Beast in the Machine

George M. Dougherty is a defense engineer and technologist who has spent his career inside the U.S. military’s robotics and autonomy programs, and Beast in the Machine is his effort to put what he has seen into a single accessible book. It is aimed at readers who want to understand how AI and robots are actually being absorbed into modern militaries, rather than what futurists imagine they might one day become.

His central argument is that the transformation underway is structural rather than spectacular. Autonomy is not arriving as a single Terminator-style breakthrough but as a steady accretion of unmanned vehicles, machine-vision targeting, networked logistics, and decision-support systems that compound on each other. Dougherty treats the loss of full human control over individual engagements as a near-certainty wherever the tempo of combat exceeds human reaction time, and he is candid about the technical limits — fragile sensors, jammable links, brittle models in contested electromagnetic environments — that will shape how fast that ceiling rises.

The body of the book is its most concrete. Dougherty walks through ground combat robots, uncrewed surface vessels, undersea autonomous platforms, swarming concepts, and human–machine teaming across services. He draws on programs such as the Army’s Robotic Combat Vehicle, the Pentagon’s Replicator initiative for attritable drones, and the Air Force’s Collaborative Combat Aircraft effort pairing manned fighters with autonomous wingmen. The war in Ukraine runs through the book as the most useful case study available — first-person-view strike drones, sea drones in the Black Sea, ad hoc counter-drone improvisation — and Dougherty uses it to show how doctrine drafted before 2022 has already been overtaken in the field. He also handles the harder, less photogenic parts: test and evaluation regimes, training data quality, adversarial deception against computer vision, and the legal scaffolding around lethal autonomy, including the long-running debate over meaningful human control and the Pentagon’s revised Directive 3000.09.

Beast in the Machine sits as a practitioner’s primer in an increasingly crowded field. It is less polemical than Paul Scharre’s Army of None and less abstract than the academic literature on autonomous weapons, with its closest neighbours being Christian Brose’s The Kill Chain and recent writing on the U.S.–China military-technical competition. Readers already deep in the subject will find some of the framing familiar, but the value of Dougherty’s account is that he writes from inside the engineering problem and treats robotics, AI, and command as one connected system rather than three separate arguments. For policymakers, journalists, and technologists trying to understand how AI is actually reaching the battlefield — and what still stands in its way — it offers a clear and grounded starting point.