Springer Nature and Google DeepMind Build AI Peer-Review Assistant

Springer Nature and Google DeepMind are developing an AI assistant to help researchers peer review manuscripts, Research Information reported. The assistant uses Gemini models to help reviewers work through complex papers, analyse large amounts of text, charts and figures, and flag parts that need closer examination.

A prototype is being tested. It is built with Google's open-source Agent Development Kit and runs entirely within Springer Nature's private Gemini Enterprise environment, so manuscripts stay inside the publisher's infrastructure. The partners said the tool will not replace human reviewers.

"That means testing what AI can add while being clear about where responsibility remains," said Saskia Steinacker, chief product officer at Springer Nature.