China versus the U.S.

The AI race between the United States and China is largely determined by their control over 90% of global computing power, 70%-80% of AI investment, and a large number of talented AI researchers. This dominance leaves the rest of the world dependent on these two powers for access to AI systems, which can be used as leverage by the U.S. and China.

This puts many countries in an uncomfortable position, particularly the Global South and middle powers, who face potential disruptions to their jobs and society due to AI advancements. Three trends exacerbate this situation: U.S. companies are adopting a more managed-access approach to their technology, compute constraints are causing demand to outpace supply, and both the U.S. and China are increasingly asserting control over access to these systems.

This could result in a handful of wealthy companies and countries having the power to determine who can access AI technology. Despite advancements in AI models, they still reflect biases rooted in the values, voices, languages, and cultures of WEIRD societies, which only represent 14%-15% of the world's population. This limits the representation of marginalized populations, including those with disabilities, in AI systems.

The first U.N. General Assembly resolutions on AI addressed the need for safe, secure, trustworthy AI systems and the application of human rights to AI. However, the resolutions were led by the U.S. and China and went it alone, without broader international cooperation. The resolutions included language affirming human rights throughout the AI life cycle, but the application of AI in military and humanitarian contexts raises further questions about the use of machine learning systems in direct control over life-or-death situations.