Anthropic's 'first AI scientific discovery' sparks backlash from biologists
MIT Tech Review AI · rss · 2026-09-29
MIT Technology Review takes a deep look at the dispute over Anthropic's molecular biology lab:
- Anthropic claimed 950 Claude agents, running for 21 hours, flagged a previously uncatalogued repeating genetic pattern near a known enzyme, calling it "reminiscent" of what led to CRISPR.
- Biologist Lucas Harrington's viral critique — endorsed by Eli Lilly's chair and CEO — argued that "finding a weird cluster of genes and repeats is often the easy part; the hard part is figuring out what the system actually does." Routine grunt work, not a discovery.
- University of Copenhagen biologist Mario Rodríguez Mestre said his team had already discovered the pattern, and wondered whether Anthropic learned it from his Claude chats; Anthropic denies it, but Mestre has stopped using Claude.
The deeper issue: AI companies insist their systems make discoveries rather than serve as tools, which clashes with how science actually works. Once the bar is "did the AI discover something," debate collapses into breakthrough-or-bust — as OpenAI saw after its million-dollar math result drew claims that the problem didn't matter and that a mathematician's work was used without credit. Whittling 200,000 candidates down to a few is legitimate science, the author notes, but vendors' ever-lower discovery bars risk diluting real progress.
More from AGI Musings
- "If You Told 2020 That AI in 2026 Solved a Millennium Problem, You'd Call It the Singularity" — aran_nayebi · 2026-09-29
- Quintin Pope: cheap finetuning lets AIs defect, making durable AI coordination—and takeover—unlikely — QuintinPope5 · 2026-09-29
- When EA Ambition Means Buying Galaxies and Digital Descendants — abhiadesai · 2026-09-29
- Miles Brundage: not taking intelligence explosion seriously was my biggest recent mistake — Miles_Brundage · 2026-09-29
- Anthropic revenue chart lands in Museum of Chart Crimes for distorted time axis — infoxiao · 2026-09-29
- Dev claims 'AI brain rot' is real: the more he uses AI, the more he needs physical books — alexcovo_eth · 2026-09-29