AI Policy Approaches in the US and China
Recent actions by the US and China highlight their strategic approaches to AI, treating it as a critical element of national security.
China's Approach
- Beijing has restricted overseas travel for top AI engineers from major firms such as DeepSeek and Alibaba.
- This move indicates a view of AI expertise as sensitive, akin to nuclear knowledge.
US Approach
- The US has directed companies like Anthropic to suspend frontier models for foreign nationals, including their employees.
- A widely used commercial AI product was removed from the market, highlighting the treatment of AI as a security concern.
Underlying Narrative
Both countries see AI as part of a zero-sum arms race, reflecting national security perspectives that liken AI to weapons of mass destruction.
- This narrative has shaped policy instruments such as export controls and chip hardware tracking.
- However, AI differs from nuclear technology as it is widely reproducible and constantly evolving.
Alternative Framing: Geopolitical Innovation Race
An alternative view considers AI as part of a global innovation race with economic and market incentives.
- Key policy instruments include compute investment, skilling, and strategic partnerships.
- Such an approach emphasizes collaboration and competition among public and private sectors worldwide.
India's Strategic Response
As a middle power, India should embrace AI as a general-purpose technology and focus on widespread diffusion.
- Adopt regulatory clarity and sector-specific deployment strategies to enhance AI adoption.
- Foster coalitions, especially with entities like the EU, to share development costs and create reliable supply chains.
- Encourage open-source and open-standard models to ensure resilience against unilateral control.
Example of AI Integration
The Supreme Court of India's draft guidelines on AI tool usage in courts exemplify potential AI utility in public services.
Priorities for Maximizing AI Benefits
- Diffusion over Denial: Prioritize infrastructure and adoption over restricting AI distribution.
- Plurilateral Resilience: Build alliances to leverage complementary strengths.
- Open over Closed: Open models and standards enhance resilience and prevent dependency on controlled systems.
