When can we say AI made a scientific discovery?
Anthropic's Claude agents just ran real biology experiments. Here's how to know if that counts as discovery.
Why it matters
Anthropic deployed Claude agents in a molecular biology lab to conjecture and guide experiments. The piece examines the epistemological boundary: when does AI-guided hypothesis generation and experimental design constitute discovery versus assistance. Practitioners need to understand attribution, reproducibility, and IP implications as AI moves from tool to co-investigator.
The key facts
5 to knowAnthropic launched a molecular biology lab earlier in 2026
Claude agents read biology problems and generate conjectures
Human scientists run experiments based on agent suggestions
Core question: attribution and definition of AI-driven scientific discovery
Implication for IP, publication credit, and validation protocols
Go to the source
MIT Technology Reviewtechnologyreview.com
Publisher excerpt: This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Last Wednesday, Anthropic announced that earlier this year it had launched a molecular biology lab, where Claude agents read and conjecture about hard biology…