In a recent announcement, Anthropic revealed that it has established a molecular biology lab where its Claude AI agents engage in reading and proposing solutions to complex biological challenges. The company claims that this AI-driven laboratory has made its first discovery, sparking debates within the scientific community regarding the nature of AI contributions to research.
Anthropic’s Claude agents, operating over a span of 21 hours, detected a previously unrecognized repeating pattern surrounding a known enzyme. While this discovery is intriguing and reminiscent of foundational work that led to CRISPR technology, critics argue that identifying genetic patterns is merely the first step in a lengthy scientific process. Renowned biologist Mario Rodríguez Mestre contested Anthropic’s claim, stating that his own team had previously identified the same pattern, raising questions about the originality of the AI’s findings.
The core of the controversy lies in the framing of AI as a self-sufficient discoverer rather than a tool that assists human researchers. This perception could undermine the collaborative nature of scientific inquiry, where breakthroughs typically emerge from teamwork and established methodologies. As AI companies like Anthropic and OpenAI continue to assert that their systems can achieve significant scientific milestones, the challenge remains to set clear standards for what constitutes a genuine scientific discovery by AI. Critics suggest that raising these standards is essential to ensure that when true breakthroughs occur, they are recognized and valued appropriately.
Source: When can we say AI made a scientific discovery? via MIT Technology Review