Opinion
Asha Jadeja is a Silicon Valley venture capitalist.
TL;DR
- The global race to dominate AI requires a parallel race to develop safeguards.
- The US and China face shared risks in AI development and must cooperate.
- A model for cooperation can be found in US-Soviet arms control agreements during the Cold War.
- Proposed safeguards include keeping AI out of nuclear launch decisions and maintaining human control over critical military systems.
- A bilateral hotline for AI-related crises is recommended.
- A narrow agreement governing powerful frontier AI systems, requiring rigorous testing and rapid notification of catastrophic threats, is necessary.
- Cooperation should not involve sharing models, intellectual property, or national security secrets.
- Potential dangers include AI-created bioweapons or cyberattacks on critical infrastructure.
- Coordination between governments and leading AI labs in both countries is crucial.
- The US should lead in AI development while simultaneously prioritizing safety and human control.