US and China Can Cooperate on AI Safety Amidst Distrust

The United States and China can establish practical cooperation on artificial intelligence (AI) safety, even while engaging in significant competition, according to an analysis published on March 11, 2024. This approach suggests that mutual distrust between the two global powers does not preclude the possibility of collaboration on critical AI safety issues. The core argument posits that a dual strategy of competition and cooperation is not only feasible but essential for managing the risks associated with advanced AI development.

The analysis highlights that AI safety encompasses a broad range of concerns, including the potential for misuse, unintended consequences, and the long-term existential risks posed by superintelligent systems. Given the rapid advancements in AI capabilities, particularly in areas like large language models and generative AI, the need for international dialogue and coordinated safety measures becomes increasingly urgent. Both nations are major players in AI research and development, making their engagement crucial for any effective global AI safety framework.

While acknowledging the existing geopolitical friction and strategic competition between Washington and Beijing, the proponents of this dual approach argue that specific, well-defined areas of AI safety can serve as a foundation for collaboration. These areas might include establishing common definitions for AI risks, developing shared methodologies for evaluating AI safety, and potentially agreeing on norms for the responsible development and deployment of AI systems. Such cooperation could be pursued through existing diplomatic channels or through dedicated bilateral working groups focused solely on AI safety.

The benefits of such cooperation extend beyond risk mitigation. By working together on AI safety, the US and China could foster a more stable global AI ecosystem, reduce the likelihood of an AI arms race, and ensure that AI development aligns with human values and societal well-being. This pragmatic approach emphasizes that shared threats, such as the potential for AI-driven instability or catastrophic accidents, can create common ground even between adversaries. The success of this strategy would depend on careful diplomacy, clear communication, and a commitment to transparency in specific AI safety initiatives, thereby building confidence incrementally despite broader geopolitical challenges.