by Nil
Katrina Manson’s Project Maven: A Marine Colonel, His Team, and the Dawn of AI Warfare is at once a work of embedded journalism, a character study, and an unsettling meditation on the technological transformation of organised violence. It is also a book that resists easy classification: part insider chronicle, part moral inquiry, part institutional critique. Manson’s central achievement lies in her ability to render the abstraction of “AI warfare” into a dense web of personalities, bureaucratic failures, ethical ambiguities, and geopolitical anxieties. Yet the book’s most provocative claim—that artificial intelligence may both mitigate and intensify the brutality of war—remains suspended in tension, never fully resolved, and perhaps deliberately so.
From the outset, Manson frames AI warfare not as a rupture but as an extension of long-standing frustrations with human fallibility. Colonel Drew Cukor’s stark assertion that “the problem with war… had always been the humans. ‘They’re materially corrupt, inefficient, and they get tired’” (p. 9) establishes a foundational premise: that automation promises not merely efficiency but a kind of moral correction. This premise is reinforced through the invocation of AlphaGo, where “a human alone would never be enough. A human plus a machine would always be better” (p. 7). Yet Manson subtly destabilises this technocratic optimism by juxtaposing it with the blunt admission from a JSOC veteran that “we were really good at killing people, but it didn’t get us very far” (p. 9). The implication is clear: improving the mechanics of killing does not necessarily address the strategic or moral bankruptcy of modern warfare.
The book’s critique of the Pentagon’s institutional dysfunction is among its most compelling strands. Manson paints a portrait of a military apparatus drowning in its own administrative inertia, where “more Pentagon employees work in acquisition and procurement than in the entire Marine Corps” (p. 37), and where even basic software systems are derided as “crap… just horrible” (p. 21). The grotesque irony that “we’ve killed more people on Office than you’d ever imagine” (p. 18) encapsulates a system in which lethality is mediated through tools never designed for war. Here, Manson’s narrative acquires a biting edge: the world’s most advanced military power is revealed to be technologically sophisticated in weaponry but archaic in information management. The resulting gap—“America’s munitions might have become precise, but the information that fed them was not” (p. 21)—becomes the justificatory space into which AI steps.
Yet if AI emerges as a solution, it is one born of crisis rather than clarity. The spectre of China looms large, shaping both the urgency and the rhetoric of Project Maven. Cukor’s warning that “if the enemy has AI and we don’t… we’re not even going to understand what went wrong” (p. 147) reframes technological development as existential necessity. Manson captures how this framing collapses ethical deliberation into strategic inevitability: “we have to do it or we will fall prey to the Chinese” (p. 150). The book is particularly incisive in exposing how this logic forecloses dissent. The comparison between US and Chinese procurement systems—where American efforts stall amid internal resistance while China’s advance is portrayed as seamless—creates a narrative of competitive determinism that risks normalising escalation.
Manson does not allow the reader to remain within the abstractions of strategy. The sections on civilian casualties puncture any illusion that improved targeting necessarily translates into humane warfare. The chilling anecdote of a drone strike that killed farmers mistaken for militants—“it turned out to be a group of farmers resting at the end of a long day” (p. 132)—serves as a reminder that the epistemic limits of war persist regardless of technological sophistication. The haunting testimony, “we kill the wrong people all the time… the machine can’t be worse than a human” (pp. 123–124), encapsulates the moral gamble at the heart of AI warfare: that incremental improvement justifies systemic risk. Yet Manson complicates this argument by documenting the scale of civilian harm, with estimates reaching “possibly… more than 8,000” deaths (p. 199). The tension between aspiration and outcome is never resolved, only deepened.
The book’s exploration of Silicon Valley’s entanglement with military power introduces another layer of complexity. The Google controversy is presented not merely as a corporate ethics dispute but as a cultural clash between technologists and the state. Internal dissenters insist that “this is going to be used to kill people” (p. 111), rejecting what they perceive as euphemistic framing. Yet the eventual outcome—“the algorithms created by Google went to war in Afghanistan after all” (p. 115)—suggests that moral resistance may alter appearances without fundamentally changing trajectories. Manson’s portrayal of this episode is particularly sharp in highlighting the limits of corporate ethics in the face of geopolitical imperatives.
If Silicon Valley appears conflicted, figures like Alex Karp embody a more unapologetic fusion of technology and militarism. His declaration that adversaries “need to wake up scared and go to bed scared” (p. 156) reveals a worldview in which deterrence is indistinguishable from intimidation. Manson does not caricature Karp so much as let his rhetoric expose itself, culminating in moments of near absurdity, such as his fantasy of weaponising drones against financial short sellers (p. 161). These passages risk veering into spectacle, yet they serve an analytical purpose: to illustrate how the language of innovation can mask, or even celebrate, the expansion of violence.
At the operational level, the book’s treatment of the “kill chain” is both technically detailed and morally unsettling. The ambition to “shorten the kill chain” (p. 227) reflects a broader drive towards speed, efficiency, and automation. Yet this acceleration raises profound ethical questions. The insistence that “it’s up to humans to decide whether to pull the trigger” (p. 237) functions as a rhetorical safeguard, but Manson’s own evidence suggests that human judgement is increasingly mediated—and potentially overridden—by algorithmic outputs. The description of autonomous systems that can “pick out… target[s]… and fire at it too” (pp. 330–331) challenges the plausibility of maintaining meaningful human control.
Nowhere is this tension more vividly illustrated than in the Ukraine war, which Manson presents as a proving ground for AI-enabled warfare. The rapid adaptation of algorithms—“it took little more than two weeks” (p. 268)—and the dramatic increase in targeting capacity—“seventy targets a day” (p. 275)—demonstrate the transformative potential of these systems. Yet the metaphor offered by one participant—“it’s like being a puppet master but the puppet has a brain” (p. 278)—captures the ambiguity of control. The suggestion that humans themselves might become the puppets introduces a note of quiet dread.
Manson’s treatment of data—the lifeblood of AI—further complicates the narrative of technological mastery. The descent into “data hell” (pp. 209–210), where poor-quality inputs undermine model performance, reveals the fragility of these systems. The blunt admission, “we fucked up data so bad” (p. 207), punctures any illusion of seamless innovation. At the same time, Cukor’s assertion that “the data is more valuable than the algorithm” (p. 216) points to a deeper transformation: the commodification of information as a strategic asset.
The book’s most profound contribution lies in its exploration of the moral paradox of AI warfare. Manson refuses to offer a definitive judgement, instead presenting a series of conflicting perspectives. On one hand, AI is framed as a tool to “save lives” by improving targeting accuracy (p. 338). On the other, it risks amplifying harm through increased efficiency, echoing the logic of the Jevons Paradox (pp. 343–344). The haunting question—“who is responsible when AI does the targeting… and you put a missile into a van with a family?” (pp. 200–201)—remains unanswered, and perhaps unanswerable.
Cukor himself emerges as a figure of contradictions: visionary and obsessive, principled and ruthless. Described as “a psychopath… in the best way” (p. 62), he embodies the intensity and moral ambiguity of the project he leads. His admission that “taking a human life… takes a piece of your soul” (p. 341) introduces a note of vulnerability that complicates the otherwise technocratic narrative. Yet his insistence, “I’d do it again, in the same way” (p. 351), underscores the persistence of conviction even in the face of doubt.
Project Maven is less a story about the triumph of technology than about the unresolved tensions it exposes. Manson shows that AI does not eliminate the fog of war; it refracts it through new forms of uncertainty, responsibility, and power. The book’s closing reflections—“are we the best custodians of it?” (p. 354)—linger as an open question. If the answer remains unclear, it is because Manson has succeeded in demonstrating that the stakes are not merely technical or strategic, but fundamentally human.
