Google's Alleged ScientistTwo Multi-Agent Framework Claims 25.2% Gains Over Top NeurIPS/ICML Papers
EngMoElgaraihy · x · 2026-09-27
An unverified third-party post claims Google Research has launched "ScientistTwo," a multi-agent framework said to run the full scientific research loop—literature review, hypothesis generation, coding, experiments, ablations, simulated peer review, and paper writing—without human intervention.
Key claims (unconfirmed):
- Benchmarked against accepted papers from elite venues like NeurIPS and ICML, reportedly beating the best human results on 86 tasks with a 25.2% performance improvement;
- Claims hallucination-free citations, working code, and publish-ready papers;
- Framed as an "independent researcher" rather than an assistant.
The post's tone is heavily hype-driven ("100% automated PhD"), and neither the official paper nor the benchmark methodology has been independently verified.
Related event: Reported Google Framework ScientistTwo Claims Fully Automated Research(2 posts)→
More from AGI Musings
- Malthus Saw It in 1798: Wealthy Avoiding Kids, Delayed Marriage Repeats Today — SydSteyerhart · 2026-11-16
- Runway CEO Predicts Fully AI-Generated Cultural Phenomenon by 2027 — c_valenzuelab · 2026-11-16
- '5% worse and 10% cheaper': the button with no vote is already everywhere — eigenrobot · 2026-11-16
- Effective Altruism is top-down and hierarchical, not a grassroots movement — ctjlewis · 2026-11-16
- Lawyers say clients' AI use creates unrealistic expectations and fake case law in settlements — hoofnagle · 2026-11-16
- Dev debunks 'rogue AI' hype: subsidized subscriptions can't fund expensive swarms — gerardsans · 2026-11-16