Carson's central argument is that fatalism about an unstoppable AI arms race is the most dangerous idea, as the West retains control over critical AI chips. He advocates for mandatory testing of frontier models and treating AI as a product, not a person, to ensure legal accountability. Duggar counters by highlighting the opacity of neural networks, using Palantir's probabilistic heat maps as an example where a model can score a target as 0.73 likely a combatant, yet no one can explain the reasoning or be held accountable when it is wrong. The discussion explores the tension between AI developers like Anthropic, who are morally opposed to lethal autonomous weapons, and the Pentagon's use of such technology. Carson argues for engaging China in dialogue and using chip leverage to avoid a destabilizing arms race, while also criticizing the American tendency to substitute capital for labor with technical solutions, insisting that wars are ultimately won by people. The conversation concludes with a call to upskill Congress and rebuild public trust in the AI industry.