The Global AI Talent Tracker
Where the world's top AI researchers were educated, where they work now, and how talent flows between nations. The most comprehensive analysis of the global AI talent pipeline.
Version 3.0 — Updated with 2025 NeurIPS, ICML, and ICLR conference data
Researchers Analyzed
Countries Represented
Latest Data
Citing Institutions
Key Findings
Our analysis of 4,622 researchers who published at top AI conferences reveals significant shifts in global talent distribution. These findings have implications for policy, investment, and the future trajectory of AI development.
China-Educated Researchers
Of the 4,622 researchers who published at NeurIPS 2024, 38% received their undergraduate education in China — up from 29% just five years earlier. The talent pipeline is shifting.
Working in the United States
Of China-educated AI researchers, 72% now work at institutions in the United States. America's AI advantage isn't homegrown — it's imported.
US Institutional Dominance
US institutions employ 59% of the world's elite AI researchers, measured by publications at top-tier conferences. This lead is built almost entirely on foreign-born talent.
Staying in China
Only 11% of top China-educated AI researchers remain working in China — down from 16% in 2019. The brain drain continues despite massive domestic investment.
Interactive Explorer
Explore where the world's top AI researchers come from and where they work now. Filter by country of origin, current institution, and track the talent flows between major AI hubs.
Where Top AI Researchers Were Educated
Note: Country of origin is determined by undergraduate institution location. Data from NeurIPS 2024, ICML 2024, and ICLR 2025 conference proceedings.
Country-by-Country Breakdown
Detailed analysis of AI talent production, retention, and migration patterns for major countries. Click on any country to see institutional breakdown and policy implications.
Trends Over Time
Tracking year-over-year shifts in AI talent distribution reveals accelerating trends. China's share of AI talent production is growing, while its retention rate continues to decline.
Year-over-Year Shifts (2017-2024)
China Production Rising
The share of top AI researchers educated in China rose from 27% in 2017 to 38% in 2024 — an 11 percentage point increase in seven years.
US Employment Growing
US institutions now employ 59% of elite AI researchers, up from 51% in 2017. This growth is almost entirely from importing talent.
China Retention Declining
China's retention of its top AI talent fell from 16% to 11% despite massive domestic investment in AI research infrastructure.
Policy Implications
These trends have profound implications for the US-China AI competition. America's AI advantage is not built on homegrown talent — it's built on visa policy. Any restriction on immigration from China directly weakens US AI capacity. Meanwhile, China's massive investment in domestic AI has increased production but failed to solve retention. The talent pipeline flows one direction: toward America.
Methodology
Our research methodology is designed for transparency and replicability. Every finding is based on verifiable data from public sources. Here's exactly how we collect and analyze the data.
Conference Selection
We analyze authors from NeurIPS (Neural Information Processing Systems), ICML (International Conference on Machine Learning), and ICLR (International Conference on Learning Representations) — the three most prestigious venues for AI/ML research. Publication at these conferences is widely accepted as a marker of elite research capability.
Researcher Identification
We identify all authors of accepted papers at these conferences for each year. For the 2024 analysis, this yielded 4,622 unique researchers. We exclude purely industry authors where possible to focus on research talent.
Education & Employment Coding
For each researcher, we identify their undergraduate institution (as a proxy for country of origin/education) and current institutional affiliation. This is done through a combination of automated scraping of Google Scholar profiles, institutional websites, and manual verification.
Validation & Quality Control
A random sample of 10% of researcher codings are manually validated. Our inter-coder reliability for country classification exceeds 95%. When undergraduate institution cannot be determined, the researcher is coded as 'Unknown' and excluded from origin-based analyses.
Data Sources
- NeurIPS 2024 proceedings (3,587 accepted papers)
- ICML 2024 proceedings (2,609 accepted papers)
- ICLR 2025 proceedings (2,891 accepted papers)
- Google Scholar profiles for researcher affiliations
- Institutional websites for employment verification
- LinkedIn profiles for educational background (when public)
Limitations & Caveats
- Undergraduate institution is used as a proxy for 'country of origin' but may not capture researchers who moved before university
- Self-reported affiliations may be outdated for researchers who recently changed institutions
- Industry researchers at closed labs (e.g., some Google DeepMind divisions) may be undercounted
- The focus on top-tier conferences may miss important applied research in industry