AI in Scientific Research: Boosting Citations but Raising Equity Concerns

Scientific papers that mention AI get a citation boost

Papers with titles or abstracts that mention certain artificial intelligence (AI) methods are more likely to be among the top 5% most-cited works in their field for a given year than are those that don't reference those techniques, an analysis has found. These papers also tend to receive more citations from outside of their field than do studies that don't refer to AI terms. But this 'citation boost' was not shared equally by all authors. The analysis also showed that researchers from groups that have historically been underrepresented in science don't get the same bump in citations as their counterparts do when they use AI tools in their work -- suggesting that AI could exacerbate existing inequalities. The findings emerged from a study that aimed to quantify the use and potential benefits of AI in scientific research. But the report, published last week in Nature Human Behaviour, also raises concerns. Scientists might be incentivized to use AI purely as a way to increase their citations -- regardless of whether the AI tools improve the quality of the work, notes Lisa Messeri, an anthropologist of science and technology at Yale University in New Haven, Connecticut. "We want to make sure that, as we are [investing] in AI, we are not doing that at the deficit of other approaches," she says. The study also provides a much-needed quantification of how AI is changing scientific research, says Dashun Wang, a co-author of the study, and a computational social scientist who studies the science of science at Northwestern University in Evanston, Illinois. "Now we finally have systematic data," Wang says, which will be instrumental for addressing disparities related to the use of AI in science. To measure scientists' engagement with AI, the authors identified AI-related terms -- such as 'machine learning' and 'deep neural network' -- in the abstracts and titles of almost 75 million papers, covering 19 disciplines, that were published from 1960 to 2019. Wang acknowledges that, because of the cut-off date, the study doesn't capture recent developments in AI, including the rise of large language models such as ChatGPT, which are already changing how some researchers do science. According to the study, scientists in all 19 disciplines have ramped up their use of AI tools over the past two decades (see 'AI use takes off'). But there is wide variation: computer science, mathematics and engineering have the highest rates of AI use, and history, art and political science have the lowest. The rates for geology, physics, chemistry and biology are in between. To estimate the potential benefits of AI for each discipline, the authors first identified research-related tasks that AI can perform. Then they tracked the rise of these capabilities over time by detecting certain verb-noun pairs, such as 'analyse data' and 'generate image', in publications about AI between 1960 and 2019. By looking at how much these terms in AI-related publications overlapped with the basic tasks of a given research field over time, the researchers were able to assess whether AI's capabilities could meet the evolving needs of that field. Again, computer science, maths and engineering were associated with the highest potential benefits, and history, art and political science with the lowest. Marinka Zitnik, a specialist in biomedical informatics at Harvard Medical School in Boston, Massachusetts, says that the paper's approach is interesting because it allows for a systematic analysis across several scientific disciplines. But it comes with limitations. "Because the authors wanted to do a very broad, systematic study, that meant that they were not able to necessarily go into a very intricate understanding behind why a specific verb or noun would appear in a paper," she says. Just because certain verbs and nouns occur together in a paper, it does not mean that, if AI can perform the task described, it will necessarily be useful for that field, she notes.

Analysis of approximately 75 million publications finds those employing AI are more likely to be a 'hit paper'

From designing new drug candidates in medicine to drafting new taxation policies in social sciences, the benefits of artificial intelligence (AI) in scientific research are all around. Just this week, two scientists known for their pioneering AI research earned the Nobel Prize in Physics, and a trio of scientists earned the Nobel Prize in Chemistry, which recognized the use of advanced technology, including AI, to predict the shape of proteins. Despite its rapid progress and broad applications, however, many researchers lack a systematic understanding of how AI may benefit their research, and skepticism remains about whether AI is capable of advancing science in every field. A new Northwestern University study analyzing 74.6 million publications, 7.1 million patents and 4.2 million university course syllabi finds papers that employ AI exhibit a "citation impact premium." However, the benefits of AI do not extend equitably to women and minority researchers, and, as AI plays more important roles in accelerating science, it may exacerbate existing disparities in science, with implications for building a diverse, equitable and inclusive research workforce. The research team, led by the Kellogg School of Management's Dashun Wang and Jian Gao, developed a measurement framework to estimate the direct use and potential benefits of AI in scientific research by applying natural language processing (NLP) techniques to these vast datasets. Wang is a professor of management and organizations at Kellogg and of industrial engineering and management sciences at McCormick, director of Kellogg's Center for Science of Science and Innovation (CSSI) and co-director of Kellogg's Ryan Institute on Complexity. Gao is a research assistant professor at Kellogg CSSI. The study, "Quantifying the Use and Potential Benefits of Artificial Intelligence in Scientific Research," was published October 11 in the journal Nature Human Behaviour. "These advances raise the possibility that, as AI continues to improve in accuracy, robustness and reach, it may bring even more meaningful benefits to science, propelling scientific progress across a wide range of research areas while significantly augmenting researchers' innovation capabilities," Gao said. Most impactful research The study found that the recent successes of AI, across fields, has been remarkable for research. There has been a growing use of AI in disciplinary research since 2015, proxied by the mention of AI-related terms (such as "artificial intelligence," "deep learning" and "convolutional neural network") in the title or abstract of publications. From 2015 to 2019, disciplines including computer science (37%), engineering (24%), physics (24%), biology (22%), psychology (24%), economics (14%), sociology (30%) and political science (27%) have all shown notably sharp increases in direct AI use scores due to the development of new AI capabilities. Researchers examine the number of times a paper is cited, and they define a "hit paper" as being in the top 5% by citations for papers published in the same field and year. Regardless of discipline, disciplinary papers that mention AI-related terms in their title or abstract receive more citations, being more likely to be a hit, and receive a higher fraction of citations from other disciplines. "In addition to its expansion, the use and benefits of AI in research is pervasive across disciplines, but we found a systemic misalignment in AI education," Gao said. "The investment in AI in higher education is not at the same pace of the AI benefit in science." These results suggest that the supply of AI talent and knowledge in most disciplines appears inadequate with the benefits these disciplines may extract from AI capabilities, highlighting a substantial AI use-AI training gap. "The use of AI in scientific disciplines has raced ahead across science, while the educational focus on AI to upskill future scientists within each discipline has lagged," Gao said. Underrepresented groups in STEM The study also highlights the unequal effects on women and minority researchers that the steadfast rise of AI use in scientific research may bring. "Historically, we know that women and minorities are less represented in some fields, especially in STEM," Gao said. "We found that as the AI use in science continues to grow, those same groups are less likely to benefit from the new technologies." Researchers suggest that an investment in making sure the training behind AI is equitable may have a positive impact on closing the demographic gap. What's next? As AI rapidly evolves, the researchers said we need to continuously monitor and update its benefit to science. "Women and minorities are benefiting the least, so how do we mitigate these disparities along demographic lines?" Gao said. The research team's analysis supports the hypothesis that collaboration between domain experts and AI researchers may represent a meaningful way to facilitate AI use across science and fill the AI use-AI training gap. "There's a benefit to increasing AI training across disciplines, which would likely help the disciplines to develop domain-specific AI expertise, allowing them to enjoy greater and timelier benefits from AI advances," Gao said.

Quantifying the use and potential benefits of artificial intelligence in scientific research - Nature Human Behaviour

The rapid advancement of artificial intelligence (AI) is poised to reshape almost every line of work. Despite enormous efforts devoted to understanding AI's economic impacts, we lack a systematic understanding of the benefits to scientific research associated with the use of AI. Here we develop a measurement framework to estimate the direct use of AI and associated benefits in science. We find that the use and benefits of AI appear widespread throughout the sciences, growing especially rapidly since 2015. However, there is a substantial gap between AI education and its application in research, highlighting a misalignment between AI expertise supply and demand. Our analysis also reveals demographic disparities, with disciplines with higher proportions of women or Black scientists reaping fewer benefits from AI, potentially exacerbating existing inequalities in science. These findings have implications for the equity and sustainability of the research enterprise, especially as the integration of AI with science continues to deepen.