Quantifying the Prevalence and Diffusion of Generative AI in Science
The project aims to quantify overt and covert usage of Large Language Models (LLMs) in science. By tracing both visible and hidden uses of Generative AI across scientific disciplines, we aim to measure its growing prevalence and assess how it is shaping scientific practices. The project also seek to model the diffusion and adoption of these technologies within scientific networks.
Generative AI has entered the lives of scientists, academics, and teachers. Papers co-written with AI are being published, and companies are releasing models that can help scientists summarize literature, solve math problems and analyze proteins. Infrastructures have been created for AI-assisted authoring of papers, and scientists now use generative AI to write code. This seems to be only the tip of the iceberg, as we do not know how much generative AI usage goes undisclosed.
Therefore, we ask: how are scientists using generative AI?
Is it only hype, or will the scientific community increasingly adopt generative AI in their work?
Our project has three core objectives:
- to examine how scientists use generative AI,
- to gauge developments in the prevalence of generative AI across disciplinary boundaries,
- to quantify the diffusion, adoption and impact of generative AI in scientific networks.
Funded by:
Project: Quantifying the Prevalence and Diffusion of Generative AI in Science
Grant type: Villum Synergy
Period: 2024 - 2026
Contact
PI 1: Roberta Sinatra robertasinatra@sodas.ku.dk
PI 2: Mathias Wullum Nielsen mwn@soc.ku.dk
Postdoc: Claudia Acciai acci@itu.dk – cla@sodas.ku.dk