Abstract for: AI sovereignty: How nations' control of agentic AI might evolve and influence national power and strategic competition

AI Sovereignty is the extent to which a nation independently controls its artificial intelligence (AI) technologies, encompassing the nation's capabilities in data, workforce, natural resources, infrastructure, and model training and hosting. Nations might strive for AI sovereignty because they expect it will improve their economic situations, competitive advantage, and overall national power. We review recent developments in AI capabilities to agentic AI. Next, we articulate micro, meso, and macro measures necessary to understand AI sovereignty. We then identify these parameters in a qualitative system model using a causal loop diagram (CLD), paying special attention to competitive dynamics between a nation-state and its strategic adversary. CLD is accompanied by reference modes for key parameter behaviors in the past and into the future. We present four notional, qualitative measures of National Power in agentic AI and two measures for AI Sovereignty. We also demonstrate how strategic competition dynamics between adversaries may seek to pull levers increasing their own national power or degrading their adversaries. We assume a future of regional hegemons versus that AI will be a globally shared utility. Under this premise, Sovereignty and National Power in agentic AI is similar to national power in aircraft. To what extent could a nation design, build, and manufacture (train) and then field at scale (implement) fleets of aircraft as instruments of national power. And to what extent were those capabilities reliant on foreign sources.