Research & education Research
OpenAI reports uneven progress in agents assisting AI research
OpenAI published updated internal measurements of coding agents in research, reporting improving task completion alongside substantial differences between kinds of work. [1]
- Stage of evidence
- Company observational report; human researchers remain involved [1]
What the researchers found
The report combines agent-use records, estimates of autonomous task success and researcher perspectives. It extends a September study rather than announcing a new model or an autonomous laboratory. [1]
Why it matters
For research teams, the practical implication is to assess assistance task by task and retain human oversight, rather than assume one productivity result applies throughout a laboratory. [1]
Key dates
- OpenAI publishes updated research-acceleration measurements [1]
Limits of the evidence
These are company measurements of its own workplace. Changing task selection and estimation methods complicate comparisons; the findings do not establish fully autonomous recursive self-improvement. [1]
Sources and verification
Catalog evidence checked How we verify records
- Primary source Research acceleration: The jagged frontier
Related profiles
Research & education
Isomorphic Labs AI drug-discovery platform
Isomorphic Labs
The official announcement identifies Greenoaks and Sequoia Capital as co-leads. Its accompanying release describes development of IsoDDE and work toward clinical trials.
Research & education
NVIDIA Nemotron for Science
NVIDIA
The collection packages technical components for researchers and scientific computing teams.
Research & education
GraphCast
Google DeepMind
Uses recent atmospheric states to predict future weather variables including wind, temperature and air pressure.