War as Phenomenon, War as Practice: From the Psychology of Conflict to the Machines Fighting It

Every generation that lives through peace convinces itself war is an aberration — a breakdown, a failure of otherwise rational systems. Every generation that lives through war discovers instead how normal it is: how easily societies organize around it, justify it, and even find meaning in it. This tension — war as disorder versus war as recurring social technology — is where any serious inquiry into conflict has to start.

Why Wars Happen: Three Ways of Answering an Old Question

International relations theory has spent a century circling the same question — why do states (and peoples) fight? — and arrived at answers that aren't mutually exclusive so much as differently focused.

Realism treats war as the natural output of an anarchic international system. There's no global sovereign to enforce order, so states are left to guarantee their own survival, and that logic — the security dilemma — pushes even defensively-minded states into arms races and preemptive conflict. War, in this view, isn't pathological; it's what you'd expect from rational actors operating without a referee.

Liberalism pushes back by pointing to trade interdependence, international institutions, and democratic norms as forces that can genuinely dampen the realist logic — the "democratic peace" thesis being the most cited (and most contested) example. War here is a policy failure, something institutions and incentives can reduce, even if not eliminate.

Constructivism — the most interesting lens for a psychology-adjacent read on war — rejects the idea that state interests are simply given by material conditions like power, wealth, or geography. Instead, constructivists argue that interests and identities are socially constructed: through discourse, shared norms, historical narrative, and the ongoing performance of national identity. Alexander Wendt's famous line "anarchy is what states make of it" captures it well — the same distribution of power can produce cooperation or hostility depending on how actors have come to understand each other. A rival isn't dangerous because of tank counts alone; it's dangerous because both sides have narrated each other into that role, often over decades, through media, education, and political rhetoric.

This matters for understanding why people — not just states — go to war. Constructivism explains how a population can be persuaded that a war is not just necessary but righteous: through the construction of in-group/out-group identity, the narration of historical grievance, and the reframing of violence as defense or restoration rather than aggression. Nationalism, in this frame, isn't a byproduct of war — it's often a precondition manufactured well before the first shot.

The Psychology Underneath the Theory

IR theory explains why states go to war; psychology helps explain why individuals comply, volunteer, or even welcome it.

A few recurring mechanisms show up across conflicts and eras. Perceived external threat — real or constructed — reliably increases in-group solidarity and out-group hostility, which is one reason leaders in political trouble sometimes benefit from external conflict. Psychologist Albert Bandura's work on moral disengagement describes the cognitive mechanisms — euphemistic language, diffusion of responsibility, dehumanization of the enemy — that let ordinary people participate in or support violence without experiencing it as a moral violation. Research on identity fusion links extreme willingness to sacrifice for a group to a felt sense that personal and group identity have merged. And conflicts are frequently driven less by rational cost-benefit calculation than by the desire to avoid a feared loss of status, territory, or honor — a dynamic that purely rational-actor models tend to underweight.

Put together, constructivism explains the macro-level manufacturing of hostile identity; psychology explains the individual-level mechanisms that make that identity actionable. War is rarely just a decision made in a war room — it's a story a society has been told well enough, for long enough, that fighting starts to feel inevitable rather than chosen.

Then the Technology Changed the Story Itself

If war has always been partly a narrative project, the tools now doing the fighting are reshaping what that narrative even needs to include — because increasingly, a war can be fought with a shrinking number of humans directly in harm's way, which changes both the political cost of initiating conflict and the psychological experience of participating in it.

The battlefield reality in 2026 is already striking. In late January, China's National University of Defence Technology publicly demonstrated a single soldier operating a formation of 200 autonomous drones. Meanwhile, the war in Ukraine has become the clearest live laboratory for this technology: drones accounted for 96% of Russia's battlefield casualties in March 2026 alone. Ukraine's own officials describe the frontier candidly — the country's Deputy Defense Minister has said fully autonomous weapons don't yet exist as a category, but have been "partially implemented in some devices."

The trend line is unmistakable even without full autonomy formally declared. A Ukrainian first-person-view drone reportedly used an AI-assisted guidance system to strike a Russian tank after losing its communications link with its human operator — exactly the scenario militaries have been building toward, since autonomous functions let drones keep operating despite jamming and loss of live communication. Analysts covering the war note that given current production pace, a battlefield with hundreds to thousands of fully autonomous drones is plausible within just a few years. The more unsettling next step already being discussed is drone swarms that hunt in packs and autonomously adjust behavior by communicating with each other, posing a much harder challenge for maintaining meaningful human control.

Economics are driving adoption as much as capability. AI-assisted terminal guidance has reportedly pushed first-person-view drone mission success rates from roughly 15% to around 60%, a jump in cost-effectiveness that helps explain why fielding capable battlefield AI at around $1,000 per drone rather than $100,000 could increase the number of autonomous weapons a country can deploy by roughly a hundredfold. That cost collapse means the technology isn't staying confined to state militaries — Colombian armed groups FARC and the ELN have reportedly adopted drone warfare directly inspired by innovations from Ukraine, with dozens of recorded attacks in the past two years.

State investment reflects how central this has become to defense planning. The Pentagon's FY2026 budget request includes $13.4 billion specifically for AI-facilitated autonomous systems, with $9.4 billion earmarked for unmanned aerial vehicles alone, alongside a further $3.1 billion allocated for counter-drone defense — underscoring that this is becoming an offense-defense arms race in its own right. Chinese military planning describes this in explicit doctrinal terms: the PLA is pursuing what it calls "intelligentized warfare," integrating AI across land, sea, air, space and cyber domains, with a stated ambition to reach that status by 2027. U.S. military leadership has stopped hedging on the direction of travel — Joint Chiefs Chairman Gen. Dan Caine said in April 2026 that autonomous weapons are going to be "a key and essential part of everything" the U.S. military does, with active work underway on applying autonomy to both drones and command-and-control systems.

The legal and ethical scaffolding is not keeping pace. Existing international humanitarian law still requires human judgment at key points of lethal decision-making, and military doctrine formally maintains that fully autonomous weapons should not operate without human input at intervention points meant to keep targeting within the bounds of the Geneva Conventions. But that formal constraint is under real strain, with recent watchdog reporting raising serious concerns about how few guardrails currently exist for the pace of investment in advanced autonomous weaponry.

What to Expect Next

A few trajectories look reasonably safe to project from where things stand now. Swarm autonomy is becoming the frontier rather than the exception — the move from single AI-guided drones to communicating swarms is the direction every major program is converging on, and it's the point where "human in the loop" oversight becomes genuinely difficult to enforce in real time. Cost collapse will keep driving proliferation beyond states, with non-state and irregular actors adopting this technology faster than treaties can regulate it. Expect an accelerating offense-defense spiral in counter-drone technology, essentially recreating the anarchy and security-dilemma dynamics realists describe, but at machine speed. And international norms will likely keep lagging capability: discussions on lethal autonomous weapons continue at forums like the UN's Convention on Certain Conventional Weapons, but binding, enforceable restrictions remain distant while battlefield deployment accelerates.

There's a genuinely uncomfortable constructivist point buried in all of this. If part of what makes war psychologically and politically possible is the human cost of initiating it — the domestic grief, the conscription politics, the soldier's own moral experience — then a shift toward machine-fought war doesn't just change tactics. It potentially lowers the narrative and psychological threshold for starting conflicts in the first place, because the story a society has to tell itself to go to war gets much easier to write when fewer of its own people are the ones telling it from the front lines.

That may be the real story here — not that drones are new, but that they may be quietly rewriting the social and psychological preconditions for war itself.