PERSON

Harry Collins

The sociologist of scientific knowledge who spent forty years embedded in communities of practice to prove that expertise is not a property of individual minds but of social groups—and who in 2026 tested a large language model against the discourse practices of gravitational wave physicists to show that the most impressive AI systems in existence still lack the collective tacit knowledge that constitutes genuine expert judgment.

Harry Collins is the sociologist who went to the laboratory and stayed. For more than four decades he has been embedded in the community of gravitational wave physicists—reading every paper, attending every conference, conducting hundreds of hours of interviews—not to study physics but to study how physicists produce the knowledge they produce, and what that process reveals about the nature of expertise itself. What he found dismantles the commonsense picture. Expertise is not a substance that individual minds contain; it is a property of social groups that individuals participate in. The knowledge required to build a working gravitational wave detector cannot be fully captured in published papers; it lives in the community of practitioners and is transmitted through apprenticeship. This finding, extended to artificial intelligence, produces Collins's central and uncomfortable claim:

Keep reading with YOU ON AI