The New Arms Race Nobody Voted For: AI, Warfare, and the Workers Already Left Behind

On 28 May 2026, Vice President JD Vance told an audience that the development of artificial intelligence worries him most in one specific context: the way it will change warfare. "The thing I worry about most with AI is how it will change warfare," he said. "AI will inevitably change warfare." The remark landed in a news cycle already crowded with hardware procurement announcements, drone procurement contracts, and autonomous-weapons debates at the United Nations. Vance's framing placed him inside a growing consensus among senior Western officials that the integration of machine-learning systems into military operations is not a future scenario but a present one — and one that the policy apparatus is struggling to govern.

The remark was short. Its implications are not.

From Sceptics to Enthusiasts

The trajectory of official attitudes toward military AI has shifted markedly in the past three years. In 2023, the prevailing bureaucratic instinct in NATO capitals was caution: procurement cycles were long, liability frameworks were unclear, and the gap between laboratory performance and battlefield reliability seemed wide. By 2025, that caution had partially dissolved. Part of the shift came from Ukraine, where commercially available quadcopter drones equipped with basic computer-vision targeting — not the sophisticated large-language-model systems being discussed in Silicon Valley — demonstrated decisive tactical impact at relatively low cost. The lesson drawn in defence ministries from Washington to Warsaw was that AI-augmented weapons did not need to be perfect to be operationally significant; they needed only to be cheap enough to be expendable and fast enough to stay ahead of adversary adaptation.

Vance's statement on 28 May fits inside that evolved consensus. He was not raising a hypothetical alarm; he was signalling that the integration of AI into US and allied military planning is now a settled strategic direction, not a question under review. The unanswered question is not whether AI will change warfare, but what the downstream consequences will be for the humans — combatants and civilians alike — caught inside those changes.

The Labour Market in the Same Frame

The conversation about AI and work has rarely been conducted in the same room as the conversation about AI and war, but they share a structural logic. Both concern the displacement of human judgment by algorithmic systems: in one case, the judgment of commanders and soldiers; in the other, the judgment of managers and workers. The displacement is not identical in character, but it is related in origin. The same computational architectures — large neural networks trained on vast datasets, capable of pattern recognition at scale — are being adapted for both military targeting and labour-market management. The dual-use character of the technology has always been present; it is now becoming institutionally visible.

A parallel development, running largely beneath the radar of high-level AI governance discussions, is the reshaping of low-skill labour markets in countries like Poland. On the same day Vance delivered his remarks, Polish-language social media accounts were circulating a different kind of story about algorithmic displacement: videos of young men who have built livelihoods — modest, physically demanding, and entirely unspectacular — around the network of bottle-return machines installed across Polish cities. The machines are a product of Poland's Deposit Return Scheme, which allows consumers to recycle beverage containers for a small monetary return. The videos document people who have made this activity — collecting, sorting, and processing returned bottles — the primary income for their working days.

This is not presented here as a direct analogue to military AI. It is presented as an adjacent phenomenon: in both cases, the structure of economic opportunity is being reorganised by forces that originate in decisions made in technology firms and government ministries far from the shop floor and the battlefield. The workers collecting bottles in Warsaw or Krakow are not being displaced by AI in any direct sense. But they are operating inside an economic environment whose skill demands, wage structures, and social-support architecture have been shaped by a technology transition that received very little public consultation. The same might be said of the soldiers — and the civilians in the zones where they fight — who will be affected by the integration of AI into the decisions that determine who fires, who moves, and who dies.

The Governance Vacuum

The most striking feature of the current moment is the gap between the speed of AI development and the slowness of the governance structures meant to contain it. The United Nations Convention on Certain Conventional Weapons process has been running since 1979 and has produced a series of protocols governing specific weapon categories. The addition of autonomous weapons systems to that agenda has been under discussion since at least 2014; substantive agreement on definitional boundaries remains elusive. At the national level, the picture is more advanced but still uneven. The United States has published a DoD directive on autonomous and semi-autonomous weapons, but critics note that its definitions are sufficiently broad to leave substantial operational discretion at the command level. Several NATO allies have published national AI strategies, but few have enacted binding domestic legislation governing the deployment of AI in lethal decision-making.

Vance's comment did not address governance. It addressed the certainty of change. That certainty is not disputed among serious practitioners. What is disputed is whether the change can be shaped, constrained, and directed — or whether it will be driven primarily by the competitive pressures between state actors who have strong incentives to deploy AI systems before their adversaries do, regardless of the frameworks that nominally govern their use.

The competitive dynamic is not unique to this moment. The introduction of nuclear weapons created a governance challenge that was never fully resolved; arms control agreements moderated the worst incentives but did not eliminate them. The parallel is imperfect — AI systems are not核武器 — but the structural logic is similar: once a technology offers decisive military advantage, the pressure to deploy it operates against the pressure to constrain it, and the latter requires sustained political will that is difficult to maintain in adversarial conditions.

What the Structural Frame Produces

The structural reality is that AI integration into warfare is accelerating along a trajectory shaped primarily by military competition, not by deliberative governance. The competitive logic is not irrational: states that deploy AI systems effectively gain tactical advantages that translate into strategic outcomes. States that decline to deploy them on governance grounds risk being outcompeted by adversaries who do not share those constraints. This logic is understood in every defence ministry that has published an AI strategy, and it is the logic that makes Vance's matter-of-fact remark intelligible.

What the structural frame produces is a set of second-order consequences that the competitive logic systematically underweights. Among them: the risk of accidental escalation when autonomous systems interact in contested airspace or contested waters, a scenario that has been war-gamed by several think tanks and found to be disturbingly plausible. The risk that AI-assisted targeting, by making precision weapons more effective, lowers the political threshold for the use of force — because the cost in allied casualties is reduced, making intervention more politically palatable even when the strategic case is weak. The risk that AI systems in adversarial hands — or loosely controlled non-state actors — introduce instability that the originating states cannot manage.

These second-order risks are not speculative in the sense of being imagined. They are documented in the academic literature on military AI, in defence-policy journals, and in the internal deliberations of governments that do not publish those deliberations. The question is not whether they exist but whether they are being given adequate weight in the decisions being made right now about procurement, deployment, and the terms of engagement.

The Workers, the Soldiers, and the Silence Between Them

Between the high-level AI governance debate and the bottle-collector videos on Polish social media, there is a shared condition: both populations are subject to structural forces they did not initiate and cannot easily resist. The soldier in an AI-assisted command chain faces a decision environment where the speed of algorithmic processing exceeds the speed of human deliberation, and where the ethical frameworks governing that interaction are, at best, incomplete. The worker in a labour market where intermediate skills are being automated finds that the pathways available to them are increasingly concentrated in the physical and interpersonal tasks that machines have not yet mastered — but those pathways are also the lowest-paid and the least insulated from economic volatility.

Neither group asked to be inside this transition. Neither group has much voice in shaping it. The governance frameworks that nominally represent the public interest are operating under conditions of informational asymmetry — the technology firms and defence ministries know more about what the systems do than the publics affected by them do — and under conditions of competitive pressure that reward speed over deliberation.

Vance said on 28 May that he worries about how AI will change warfare. That worry is legitimate and documented. What the public conversation has not yet adequately addressed is the political economy of that worry: who bears the cost of the transition, who captures the benefit, and what the decision-making processes look like when those questions are decided without public deliberation. The soldiers and the workers are already inside the transition. The rest of the conversation is about whether it will be governed — and by whom.

This article drew on the published remarks of Vice President JD Vance on 28 May 2026 and contemporaneous social-media reporting on labour-market conditions in Poland as a parallel structural context. The governance-gap analysis is grounded in publicly available UN and defence-ministry documentation.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://t.me/disclosetv
• https://en.wikipedia.org/wiki/United_Nations_Convention_on_Certain_Conventional_Weapons
• https://en.wikipedia.org/wiki/Lethal_autonomous_weapon
• https://en.wikipedia.org/wiki/AI_military_applications
• https://en.wikipedia.org/wiki/Poland_Deposit_Return_Scheme