Researchers from China have been pivotal in advancing artificial intelligence (AI) at American institutions, contributing significantly to the body of research produced in the United States. An analysis of top AI research papers reveals that researchers of Chinese origin have been contributing as much, if not more, than their American counterparts, raising concerns amid escalating tensions between the two nations: Are these U.S.-based Chinese researchers now returning to China in large numbers, potentially bolstering the capabilities of a rival superpower?
To explore this question, researchers updated a dataset originally compiled by the Paulson Institute in 2020, known as the Global AI Talent Tracker. This dataset focused on 675 top-tier AI researchers whose papers were accepted at NeurIPS 2019, a prestigious AI conference with a mere 20 percent acceptance rate. Among these, 100 were identified as Chinese-origin researchers affiliated with U.S. institutions as of 2019. Recent investigations aimed to discover their current affiliations and assess whether the U.S. has managed to retain this pool of talent.
The results showed that 87 of the 100 researchers remain in the United States, while only 10 have moved to work for Chinese companies or universities, with the remaining three in other countries. This suggests a stronger retention rate than anticipated, especially in light of the geopolitical climate. Historically, Chinese researchers in the U.S. have a high stay-rate, around 90 percent, but recent years have seen a notable uptick in those choosing to return to China, raising questions about the sustainability of America’s competitive edge in AI.
Despite the positive retention numbers, challenges loom. The data indicates that the U.S. may be struggling to attract new talent from China. While the majority of researchers in the sample have chosen to stay, an increasing number of talented Chinese AI researchers appear to be opting to remain in China for their careers. Factors influencing these decisions include heightened scrutiny and challenges surrounding visa processes for Chinese nationals, especially following actions taken by the U.S. government since 2018 that have made it more difficult for Chinese students and researchers to secure or renew visas.
High-profile indictments of Chinese researchers have contributed to a climate of mistrust, with many feeling the weight of racial profiling, as reported by 42 percent of participants in a 2021 survey. This atmosphere, compounded by travel restrictions during the COVID-19 pandemic, has further dissuaded Chinese researchers from pursuing opportunities in the U.S. In contrast, China’s burgeoning AI industry has increasingly attracted these talents back home, providing opportunities to work at the forefront of AI research without the necessity of relocating to the U.S.
Interestingly, the majority of the 87 researchers who have remained in the U.S. have transitioned from academia to the private sector, with 41 now employed by U.S. companies, 40 holding academic positions, and six engaged in doctoral or postdoctoral research. Notably, more than half of those working in the private sector are affiliated with major tech firms such as Google, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla. Additionally, three of the 87 have even founded their own startups in the U.S.
Among the returnees is Yang Zhilin, who completed his undergraduate degree at Tsinghua University before pursuing a PhD at Carnegie Mellon University, where he became a lead author on influential AI research papers. In 2023, Yang returned to China to establish Moonshot AI, which has since raised over a billion dollars and produced the Kimi model series, a set of high-performing open-source models. This shift highlights the complex interconnection between the two nations’ AI research communities, as U.S. companies now adopt and build upon models developed in China.
As the landscape of talent shifts, the implications for U.S. competitiveness in AI become apparent. Although the U.S. retains many seasoned Chinese researchers, a trend has emerged showing a dwindling ability to attract new arrivals. The original Global AI Talent Tracker indicated that in 2019, Chinese-origin researchers accounted for 29 percent of authors at NeurIPS, surpassing both American and European researchers. By 2022, this figure had risen to nearly half, with Chinese institutions more than doubling their share to 28 percent, indicating a rapid improvement in China’s research output.
If these trends continue, the U.S. may face a precarious future in AI innovation. Historically, America has benefited from a substantial influx of elite Chinese researchers who have established long-term careers in the country. However, if this flow diminishes or reverses, the United States could struggle to fill the resulting talent gap.
To mitigate such risks, an “all of the above” strategy is essential. This would involve bolstering investments in education at various levels to foster homegrown talent, ensuring research funding and visa certainty to attract international scholars, and creating an environment appealing to top researchers worldwide, including those from China. While none of these strategies are simple to implement, the importance of addressing these issues is clear as the stakes for future AI leadership escalate.
See also
Cisco-OECD Study Reveals Emerging Economies Lead Generative AI Adoption with Youth Engagement
University of Toronto Launches $20M Hinton Chair in AI with Google’s Support
Deep Learning Model EfficientECG Achieves 95% Accuracy in Rapid ECG Classification
Apple Reveals M4 Chip with 38 Trillion Operations/Second, Boosting iPad AI Performance
