Big Tech-Funded AI Papers Have Higher Citation Impact, Greater Insularity, and Larger Recency Bias

Over the past four decades, artificial intelligence (AI) research has flourished at the nexus of academia and industry, while a small group of Big Tech companies increasingly control computational resources, data, and talent. So far, it has been unclear how many papers industry funds in top AI venues, how their citation impact compares to other papers, and how their citation practices differ. We address this gap by analyzing ~ 49.8 K papers from 10 top AI conferences, ~ 1.8 M citations from AI papers to other papers, and ~ 2.3 M citations from other papers to AI papers (1998-2022, Scopus). We study the evolution of industry funding, the citation impact of funded papers, the diversity and temporal range of their citations, and the subfields where industry is most active, framed through 7 research questions. Industry presence grows markedly after 2015, from less than 2 % of papers to more than 11 % in 2020, and then stabilizes around 9 %. Between 2018 and 2022, 12 % of industry-funded papers achieve high citation rates (h5-index), compared to 4 % of non-industryfunded and 2 % of non-funded papers. We propose the Citation Preference Ratio (CPR) to measure how much top AI venues engage with different funding types and find that they cite industryfunded research more than expected given its volume. Industryfunded research is also more insular: it cites other industryfunded work more often and non-funded work less often than expected. Finally, industry-funded papers show a stronger recency bias, citing more recent work and fewer older papers than nonfunded work. Our findings point to three intertwined trends in AI research: (1) a growing citation impact of industry-funded papers, (2) greater insularity of industry-funded works than nonfunded works, and (3) a stronger preference for recent work in industry-funded research. While industry funding substantially contributes to AI advances, these trends raise questions about Big Tech's influence over research agendas and the long-term diversity of the field. All data and code are publicly available: https://github.com/Peerzival/impact-big-tech-funding.