Exploring How Acquired Diversity of Teams Influences the Innovation Diffusion in Artificial Intelligence

ABSTRACT This study explores innovation diffusion in Artificial Intelligence (AI) over the past 70 years, focusing on the impact of acquired diversity(including discipline diversity and topic diversity) of teams. We obtained 906,065 AI papers with 18,357,761 citations written by 1,316,936 authors between 1950 and 2015 from the Microsoft Academic Graph. Using citation analysis, we examine how AI innovations spread and the influence of the acquired diversity of teams on this process. We measure innovation diffusion from four aspects: innovation speed, innovation diffusion power, innovation diffusion breadth, and innovation diffusion breadth diversity. Our findings show that: (1) AI papers published before and after 2000 demonstrate distinct diffusion patterns. AI papers published before 2000 show a bimodal citation trend, while those released after 2000 present only a single peak in citations. (2) The first citation interval can be reduced by 2.35 years before 2000 and by 0.27 years after 2000 by enhancing topic diversity. (3) Both discipline diversity and topic diversity effectively promote innovation diffusion power, innovation diffusion breadth, and innovation diffusion breadth diversity in the field of AI.