Academic publisher Wiley finds researchers worldwide expect to use generative artificial intelligence more in future
Researchers believe artificial intelligence can outperform humans in more than half of research tasks, a study has found.
In a report launched on 4 February, the academic publishing company Wiley assessed how nearly 5,000 researchers across a range of countries, career stages and fields of study use AI. The majority believe AI already outperforms humans in 23 out of 43 use cases considered.
“AI processes more and faster information compared to humans,” said a mid-career researcher in Canada participating in the study, although other researchers in Germany and Croatia stressed that all AI outputs must be reviewed manually by humans.
The use of AI tools in research tends to be focused on writing tasks, such as translation and proofreading, with very few researchers using them for administrative tasks or post-publication support, the study found.
It focused on two key dimensions—interest in AI and human performance—which the authors tested over five elements of the research process: determining what to research, conducting the research, preparing for publication, peer review, and promoting and sharing outputs.
Guidance needed
While AI is seen as a powerful tool for researchers, one key finding was that the technology’s use is still fairly limited, with 63 per cent of researchers citing lack of guidance as a key barrier to wider adoption.
Seven in 10 researchers expressed a “strong desire” for publishers to provide clear guidelines on which uses of AI are acceptable when publishing scholarly research.
Wiley said it is “actively developing AI guidelines for authors”, which will be launched in the coming months.
“Our study reveals a striking pattern—while researchers are eager to embrace AI’s benefits, they face significant barriers including a lack of clear guidelines on acceptable use,” said Jay Flynn, Wiley’s executive vice-president and general manager for research and learning.
Future skills
The study found some differences in the take-up of AI between countries and career stages.
Researchers in China and Germany were more likely than other countries to use AI, at 59 per cent and 57 per cent respectively, and early career researchers were more likely to use the technology than those further down their career paths.
Overall, researchers in the study said they thought developing AI skills will be important to them in the near future.
“While researchers do have some concerns about the models themselves, they also have a favourable view of the available generative AI technology and can envision a lot of possibilities for current use,” the study found.
Even researchers “who are less enthusiastic about AI still expect that they will need to adapt to increasing AI use within the next couple of years”.
