Researchers of Chinese Origin Dominate the World's Top AI Talent

[Riaz Haq]

Recent launch of DeepSeek AI model has brought to light the large and growing AI talent in China. The researchers working for the Chinese startup have shown that human creativity and problem-solving skills can overcome limitations such as access to high-performance hardware. It confirms that the most important resource needed for breakthroughs in AI is the human resource. 

The people of Chinese PRC origin account for 47% of the top 20% AI talent in the world based on undergraduate degree, according to a survey.  Americans make up 18%, Europeans 12% and Indians 5% of the global AI researchers. In terms of the countries they serve, 57% of them work in the United States, 12% in China, 8% in the UK, 4% each in France and Germany and 3% in Canada as of 2022. While the US still has the lion's share of the top talent, its share has declined from 65% in 2019 to 57% in 2022. Marco Polo talent tracker lists Pakistan among a dozen countries for top AI talent in Asia. 

Top Global AI Talent. Marco Polo AI Talent Tracker

More than half (15 out 25) of the institutions (companies and universities) where the top AI researchers work are located in the United States, while 6 are in China. The remaining four are in the UK, Switzerland, Singapore and Canada, according to Marco Polo Global AI Talent Tracker. 

Top AI Talent in Asia Pacific. Source: Marco Polo

The Chinese from PRC dominate the Asia Pacific region with 81.9% of the top AI talent. Indians account for 8.2%, South Korea 4% and "others" 5.8%.  "Others” include Taiwan, Australia, Vietnam, Singapore, Japan, Bangladesh, the Philippines, Malaysia, Pakistan, Mongolia, and Sri Lanka. 

The fact that a number of large language models, including Chinese DeepSeek and Meta's Llama 3, are open source will help develop more global AI talent and spur greater innovation around the world. In the end, it is much more likely that the open source offerings will see greater success than the closed source models like OpenAI's.  

Related Links:

Haq's Musings

South Asia Investor Review

Chinese AI Lab's DeepSeek Shocks Silicon Valley

Pakistan to Develop Urdu Language LLM

Pakistani Military Launches Defense AI Program

Algorithm: Origins of Artificial Intelligence in Islamic Age

Digital Pakistan 2022: Broadband Penetration Soars to 90% of 15+ Po...

STEM Enrollment in Pakistan Exceeds One Million

Digital Public Infrastructure in Pakistan

Generative AI Buzz in Pakistan

Is Pakistan Ready for the AI Revolution?

Growing Presence of Pakistani Women in Science and Technology

Riaz Haq's Youtube Channel

 

PakAlumni Social Network

[Riaz Haq]

Meet Cursor: How Anysphere’s MIT-born AI startup hit a $9.9B valuation in 3 years — TFN. Founded by Sualeh Asif, Pakistani-American cofounder

https://techfundingnews.com/meet-cursor-how-anyspheres-mit-born-ai-...

----------
From Karachi to Silicon Valley: The Remarkable Journey of Saleh Asif

From Karachi to Silicon Valley, Saleh Asif’s journey is a shining example of Pakistani talent on the global stage. A former International Math Olympiad participant and MIT graduate, Saleh is now the co-founder of Cursor AI — a $10 billion platform revolutionizing software development. Backed by global giants like OpenAI and Stripe, his success reflects the untapped potential of Pakistan’s youth in STEM and serves as an inspiring blueprint for the nation’s future.


------

https://www.sualehasif.me/

Sualeh Asif
tweets :)
I’m building Cursor to discover a new way to write code. I owe much of my fun to my friends and MIT. I am extremely excited about the new capabilties of LLMs and applications to code tools.

Previously:

Studied machine learning, number theory, performance engineering, and theater at MIT.
Made early contributions to metaphor, an end-to-end LLM powered search engine (before it was cool).
Represented Pakistan at the IMO 2016-2018. Studied and taught competitive math at the Pakistani math camps.
Worked on translation at IBM Watson ML.
Selected Publications:
"Computing L-Polynomials of Picard Curves from Cartier-Manin Matrices", Mathematics of Computation, 2021.
"Arithmetic Expression Construction", presented at 2020 International Symposium on Algorithms and Computation.
"Tetris is NP-Hard Even with O(1) Columns", presented at 2019 Japan Conference on Discrete and Computational Geometry, Graphs & Games (JCDCG^3).
-- Sualeh