Who Uses Artificial Intelligence in Research – And for What?
Chugunova et. al: "Who uses AI in research, and for what? Large-scale survey evidence from Germany" Research Policy, Volume 55 (2026)
In June 2024, all employees of the Max Planck Society and the Fraunhofer Society were invited to participate in an anonymous survey on their use of AI tools at work. As researchers and support staff differ substantially in their task profiles and opportunities to use AI, they are analyzed separately. This summary focuses on 6,215 complete responses from researchers, which are broadly representative of the two organizations. The survey covered AI tools in general rather than generative AI specifically, although the latter likely featured prominently in respondents’ considerations given their visibility in public discourse.
AI use among researchers is already widespread. Over 42% report being very or fairly familiar with AI tools, and 44% have used them at least a few times. Nearly one in four researchers (25.9%) use AI daily or more frequently, while only 22.2% report never using AI for work.
Clear patterns emerge in who adopts AI. Younger researchers use AI more frequently than older colleagues, and familiarity increases with educational attainment. Consistent with previous studies, we observe a gender gap in AI use for research tasks. Our data show that this gap is largely explained by differences in familiarity with AI tools rather than by ability or negative attitudes. Once female researchers use AI, they find it just as helpful as men do.
Researchers who use AI tend to hold more positive views about its impact on research quality, skill development, and society. While 69.2% expect AI to transform or even revolutionize their field within the next decade, opinions about societal effects are more mixed: 40.6% see more opportunities than risks, whereas 22.2% perceive more risks than opportunities.
AI is increasingly used throughout the research process. Common applications include testing ideas, writing code, and drafting research papers. Notably, researchers tend to use AI most for the tasks on which they spend the greatest amount of time, suggesting that AI is becoming a co-creator rather than a peripheral support tool.
Efficiency is a key motivation: 50.4% of researchers report using AI primarily to speed up their work. However, effective use requires skill. To proxy prompting ability, respondents were asked to create a prompt that would allow a large language model to identify a visual phenomenon. Only 21% succeeded, highlighting that prompting is a skill in itself. Training substantially increases the likelihood of producing effective prompts.
Several barriers to wider AI adoption could be addressed through institutional action. The most frequently cited obstacles are legal uncertainty (17.6%), lack of knowledge (17.4%), and limited access to suitable tools (16.6%). Researchers also express a strong demand for guidance: 58.7% expect high-level direction from supranational bodies such as the EU, and 51.3% from their own research organizations.
Overall, AI is rapidly becoming an integral part of research practice. Understanding who uses AI, for which tasks, and under what constraints is essential for designing policies that support effective and responsible adoption while safeguarding scholarly standards.
Link: Who uses AI in research, and for what? Large-scale survey evidence from Germany
Authors: Marina Chugunova (Max Planck Institute for Innovation and Competition), Dietmar Harhoff (Max Planck Institute for Innovation and Competition), Katharina Hölzle (Frauenhofer IAO), Verena Kaschub (Frauenhofer IAO), Sonal Malagimani (Frauenhofer IAO), Ulrike Morgalla (Max Planck Institute for Innovation and Competition), Robert Rose (Frauenhofer IAO)
