Peer review asks scientists to judge work produced by their competitors, usually under severe time pressure. AI may help authors produce and submit more papers, but it may also help reviewers check arguments, code and data. The effects depend on how authors, reviewers and journals respond to one another.

This project will build a game-theoretic or computational model of scientific publishing. Possible questions include whether AI-assisted review can keep up with AI-assisted writing, which disclosure rules can be sustained, and how a journal's objectives affect the behaviour of authors. The student may extend an existing model or develop a simpler model around a new question.