An international research team from the University of Chicago and Tsinghua University in Beijing investigated the impact of using AI tools in research and science. They were surprised by the “dramatic extent” of the results.
The analysis of 68 million studies across six scientific fields showed that research using AI for data analysis or modeling was cited more frequently. Being often cited in recognized journals can significantly boost a scientist’s career.
However, AI-supported research focused on a narrower range of topics and repeated themes more often. The researchers noted that the more scientists use AI, the more they tend to focus on similar issues that can be addressed with large existing datasets.
James Evans, co-author of the study and director of the Knowledge Lab at the University of Chicago, highlighted the strong incentive for individuals to incorporate such systems into their work. This determines whether scientists succeed in a competitive research field.
Yet, fundamental questions that could lead to new research areas are being neglected. According to Evans, research is becoming increasingly dependent on technologies like machine learning or neural networks. The system of science using AI is shrinking.
The differences between AI-supported research and studies without AI were significant. Scientists using AI published 67% more studies per year and their work was cited more than three times as often. Additionally, researchers relying on AI support had more career success; they were 32% more likely to reach a leading position in a research team.
Evans noticed similar effects on scientific work with the increasing online publication of journals due to internet development. This fundamentally changed how and what was cited, narrowing research diversity.
Evans is not against new technologies and uses AI regularly in his research. However, he warns that the success of AI-supported work comes with a “lack of imagination.” He calls for slowing the shift to AI-supported research to preserve existing alternative approaches.
This shift in research methodology highlights the need to balance technological advancement with diverse and imaginative scientific exploration. The rapid adoption of AI in research has brought substantial benefits but also poses challenges to creativity and the exploration of new scientific fields.
As the scientific community continues to embrace AI, it is crucial to consider the broader implications on research diversity and innovation. Balancing AI use with traditional research methods could ensure a more comprehensive and imaginative scientific landscape.