The Generative AI Gold Rush in Theoretical and Computational Research

Summary

A synthetic-control study of 2,080 monthly observations across twenty arXiv archives (Jan 2018 - Aug 2026) documenting a large 2026 output regime shift concentrated in Mathematics: 47,127 list entries, 33.5% above 2025 and 11.9% above the synthetic counterfactual, robust across donor refits, placebos and alternative pre-periods. An author panel shows the repeated-output tail thickening (share producing 5+ submissions rising 2.45% to 3.80%, 10+ from 0.21% to 0.49%). Directly relevant as quantitative evidence on AI's productivity effect in fields with verifiable tasks, and it names verification, selection and attention as the binding governance constraints.

Contribution

Queued — not yet read.

Method

Queued — not yet read.

Relevance to RISE

Surfaced by the 2026-09-07 literature sweep as a high-priority candidate for the RISE knowledge base (angle: research productivity of AI). Verified against: Opened https://arxiv.org/abs/2609.04872 on 2026-09-07; verbatim abstract read (synthetic control, 47,127 entries, 33.5%/11.9%, author tail statistics), three authors, submitted 4 Sep 2026. Not in dedupe_titles.txt.

Critique / open questions

Queued — not yet assessed.

Key quotes

Generative AI is changing the production conditions of theoretical and computational research, but its sys tem level effects require measures that separate plat form growth, field specific divergence, and production structure. (abstract)

We assemble 2,080 monthly observations for twenty arXiv archives from January 2018 through Au gust 2026 and a separate pseudonymized Mathematics author panel. (abstract)