Nathan Goes to China #3: US-China Relations, the Art of the AI Deal & the Road to Pax Robotica

In this talk

1. 8:57

   Set a shared goal for US-China relations

   Nathan argues that policy should aim for peace and shared prosperity, which he calls Pax Machina or Pax Robotica. He wants people worldwide to benefit from AI, including medical advances and household robots, while avoiding an AI arms race or catastrophic risks.

2. 18:56

   Treat China as a lasting AI power

   Nathan describes China’s return to technological leadership as a historical pattern, not a recent anomaly. He points to China’s AI investment dating to its 2016 five-year plan, its deep research talent, and Chinese firms’ presence on pre-ChatGPT benchmarks. He also notes constraints, including an aging population, succession uncertainty, and limited trust abroad.

3. 1:07:10

   Understand the Chinese view of technology controls

   Nathan says China’s experience of the Century of Humiliation informs its determination to stay near the technology frontier and avoid dependence on foreign suppliers. He argues that US chip controls and rhetoric about winning the AI race can be read in China as efforts to keep it down. He also acknowledges US grievances over IP theft while pointing to Chinese calls for cooperation and open-weight model releases.

4. 1:42:39

   Build practical cooperation and keep AI legible

   Nathan urges the US to build domestic capacity while learning from Chinese strengths in areas such as batteries, solar, electric vehicles, and robotics. He proposes trading access to chips for technology transfer and US manufacturing, clarifying that safety research collaboration is allowed, and exploring shared compute, confidential audits, and software verification. He warns that diverging hardware and AI systems could make each country harder to understand and increase the risk of miscalculation.

5. 3:01:12

   Compete on benefits and limit military risks

   Nathan suggests beginning international talks on autonomous weapons and avoiding AI systems trained to achieve military goals by any means. He proposes positive-sum contests, such as tracking disease cures, battery advances, and traffic safety, to channel competition toward public benefits. He argues that the aim should be global prosperity rather than national dominance.