent use. Author daveholtz also credited Chris Ong and apologized for tagging the wrong Chris Ong.

Related event: OpenAI's Research Acceleration Report: Agents Now Complete Nearly Half of Long Tasks, Calls for Mandatory RSI Disclosure(10 posts)→

Original post →

More from AGI Musings

AGI Musings channel →