The Frankfurter Allgemeine Zeitung has published an article by Nils Althaus on how deeply AI has moved into the research process — writing papers, reviewing them, and increasingly doing the research itself. It reports on submission volumes rising sharply since ChatGPT, on surveys showing that a majority of reviewers now use language models despite publisher bans, and on findings that AI-generated reviews were rated higher than human ones on most quality criteria at a major AI conference. It also points to a countervailing effect: AI-assisted research tends to concentrate on fields where large datasets already exist rather than opening up new ones.

Prof. Iryna Gurevych is quoted throughout on the scale of the shift — students now submitting to leading AI conferences at a stage that would previously have required years of research experience — and on what follows from it: that science will be “AI first”, and that the central challenge for the research system will be making these systems verifiable and controllable so that responsibility stays with people.

FAZ, „Wenn KI die eigene Forschung durchwinkt", September 10, 2026