Has Scientific Talent Shifted from Depth to Breadth?Evidence across Papers, Knowledge Inputs, Careers, and Teams

Details

A persistent narrative holds that generative AI will push researchers toward breadth over depth, using AI tools to span multiple domains rather than specializing deeply. This empirical study tests that claim using 47,959 articles from six fields over 2010–2025, 51,736 cited works, and reconstructed prior publication histories for 1,754 contributors. From 2010 to 2022, team sizes grew 37.3% (95% CI: 34.4–40.3%) while paper topic breadth declined by 0.0144 on a 0-1 hierarchical distance scale — outputs became slightly more focused even as teams expanded. Cited knowledge was stable to modestly broader, revealing a divergence: knowledge inputs are diversifying while paper outputs focus. One standard deviation of focal depth is associated with 8.2% higher citation impact (FWCI). Post-2022 deviations from earlier trends were not systematic and did not vary clearly with baseline AI intensity across 83 subfields. The data challenge the 'AI drives breadth' narrative and suggest focused individual accumulation coexists with expanding team-level collaboration — a more nuanced picture than enthusiasts or critics typically present. The post-ChatGPT window (post-2022) has limited data; longer observation periods are needed to detect structural AI-driven shifts.