Research led by James Evans at the University of Chicago has revealed a strong link between the utilization of artificial intelligence (AI) tools and academic productivity. Scholars using AI in their research publish nearly three times as many papers and gain close to five times the citations compared to their non-AI-using counterparts. This analysis, spanning over four decades and involving over 41 million research papers, was conducted in collaboration with researchers in China.
The study categorized the research into three periods: traditional machine learning (1980–2014), deep learning (2015–2022), and the current generative AI era (2023 onward). It focused on various natural sciences but excluded domains such as mathematics and computer science. Despite the positive findings regarding productivity, Milad Abolhasani of North Carolina State University pointed out that the study might miss certain papers where AI usage is not explicitly mentioned, potentially underrepresenting its prevalence in academia.
Additionally, the dataset showed that smaller teams employing AI tools have fewer junior researchers; however, early-career scientists in these teams are 13% more likely to remain in academia and achieve established researcher status about 1.5 years sooner than their peers who do not engage with AI. Caution was expressed by neuroscientist Molly Crockett from Princeton University, who noted that citation counts do not necessarily equate to research quality or impact.