How AI Could Hollow Out the U.S. Military

CSET’s Emelia Probasco argues that the U.S. military’s growing use of AI could change how decisions get made, and not necessarily for the better. Her op-ed in Foreign Affairs focuses on something less flashy than killer robots: whether soldiers and commanders keep using their own judgment as AI systems become more capable and more embedded in training and operations.

That is the core worry. The military may think it is adopting a tool, but tools shape habits. If the Pentagon keeps integrating AI into more of what it does, the pressure to trust the machine can start to creep into human decision-making itself. Probasco’s point is that this is where the danger lives.

She puts it bluntly: avoiding “Terminator-like dystopias” is not only about controlling AI. It is also about controlling human behavior. That framing matters because the easy version of the debate is all about software safeguards, testing, and rules. Her argument is that those are not enough if the people using the systems stop thinking for themselves.

The broader message is simple and uncomfortable. A military can become dependent on smart systems without ever handing over the formal keys. And if human judgment atrophies, the hollowing out happens quietly, inside the chain of command rather than in some dramatic sci-fi crash.