Scientists have a bad case of AI FOMO, Nature poll reveals
Skip to main content
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain<br>the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in<br>Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles<br>and JavaScript.
Advertisement
Bluesky
Many researchers use AI tools for research, despite feeling that AI use is having a negative effect on science.Credit: Oscar Wong/Getty<br>When it comes to artificial-intelligence tools, scientists have a serious case of FOMO — fear of missing out. A Nature poll of more than 1,900 researchers uncovered concerns about the negative impacts of AI, but revealed that many people feel pressured to adopt the technology owing to concerns about being left behind.<br>Nearly half of poll respondents said that they feel broadly negative towards AI, with 63% saying that the risks of using tools such as large-language models (LLMs) to analyse data and scientific literature outweigh the benefits. Despite these concerns, nearly 60% of respondents felt that they would be left behind if they did not use AI tools in their work. The poll was posted on the Nature website, on social media and in the Nature Briefing e-mail newsletter.<br>“If you want to stay in the fight, you are left without a choice to adopt the use of some tools — particularly LLMs — to keep up with others who are using them,” says Alexander Gibson, a PhD candidate studying clinical prediction models at the Queensland University of Technology in Brisbane, Australia.<br>Gibson says that he has felt a “passive push” to use AI tools because they promise to make everything more efficient. However, he feels that, although AI models can produce work faster, the quality of that work is not always better — and is sometimes worse — than work done by humans. For example, Gibson says that he used an LLM to extract data from PDF articles for his research. This made his research feasible because extracting the data manually would have been challenging, but the LLM made mistakes. “The quality of the work had not improved.”<br>Opinions on AI<br>Nature asked researchers about their perceptions of AI, whether AI models are having a broadly positive or negative effect on science and how often they use these tools. The respondents came from 75 nations, with the highest percentage of answers coming from the United States (38%). Of the 1,907 scientists who responded, nearly 48% said that they feel negative towards AI. Meanwhile, 30% of respondents to the poll said that they feel positive about AI overall, and 22% were neutral.<br>Only 23% of the respondents felt that AI tools were having a positive impact on research, whereas 31% said that the technology was negatively affecting science. Nearly half of the researchers felt that the impact of AI use depends on how the tools were used. Models designed for specific scientific tasks tended to be more popular than were general-purpose LLMs such as those behind OpenAI’s ChatGPT chatbot.<br>Meet the academics refusing to use generative AI
Despite their scepticism, most poll respondents said that they use some form of AI tool in their research, with 25% stating that they use them every day. 26% use AI models weekly, 15% monthly and 34% said that they never use AI tools.<br>The results are similar to those from a 2023 survey of Nature readers that asked people about their perceptions of early AI adoption in science. That poll found that researchers were concerned about AI models’ capacity to spread misinformation, entrench bias in data and contribute to rising unemployment.<br>Irene Kaplow, a computational biologist at Carnegie Mellon University in Pittsburgh, Pennsylvania, says that researchers aren’t sceptical enough about AI because people often use tools without fully understanding their limitations.<br>For example, Kaplow says that she uses AI tools to discover gene transcriptional regulatory elements — specific DNA sequences that control when, where and to what degree a gene is expressed. She often uses Google DeepMind’s AlphaGenome software, which can predict how genetic mutations will affect an organism, for her research. But the model has only been trained on human and mouse data and does not perform as well on other species. Still, Kaplow is aware of colleagues who continue to use the model on organisms beyond its remit.
Enjoying our latest content?
Log in or create an account to continue
Access the most recent journalism from Nature's award-winning team
Explore the latest features & opinion covering groundbreaking research
Access through your institution
or
Sign in or create an account
Continue with Google
Continue with ORCiD
doi: https://doi.org/10.1038/d41586-026-01690-7
Reprints and permissions
Related Articles
Are your students using AI to write papers? Take...