Why Are 84% of Researchers Now Using AI in Their Work?

A recent Wiley study, released in October, found that 84% of researchers are now actively using AI tools to support their work, marking a significant increase from 57% in 2024.

The comprehensive survey tracked 2,430 researchers this summer, revealing steep growth driven by rising efficiency and changing attitudes toward technology.

With nearly three-quarters of respondents saying AI has improved both the quantity and quality of their output, enthusiasm is high.

Yet caution is growing as researchers reconsider AI’s limits, bringing nuance to the adoption trend and reframing expectations around the capabilities of artificial intelligence.

How rapidly has AI adoption grown among researchers?

AI use among researchers surged in just one year, increasing from 57% in 2024 to 84% in 2025, according to the ExplanAItions study. This rate of adoption is unprecedented for new technology in academic settings, indicating broad utility and a snowball effect as peers share successful practices.

Usage has expanded beyond technical fields, with researchers in the humanities, social sciences, and business now integrating AI into literature reviews, data analysis, and writing.

The surge has also seen AI adoption for a wide array of tasks, from designing experiments to publishing workflows and data preparation.

Did you know?
Despite widespread adoption, only 11% of researchers recognize top specialized AI research tools by name, highlighting a major gap in sector-specific awareness.

What is driving researchers to embrace AI tools?

Efficiency is the top motivator: 85% of surveyed researchers reported working faster and smarter with AI tools. Nearly 62% now rely on AI for research and publication tasks, up from 45% just one year ago.

Popular general-purpose tools, especially ChatGPT, dominate academic use, with 80% citing it as a leading resource, while only 25% use specialized research assistants.

Researchers noted that AI helps with repetitive tasks, offers quick subject overviews, and streamlines data collection.

These practical benefits lead to stronger output and allow for more time to devote to thought-intensive aspects of their work.

Are expectations about AI’s capabilities changing?

Despite increased usage, the community is recalibrating expectations. In 2024, most believed AI could outperform humans in more than half of research scenarios; now, fewer than one-third agree.

Researchers are transitioning from initial enthusiasm to informed skepticism based on practical experience.

A “reality check” is also underway, with more users recognizing AI’s contextual limitations, the risk of hallucinations, and the missing domain knowledge in highly specialized fields. The shift reflects a mature understanding of both the strengths and boundaries of AI.

ALSO READ | How Is Google’s Opal Changing AI App Creation Worldwide?

What risks and concerns do researchers now prioritize?

Concerns about AI accuracy and hallucinations increased from 51% to 64% within a year, underscoring the need for greater vigilance over the quality of results.

Privacy and data security worries also rose from 47% to 58%, while many academics voice doubt about the reliability of outputs created with generative tools.

Researchers call for greater transparency, guidance on tool limitations, and institutional standards for the responsible use of these tools.

The heightened caution mirrors broader debates about the reliability and ethical deployment of AI in research.

How can institutions support effective AI use in research?

Although researchers welcome AI, only 41% feel they have adequate support from their institutions. Most respondents believe publishers should set guidelines, with 73% actively seeking more direction.

Corporate researchers benefit from better access (58% versus 40% overall) and exhibit higher confidence in AI’s capabilities, likely due to the structured rollout and allocation of resources.

Expanded access and publishing guidelines could support more productive adoption and help close the gaps created by rapidly evolving technology.

Training, community sharing, and clearer policies may allow cautious experimentation to give way to stronger integration and collaboration across sectors. Researchers now face an era where AI is inseparable from the research process.

What began as cautious trial and error has transformed into a nuanced, collaborative approach to integrating technology, setting the stage for further innovation, deeper trust, and a new generation of research powered by intelligent tools.