Compute inequality is the unequal distribution of computational resources needed to develop or use advanced artificial intelligence. Access to training-scale models, multilingual datasets, and inference application programming interfaces is concentrated in a small number of countries and corporations. section 11.5
Unequal infrastructure limits who can shape, deploy, and benefit from AI. Investment may build national capabilities, but it does not automatically provide broad access; workers and regions supporting the ecosystem can remain overlooked. section 11.5
Responses discussed include:
- Funding, education, and infrastructure-sharing through multilateral initiatives. section 11.5
- Open-source alternatives, public laboratories, and safety-focused funding, despite resource and political constraints. section 11.5
