How experience moderates the impact of AI suggestions on researchers' perceptions of their ideas

At the heart of scientific discovery are researchers who identify ideas worthy of inquiry. In performing research tasks, researchers increasingly use generative artificial intelligence (AI) technologies. While this technology offers the promise of increased efficiency across the research process, little is known about how generative AI assists researchers with the generation of initial research ideas—one of the most fundamental tasks in science—and their willingness to adopt it for such purpose. In a pre-registered randomized online experiment with 310 researchers and a follow-up ideation study with 28 researchers, we show how one's research experience plays a role in receiving generative AI suggestions when formulating new research ideas. Experience negatively moderates both the incorporation of generative AI suggestions as well as its effect on perceived novelty and impact of the researcher's idea. We also find that generative AI may be reshaping research ideation by adding a layer of non-human exchange that precedes feedback from human experts. By focusing on the critical moment at which research ideas spark, we provide an initial understanding of how researchers respond to generative AI at a time when the paradigm of human knowledge production may be shifting from a human-only to a joint human-AI model.