AI GEOPOLITICS
When your theory of mind > their GPU budget
Song-Chun Zhu moved from a rural 1970s China childhood to a leading role in AI. Early hardship sparked his curiosity about how intelligence works.
He studied at Harvard, then taught at Stanford, Ohio State, and UCLA, where he ran a major AI lab.
His work on statistics and pattern recognition influenced today’s systems like ChatGPT, even as he grew sceptical of the neural-network trend.
In 2020, after nearly 30 years in the US, Zhu moved back to China to lead the Beijing Institute for General Artificial Intelligence.
At home, his return was seen as part of China’s AI push; abroad, it raised questions about research ties and geopolitics.
In 2023, he joined China’s top political advisory body and urged the country to treat AI with nuclear-level urgency.
Zhu disagrees with the idea that more data and compute will deliver artificial general intelligence (AGI).
He argues real intelligence needs reasoning, common sense, and adaptability, “small data, big task.”
TL;DR
Zhu’s career highlights a split between scale-driven AI and approaches focused on reasoning.
His return mirrors a broader move of Chinese-born scientists leaving the US amid strain.
The US–China contest now includes not just research output but competing visions for AI.
Brains on the move
His team’s virtual child, “TongTong,” is built to show practical problem-solving rather than test-score gains.
His story sits inside a wider shift. For decades, the US attracted top scientists and powered breakthroughs.
Now, political pressure, tighter funding, and US-China tensions are changing that flow.
China is offering resources and roles to returning and foreign researchers, while many US academics face new limits. Who wins the AI arms race? We still don’t know. Maybe we’re asking the wrong question.
This man is like the cheat code for AI labs.