Google's PaperOrchestra uses 5 AI agents to write a full paper in ~40 minutes
alex_verem · x · 2026-11-15
Google introduced PaperOrchestra, a multi-agent framework that turns a scientist's rough notes and raw results into a submission-ready LaTeX paper. Five agents split the work: outline planning, figure/diagram generation, citation-verified literature search, section writing, and simulated peer-review-driven revision.
Key results from reconstructing 200 top AI conference papers:
- 40 minutes per paper on average
- Simulated acceptance rates of 84%/81% vs 86%/94% for human papers (AI reviewers)
- 46-48 citations per manuscript vs 59 for humans and 9-14 for other AI writers
- Human judges preferred its literature reviews over other AI systems by 50-68%; it tied or beat human-written reviews 43% of the time
Limits: a second AI reviewer scored it 48%/22% vs 71%/63% for humans, and 11 AI researchers preferred human full papers 81% of the time. Acceptance rates come from AI reviewers and should be treated as estimates; humans still supply the idea, run experiments, and stay accountable.
Related event: Google's PaperOrchestra Turns Research Notes into Full Papers(2 posts)→
More from coding & agent
- An AI agents developer roadmap: programming, math, and LLM fundamentals — blaizedsouza · 2026-11-15
- ABREKA: an autonomous research agent that runs for days, one experiment at a time — generativist · 2026-11-15
- Personal AI agents need their own social network to coordinate for you — StewartalsopIII · 2026-11-15
- The four-layer playbook for agents that repeat the same mistake all night — blaizedsouza · 2026-11-15
- How I rebuilt my daily schedule around AI agents: overnight runs and high-intensity check-ins — vinvan · 2026-11-15
- Haskell dev vibe-codes himself a better ghcid, and so can you — tomjaguarpaw · 2026-11-15