Jeffrey Ding on Technology Diffusion and US-China AI Competition

Jeffrey Ding's Theory on Tech Supremacy

• 💡 Economic and technological superpower status depends less on being first to invent new technologies and more on two factors.
• 🔑 The first factor is developing General Purpose Technologies (GPTs), which are innovations that enhance productivity and efficiency across all economic sectors.
• 🎯 The second crucial factor is a country's ability to effectively diffuse and spread these GPTs throughout its entire economy.

US vs. China in AI Competition

• 🇨🇳 China's perceived "AI prowess" is an illusion, as it remains a "fast follower" behind the US in artificial intelligence development.
• ⚠️ Challenges for China's AI include data limitations, US export controls on semiconductors, and government censorship impacting large language models.
• 📊 China exhibits a "diffusion deficit", where its innovation capacity in science and technology significantly outpaces its ability to spread those technologies broadly.

Key Factors for Technology Diffusion

• 🧠 Human capital development is paramount, focusing on a wide base of engineering skills and technical literacy, not just "heroic inventors."
• 🎓 A decentralized university system like that in the US fosters adaptability and a diverse skill base, unlike China's top-down approach.
• 🌐 Trade openness and strong university-industry linkages are vital for plugging into global innovation networks and facilitating technology transfer.

AI as a General Purpose Technology

• 🚀 AI demonstrates characteristics of a GPT, including continual improvement, diverse use cases across industries, and broad patent citations.
• ⏳ The full economic impact of AI, like past GPTs such as electricity, will likely take multiple decades to diffuse and require significant organizational adaptation.
• 💡 Advances in AI can transform industries like electric vehicles, but the reverse is not true, indicating AI's foundational nature.

Policy Recommendations and Risks

• ✅ Effective policy should prioritize STEM workforce development and support institutions like community colleges to broaden AI engineering skills.
• 📈 The current US "Fortress America" approach, relying on export controls, is based on a mistaken assessment of technological leadership.
• 🚨 A "run faster" model, focusing on diffusion and internal strength, is preferable, as a weak or "peaking" China could pose a greater national security threat.