The diverging effect of prestige and experience on the use of artificial intelligence knowledge

Abstract This paper examines the relationship between scientists' experience and prestige and their use of artificial intelligence (AI) knowledge. Analyzing citation patterns from 5 million papers (1990–2024) citing AI literature and 1.1 million authors, we find that prestige positively correlates with citing highly visible AI work, while experience shows the opposite pattern. This divergence persists both within and beyond Computer Science contexts. We interpret these findings as suggesting that prestige is associated with social mechanisms linked to knowledge convergence toward mainstream work, while experience may be associated with knowledge diversity. These findings also reveal concerning concentration trends in AI knowledge utilization and inform strategies for more effective scientific knowledge use.