Field diffusion · General science · Social science

OpenAlex study links AI use to higher measured scientific creativity

A publication-level analysis of AI adoption and scientific creativity measures across more than one million papers.

Summary

Ding, Lawson, and Shapira analyze more than one million OpenAlex publications and report that AI publications are more likely than non-AI publications to reach top-decile creativity measures. They distinguish tool-oriented AI use, associated with recombinant novelty, from adaptation-oriented AI use, associated with object-based creativity.

AI role

AI adoption was measured in publications and compared across tool-oriented and adaptation-oriented modes of AI use.

Narrative role

This adds a direct meta-science measurement of whether AI-linked papers differ in creativity and impact, complementing adoption-count and survey evidence.

Caveat

The source is a new preprint, and creativity metrics based on publication data cannot prove that AI use caused the measured differences.

Event

Date: 2026-06-03
Timeline role: Supporting
Progress relevance: Field impact
Validation: Downstream measured

Evidence

Preprint · Bibliometric analysis

Observed metrics

publications analyzed: over one million; The paper uses OpenAlex publications as the primary researcher-output dataset.
top-decile creativity lift: 5.5-10.2 percentage points; The authors report higher likelihood for AI publications to rank in top-decile creativity than non-AI publications.

Source: arXiv ↗