Elsevier's global survey of over 3,200 researchers reveals that AI adoption is accelerating—58% now use AI tools, up sharply from 37% in 2024—but trust and governance lag far behind. While most researchers see AI as a time-saver for literature review and data analysis, only a quarter believe AI tools are ethically developed or trustworthy, and regional confidence varies dramatically, with Chinese researchers far more optimistic than their US and UK counterparts. The findings highlight a critical gap: researchers need institutional governance, training, and AI systems built on transparent, verified data and human oversight to confidently integrate these tools into their work.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Elsevier surveyed more than 3,200 academic and corporate researchers across 113 countries and found that 58% now use AI tools in their work, up from 37% in 2024. However, only 32% believe their institution has good AI governance, and just 27% feel they have adequate training in using AI.
Why it matters
Researchers see AI as transformative—61% use it to find and summarize research, and 51% for literature reviews—yet they remain deeply skeptical about ethics and reliability. Only 23% believe AI tools are ethically developed, and just 22% find them trustworthy; this gap between adoption and confidence suggests institutions need to build guardrails and training to unlock AI's full potential safely.
What to watch
Regional confidence in AI diverges sharply: 68% of Chinese researchers think AI gives them more choice versus 29% in the US and 26% in the UK. Researchers globally point to five trust markers as essential—automatic citation (59%), up-to-date training data (55%), explicit safety training (55%), high-quality peer-reviewed content (55%), and human expert review (49%).
Researchers face mounting pressures that are transforming how they work: only 45% say they have sufficient time for research, 68% report greater pressure to publish than two or three years ago, and just 33% expect funding to increase in the next two to three years. Against this backdrop, AI adoption is outpacing institutional readiness. While 58% of researchers now use AI tools—a significant jump from 37% in 2024—fewer than one-third of institutions have implemented governance frameworks or provided adequate training, suggesting a widespread mismatch between supply and institutional support.
The trust problem is acute and geographically uneven. Globally, 39% view AI tools as unreliable and 38% believe they are unethically developed, yet researchers continue to adopt them anyway, likely out of necessity. This divergence is sharpest between China and the West: Chinese researchers express substantially higher confidence across all dimensions—from believing AI offers greater choice to optimism about its ability to save time and improve work quality—while US and UK researchers are notably more skeptical. The body identifies a clear pathway to bridging this gap: researchers want AI systems built on transparent, recent, peer-reviewed sources with explicit safety training and human validation. These trust markers—transparency, recency, safety, quality, and human oversight—appear to be the conditions under which researchers would more confidently deploy AI for higher-stakes tasks like hypothesis generation and study design, moving beyond current use cases focused on summarization and literature review.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
