The Algorithm and the Battlefield: Jean-Marc Rickli and the New Grammar of War

By Matthew Parish, Associate Editor

Sunday 30 August 2026

There is a recurrent temptation in thinking about warfare to believe that the next technology will change everything. The machine gun would abolish manoeuvre. Strategic bombing would make armies obsolete. Nuclear weapons would make conventional war irrational. Precision-guided munitions would eliminate the fog of war. Cyber warfare would render tanks and artillery antiquated. Now artificial intelligence is presented as the technology that may finally transform not merely the instruments with which human beings fight but the nature of warfare itself.

Dr Jean-Marc Rickli, Head of Global and Emerging Risks at the Geneva Centre for Security Policy, is considerably more sophisticated than most technological prophets. His work does not simply predict a science-fiction battlefield populated by autonomous machines. Instead he has developed a broader argument about the consequences of artificial intelligence, autonomous weapons, drone swarms, neurotechnology, cognitive warfare and other emerging technologies for the organisation of political violence.

His central concern might be expressed in deceptively simple terms: technology is lowering the cost of power.

That proposition deserves to be taken extremely seriously. But it also deserves qualification — particularly after four and a half years of full-scale war in Ukraine, which has become perhaps the greatest experimental laboratory in military technology since the Second World War.

Rickli’s work sits at the intersection of strategic studies and technological foresight. He is interested not merely in weapons but in what happens when technological capabilities previously confined to governments become available to corporations, irregular organisations and ultimately individuals. His writings describe a world in which emerging technologies progressively “democratise” access to coercive power. The monopoly of the state over sophisticated violence becomes increasingly porous.

The implications are profound.

A twentieth-century air force required airfields, pilots, mechanics, aircraft factories, enormous budgets and elaborate command structures. A twenty-first-century organisation may be able to achieve some of the same tactical effects with commercially manufactured drones, software engineers, satellite communications and components ordered through global supply chains.

A cruise missile once represented the accumulated industrial capacity of a great power. A long-range attack drone may cost tens of thousands of dollars.

The direction of travel is therefore unmistakable.

Rickli has been warning about one particularly important consequence for years — swarming.

A swarm is qualitatively different from merely possessing a large number of drones. Once dozens, hundreds or potentially thousands of autonomous systems cooperate, the human operator cannot individually control them. Algorithms must allocate tasks, coordinate movements, respond to threats and reorganise the swarm when individual machines are destroyed. Rickli argued as early as 2020 that the combination of autonomous algorithms and drone swarms offered enormous scalability and could overwhelm existing defensive systems.

The military logic is straightforward. Every defensive system possesses a finite capacity. A radar can track only so many targets. An interceptor battery contains only so many missiles. A soldier can engage only so many drones simultaneously.

If the attacker can manufacture autonomous weapons more cheaply than the defender can destroy them, quantity becomes a strategic weapon.

This phenomenon is already visible in Ukraine. Electronic warfare has produced an evolutionary contest between drone operators and jamming systems. Fibre-optic drones have emerged precisely because radio-frequency control can be disrupted. Autonomous navigation and target-recognition technologies promise another solution — removing the requirement for continuous communication between operator and machine. Rickli has identified precisely this direction of technological development and anticipates increasingly autonomous swarms designed to saturate enemy defences.

Yet here Ukraine also supplies the first important qualification to technological determinism.

Technology matters enormously. Technology alone does not win wars.

Indeed Rickli himself, together with Federico Mantellassi, has made this point particularly well. Their study of the Russo-Ukrainian war concludes that although drones, artificial intelligence, cyber capabilities and the digital information environment are changing warfare, the traditional determinants of military power have stubbornly refused to disappear. Soldiers matter. Ammunition matters. Artillery matters. Tanks matter. Logistics matters. Organisation matters. Above all, the capacity of institutions to learn and adapt matters.

This observation may prove more important than the more spectacular predictions about autonomous warfare.

Russia invaded Ukraine in February 2022 possessing many impressive technologies. What it initially lacked was competent organisation.

The failure was spectacular.

Columns of armoured vehicles became immobilised. Logistics collapsed. Command structures proved inflexible. Intelligence was poor. Political assumptions substituted for military planning. Ukraine, meanwhile, combined comparatively modest resources with extraordinary institutional improvisation.

War remained what Clausewitz understood it to be — a contest between organised political communities attempting to impose their wills upon one another. Silicon had not repealed Clausewitz. And this exposes a recurring difficulty in the emerging-technologies school of strategic thought. Technological capability is not the same thing as military effectiveness.

A drone is useful because somebody knows where to send it.

An artificial intelligence system is useful because somebody has supplied it with relevant information. A targeting algorithm is useful because intelligence networks identify targets.

A swarm is useful because military doctrine explains what the swarm is supposed to accomplish.

The decisive military technology may therefore not be artificial intelligence itself but the institutional machinery within which artificial intelligence is embedded.

Ukraine has demonstrated this repeatedly. Her extraordinary drone ecosystem did not emerge because Kyiv possessed some secret technological breakthrough unavailable elsewhere. It emerged from thousands of engineers, soldiers, volunteers, entrepreneurs, workshops and military units continually exchanging battlefield experience and modifying equipment.

Innovation became evolutionary.

A drone design succeeded for several weeks. Russia developed a countermeasure. Ukraine modified the drone. Russia changed its electronic warfare system. Ukraine altered frequencies, navigation methods or communications. Russia adapted again.

The decisive advantage was not technological superiority in the abstract. It was the speed of adaptation. This distinction matters when assessing Rickli’s darker warnings about artificial intelligence.

His recent work has increasingly focused upon agentic AI — systems capable not merely of responding to instructions but of pursuing objectives through sequences of actions with comparatively limited human supervision. In military applications, such systems might gather intelligence, formulate plans, select tools, coordinate other systems and adjust their behaviour dynamically. Rickli argues that their deployment creates substantial risks involving proliferation, accountability, strategic stability and unintended behaviour.

These concerns are entirely legitimate. Yet there is a danger of treating autonomy as though it were a binary concept. It is not.

Weapons have possessed varying degrees of autonomous behaviour for decades. Mines explode without asking permission. Torpedoes acquire targets. Air-defence systems automatically track incoming threats. Fire-and-forget missiles pursue targets after launch.

The meaningful question is therefore not whether machines should ever make decisions. Machines already do.

The question is which decisions should machines be permitted to make, under what circumstances, according to which rules and subject to whose responsibility. That distinction is crucial.

A defensive system automatically destroying an incoming missile travelling towards a city is morally and strategically different from an autonomous drone searching an urban district for people whose characteristics correspond to an algorithmic target profile.

The word “autonomy” obscures this distinction unless it is carefully unpacked. Rickli is nevertheless right to identify a profound strategic danger arising from speed. Human beings are slow. Machines are fast.

During the Cold War there were already terrifying moments in which political leaders had minutes to determine whether indications of nuclear attack were genuine. Artificial intelligence could compress decision-making cycles still further.

Imagine competing autonomous military systems operating across cyber networks, satellite constellations, electronic warfare environments and thousands of unmanned platforms.

One system identifies behaviour interpreted as preparation for attack. It responds.

The opposing system interprets that response as confirmation of hostile intent.

It retaliates. The first system escalates again. The entire process might unfold faster than political leaders could understand what was happening.

Rickli’s concern that autonomous weapons may therefore favour offensive action and encourage pre-emption is intellectually substantial. His earlier work explicitly argued that autonomous weapons could destabilise strategic relationships because states might fear that waiting placed them at a disadvantage.

Yet even here caution is necessary. Technologies rarely favour attack or defence permanently. The machine gun initially appeared to favour defence. Tanks restored manoeuvre. Anti-tank weapons strengthened defence again. Precision weapons altered the balance once more. Drones currently illustrate precisely this dialectic.

Cheap attack drones threaten expensive military equipment. But cheap interceptor drones may eventually threaten attacking drones. Artificial intelligence may enable swarms — while simultaneously enabling defensive systems capable of identifying and destroying swarms.

Technology generates counter-technology. Hence predictions of permanent offensive dominance should be treated sceptically.

There is another element of Rickli’s work that deserves particular attention — cognitive warfare. Modern conflict increasingly concerns not merely territory but perception. Social media, generative artificial intelligence, synthetic video, behavioural profiling and algorithmically targeted propaganda make it possible to influence populations on a scale unimaginable during the twentieth century.

Rickli and his collaborators have described the emerging problem as one of cognitive security — the defence of societies against manipulation of the informational environment. Democracies may be peculiarly vulnerable because their information systems are deliberately open.

This is persuasive up to a point. Russia has spent years attempting to exploit precisely these vulnerabilities — amplifying political divisions, manufacturing narratives and flooding information environments with contradictory accounts until citizens cease believing that objective truth exists. Artificial intelligence makes this dramatically cheaper.

Once generating a convincing article, photograph, recording or video costs almost nothing, propaganda acquires industrial scale. But there is a danger in the concept of cognitive warfare. If defined too broadly, it becomes almost indistinguishable from politics. Governments have always attempted to persuade foreign populations. Newspapers have always published propaganda. Intelligence agencies have always conducted influence operations. Political parties have always manipulated emotion.

A democratic government that begins treating disagreement as enemy cognitive warfare risks becoming the very authoritarian system it claims to resist. “Cognitive security” must therefore mean protecting the integrity of the informational environment — not protecting governments from inconvenient opinions.

The distinction is fundamental. Rickli’s wider intellectual project nevertheless identifies something genuinely historic. For centuries the technological development of warfare generally favoured concentration. Industrial warfare required factories. Factories required capital. Capital required states.

Nuclear weapons carried this logic to its extreme. Only enormously powerful states could manufacture thermonuclear arsenals, ballistic missiles, submarines and global early-warning networks.

Digital technology may be reversing this relationship. Software replicates almost without cost. Commercial drones are ubiquitous. Satellite imagery can be purchased. Artificial intelligence models can analyse enormous datasets. Synthetic biology may eventually place capabilities once confined to state laboratories within reach of comparatively small organisations. Neurotechnology introduces still more unsettling possibilities; Rickli’s recent work explores the dual-use military implications of technologies designed to interact with or influence the human brain.

Power is becoming distributed. That does not necessarily mean states are becoming irrelevant. Indeed the opposite may occur. Ukraine demonstrates why. A decentralised ecosystem can innovate extraordinarily quickly. But ultimately somebody must manufacture ammunition by the million, maintain electrical grids, mobilise soldiers, operate air-defence networks, negotiate alliances and finance years of warfare.

Only states can reliably perform these functions at enormous scale. The future may therefore not be the disappearance of the state but something more complicated — states surrounded by vast technological ecosystems of private companies, volunteers, software developers, intelligence networks and semi-autonomous machines.

Rickli has described a related phenomenon as “surrogate warfare”: states and other actors increasingly employing technological and human intermediaries to project coercive power while reducing financial, political and human costs. Ukraine again offers an extraordinary preview. A soldier may operate a commercially derived drone manufactured by a private company, modified by volunteers, financed through crowdfunding, using software written by civilians, guided by satellite communications owned by another private corporation and directed towards a target identified through a combination of military intelligence, commercial satellite imagery and battlefield reconnaissance.

Where, precisely, does the state end? The answer is increasingly difficult to identify. Yet there is one final reason to resist technological fatalism. War is not ultimately about machines. It is about political purpose. An autonomous drone does not hate anyone. An artificial intelligence system does not covet territory. A swarm possesses no ideology. A neural network does not dream of empire. Human beings supply the objectives.

This matters because much contemporary discussion about artificial intelligence subtly transfers responsibility from people to technology. We speak of the danger that AI “might” start wars, “might” select targets or “might” escalate conflicts. But machines operate inside political systems created by human beings. If Russia builds autonomous weapons and sends them against Ukrainian cities, the fundamental problem is not that artificial intelligence has become malevolent. The problem is that the Kremlin decided to attack Ukraine.

International discussion about autonomous weapons is nevertheless accelerating. In August 2026 the United Nations Secretary-General and the President of the International Committee of the Red Cross renewed their demand for legally binding rules governing autonomous weapons, arguing that decisions over human life must remain subject to human control. The regulatory debate with which Rickli has long been associated has therefore moved from academic foresight towards an immediate question of international politics.

Regulation is desirable. But regulation will encounter the oldest problem in arms control.

Responsible states may regulate themselves. Irresponsible states may not.

A democratic country may insist upon elaborate human oversight while an authoritarian adversary deploys fully autonomous systems without comparable restrictions. The resulting ethical asymmetry might perversely reward the actor least interested in ethics.

Hence governance cannot substitute for deterrence. Rules governing military artificial intelligence will work only where states believe that obeying them does not make them defenceless. That is perhaps where Rickli’s technological foresight should be combined with the harder traditions of strategic theory.

The coming military revolution will not be governed solely by algorithms. It will be governed by algorithms, institutions, industrial capacity, alliances, political legitimacy, deterrence, geography and human courage. Artificial intelligence will change warfare profoundly. Autonomous machines will become ubiquitous. Drone swarms will grow larger. Decision cycles will accelerate. Propaganda will become synthetic. The boundary between civilian and military technology will continue to dissolve.

Rickli is right to insist that these developments deserve attention before rather than after they become uncontrollable. But Ukraine supplies the necessary corrective. A technologically superior army can still lose. A technologically inferior society can still resist.

Machines can multiply power — but they cannot manufacture political purpose.

The deepest lesson of the technological transformation of warfare may therefore be paradoxical. The more sophisticated our machines become, the more important the distinctly human elements of strategy become: judgement, organisation, adaptability, legitimacy and will.

The battlefield of the twenty-first century may indeed be populated by algorithms. But the wars fought upon it will remain ours.