Skip to content

China's Not the Problem. We Are.

Source: China's Not the Problem. We Are. by Ross Douthat (NYT Opinion — "Interesting Times" podcast) with Kyle Chan

TL;DR

A New York Times podcast transcript turned column. Kyle Chan (Brookings Institution) argues that the US and China have fundamentally different AI goals, making it difficult to frame the situation as a simple "race." The US focuses on frontier breakthroughs and AGI, while China's approach is more state-directed with different priorities. The conversation explores the Cold War atmosphere around the Trump-Xi meeting in Beijing and what "winning" even means in the AI context. The title encapsulates the central argument: America's own limitations and blind spots may be a bigger obstacle than Chinese competition.

Not a Race

The dominant framing of US-China AI competition is that of a race — two runners sprinting towards the same finish line, with the winner capturing the prize of global technological leadership (and perhaps AGI itself). Kyle Chan challenges this framing at its root:

The US and China have fundamentally different AI goals.

The US approach — driven by private sector frontier labs (OpenAI, Anthropic, Google DeepMind, Meta) — is oriented toward capability maximisation. The goal is to push the frontier: larger models, longer contexts, more general intelligence, and ultimately AGI. Investment is measured in dollars, compute, and researcher headcount.

China's approach is more state-directed and application-oriented. The Chinese government prioritises AI applications that serve state interests: surveillance, social control, industrial automation, and military applications. The goal is not necessarily to reach AGI first, but to deploy AI effectively within the Party's governance framework.

Different Metrics of Success

If the US "wins" by achieving AGI first, but China "wins" by integrating AI into state capacity more effectively, then the race metaphor collapses. They are running towards different finish lines.

Chan's argument implies that the US could "win" the capability race and still "lose" the geopolitical competition if China deploys AI more effectively for state purposes.

America's Blind Spots

The title's provocation — "China's Not the Problem. We Are." — points to several US vulnerabilities that Chan argues are more significant than Chinese competition:

1. Political Instability. The US political system's ability to sustain long-term technology policy is questionable. Every administration reverses the previous one's priorities. China's one-party system provides policy continuity that the US cannot match.

2. Infrastructure Constraints. Building AI data centres and the energy infrastructure to power them faces local opposition, permitting delays, and grid interconnection challenges in the US. China can build faster.

3. Talent Pipeline. US reliance on immigrant talent is politically contested. Restrictive immigration policies could choke the US AI sector's most important input.

4. Short-Termism. Public company pressure for quarterly results may lead US firms to underinvest in long-term, high-risk AI research. Chinese state-backed firms can take longer time horizons.

The Cold War Atmosphere

The podcast was recorded against the backdrop of the Trump-Xi meeting in Beijing — a summit that took place in an atmosphere of mutual suspicion and strategic competition reminiscent of the Cold War. Both sides have imposed technology export controls, investment screening, and talent restrictions.

But Chan and Douthat question whether the Cold War framing is productive. The US-Soviet competition was between two systems with fundamentally incompatible ideologies but roughly comparable technological capacity. The US-China competition is between deeply economically intertwined powers with overlapping interests (climate, trade, pandemic preparedness) and asymmetric technological strengths.

What Does "Winning" Mean?

The conversation closes on an unresolved but essential question: what does winning the AI race actually mean?

  • Is winning achieving AGI first?
  • Is winning deploying AI most effectively across the economy?
  • Is winning building an AI system that aligns with liberal democratic values?
  • Is winning avoiding catastrophic AI outcomes?

Depending on the answer, the US may be ahead, behind, or playing a completely different game from China.

Key Takeaways

  1. The "AI race" frame is misleading — the US and China have fundamentally different AI goals.
  2. America's weaknesses are the bigger problem — political instability, infrastructure constraints, talent pipeline risks, and short-termism may matter more than Chinese competition.
  3. The Cold War framing obscures interdependence — the US and China remain deeply economically entangled, even as they compete.
  4. "Winning" is undefined — without clarity on what victory looks like, it is impossible to know whether the US is ahead or behind.
  5. Self-awareness is the first step — recognising America's own limitations is more productive than fixating on Chinese competition.