Scientists Broadly Negative on AI Yet Driven by FOMO to Adopt

Almost half the researchers in a new Nature poll of 1,907 scientists said they feel broadly negative about artificial intelligence, with only 30% reporting positive feelings. Yet the same poll found that a majority are using AI tools anyway, and that competitive anxiety, more than genuine conviction, appears to be driving adoption.

The usage numbers are striking: 25% of respondents use AI models daily for research, 26% weekly, and 15% at least monthly, leaving just 34% who say they never use them. At the same time, 63% said AI risks outweigh benefits for data analysis and scientific literature review -- two of the most routine research tasks. The tension is captured plainly by Alexander Gibson, a PhD candidate at Queensland University studying clinical prediction models: "If you want to stay in the fight, you are left without a choice to adopt the use of some tools."

What the poll measured isn't outright rejection of AI -- it's closer to resigned adoption. When asked about AI's impact on scientific research overall, only 23% of respondents thought the effect was positive; 31% said it was negative. And 59% said they fear they would lag behind without using AI tools at all.

The honest caveat: Nature's readership is self-selected, and the available reporting does not describe the full survey methodology or how scientific disciplines were distributed in the sample. What this poll does not tell you is which specific AI tools researchers find acceptable versus problematic, or whether the skepticism is concentrated in particular fields. But with nearly 1,900 researchers responding, the signal is hard to dismiss: in science right now, AI adoption appears to be pushed more by fear of falling behind than by confidence that the tools are actually improving research.