DeepMind is building an AI to peer-review science — Springer Nature signed up

Springer Nature and Google DeepMind said on Monday, October 26, that they will co-design an AI peer-review assistant built on DeepMind's Gemini models, to help reviewers work through large volumes of text, charts and figures.

A prototype is already being tested inside Springer Nature's secure systems, running on Google's Agent Development Kit in its private Gemini Enterprise platform; the publisher says no research content leaves its secure environments.

The tension is the timing: AI-assisted reviewing is already surging and controversial — funders such as UK Research and Innovation are trialling it — and the assistant is pitched as keeping expert human judgment at the center while submission volumes keep climbing.