AI War Won’t Be Won: Seven Paths to a Global Public Good

Gary Marcus argues that China has nearly caught up to the US in AI and that a decisive US victory in the AI race is unlikely. He outlines seven strategic options for policymakers: do nothing and let OpenAI/Anthropic stand on their own; outlaw open source; build a regulatory moat; bail out the big AI labs and turn them into national labs; ban Chinese models; buy out the big AI labs and convert them into national or international labs; and finally give up on a zero-sum win and instead invest in making AI a global public good, potentially via an international CERN-for-AI-style mission. He notes disruptive developments like Moonshot.AI’s Kimi K3 (open-weight) and GLM 5.2 from Z.ai, which challenge current US business models, and suggests a non-zero-sum, globally collaborative approach could better benefit humanity than a pure competition.