Aug 18, 2026 · 23 min
America’s AI lead depends on adoption, not models alone
The Hidden Weakness in America's AI Lead (#315)
The United States may build the strongest AI systems yet lose strategic ground if organizations, governments, and the public cannot trust and deploy them quickly.
- 1AI’s nonlinear progress makes long-term forecasting difficult and turns rapid adaptation into a national advantage.
- 2Successful adoption depends on leaders using AI themselves and redesigning workflows, not assembling disconnected model options.
- 3Trust is the central constraint on whether AI strengthens public services, national security, and America’s position relative to China.
Don't miss
Mulligan connects a Pennsylvania public-sector deployment’s eight-hour weekly savings to a larger argument for redesigning work around AI.
The brief
Katrina Mulligan, OpenAI’s Head of National Security Partnerships, argues that America’s AI race is not only about model capability; it is about whether institutions can trust and adopt the technology.
Her move from the Pentagon to OpenAI exposed a radically faster development culture, where recursive self-improvement and nonlinear progress make conventional forecasts unreliable.
The larger transformation extends beyond ChatGPT: reasoning models and enterprise applications could reshape government work, but only if leaders redesign workflows instead of adding chatbots.
A Pennsylvania deployment saved public servants an average of eight hours per week, illustrating the immediate productivity gains—and the bigger opportunity to rebuild public services around AI.
Mulligan’s central warning is geopolitical: the United States can lead in model development yet fall behind China and others if trust slows adoption across society.
Featuring
OpenAI
ChatGPT
Pennsylvania
United StatesListen to the full episode and explore every guest, topic, and moment on PodLume.