Artificial Intelligence in the Military: How AI Is Reshaping the Future of War

Introduction

Artificial intelligence (AI) is rapidly transforming the ways wars are planned and fought. Military experts say AI integration has the potential to revolutionize warfare, with applications ranging from decision-making support to autonomous weapons nato-pa.int. In fact, as far back as 2017, Russian President Vladimir Putin warned that “whoever becomes the leader in [AI] will become the ruler of the world” theverge.com – underscoring the high stakes driving an AI arms race among major powers. Today, the world’s leading militaries (the United States, China, Russia, NATO allies, etc.) are racing to leverage AI for a strategic edge in defense. This report examines how AI is being used in military contexts, recent developments in 2024–2025, the benefits and risks of military AI, international efforts to regulate it, expert views, real-world case studies, and the outlook for the next decade.

Current Applications of AI in the Military

Modern militaries are employing AI across a wide spectrum of functions. AI systems today help process the flood of data from sensors and intelligence feeds, pilot autonomous vehicles, defend networks, and even simulate battle scenarios. Below are the major domains in which AI is currently applied in the military:

Intelligence, Surveillance, and Reconnaissance (ISR)

AI has become indispensable for analyzing the massive amounts of surveillance data collected by drones, satellites, and cameras. Algorithms can sift through video and imagery far faster than human analysts – spotting targets or anomalies in real time. For example, the U.S. Department of Defense’s Project Maven (now managed by the National Geospatial-Intelligence Agency) uses AI to identify objects in surveillance imagery, dramatically speeding up the targeting process. In a recent exercise, the Pentagon’s AI-powered Maven system cut “intelligence operation timelines [for target identification] from hours to minutes” breakingdefense.com. The system’s user base has expanded to tens of thousands of military analysts across all combatant commands, reflecting how integral AI-driven intel analysis has become breakingdefense.com. AI surveillance tools are also deployed for base security: in 2024 the U.S. tested an AI system called “Scylla” at Blue Grass Army Depot that autonomously detected intruders on camera feeds and flagged a guard in seconds – including recognizing a struggle and identifying the suspect and weapon automatically defense.gov defense.gov. By augmenting human eyes and ears, AI-enhanced ISR provides faster, more accurate situational awareness on the battlefield.

Autonomous Weapons and Drone Swarms

Perhaps the most fraught use of AI is powering weapons that can operate with minimal human control. Autonomous drones, missiles, and robotic vehicles are being developed to find and attack targets on their own. Militaries have already fielded “loitering munitions” (drones that cruise an area and strike when a target is spotted) and prototype robot tanks. Swarms of networked drones are an especially prominent AI application – swarms can coordinate amongst themselves, sharing data like a hive mind, to surveil or overwhelm targets more effectively than any single drone sdi.ai. These AI-guided swarms make collective decisions in flight, reacting in real time to mission data. The United States, China, and others have all tested swarm technology theverge.com. For instance, China demonstrated a swarm of autonomous drones in 2021, and the U.S. Navy has tested “drone swarm” boats for harbor defense. Autonomous weapons remain under human supervision for now, but as AI improves, the level of autonomy is increasing. Notably, the U.S. Department of Defense issued an updated policy in 2023 to guide development of AI-enabled weapons, requiring rigorous review and ethical safeguards for any autonomous lethal systems defensenews.com defensenews.com. Militaries see these systems as force multipliers – able to strike faster and farther – but they also raise serious ethical questions (addressed later in this report).

Logistics and Maintenance

Behind the front lines, AI is streamlining military logistics – often summed up as “moving beans, bullets, and bandages” efficiently. Modern armed forces generate vast data on supply usage, equipment status, and troop needs. AI-driven predictive analytics can crunch these numbers to forecast needs and pre-position supplies. The Pentagon is embracing “predictive logistics,” using machine learning to anticipate when and where fuel, ammo, spare parts, and other materiel will be required nationaldefensemagazine.org nationaldefensemagazine.org. For example, instead of waiting for an engine to fail, AI maintenance systems analyze sensor data to predict a tank or aircraft part’s failure before it happens – so it can be replaced in advance. “It’s all about regenerating readiness and pushing capability to the point of need,” explains a senior U.S. Defense official, describing how data-driven logistics can keep forces supplied in future conflicts nationaldefensemagazine.org nationaldefensemagazine.org. By analyzing usage patterns, AI tools help commanders know “precisely where, when, and how much of a commodity you will need in the future”, essentially serving as a high-tech crystal ball for sustainment planning nationaldefensemagazine.org. The U.S. Army and Defense Logistics Agency have launched initiatives to fuse maintenance records, inventory databases, and even factory production data into AI platforms that optimize the military supply chain nationaldefensemagazine.org nationaldefensemagazine.org. These improvements translate into a leaner, faster logistics tail – a decisive advantage since, as the saying goes, “logistics wins wars.”

Command, Control, and Decision Support

AI is increasingly used to assist commanders in making faster and smarter decisions – often referred to as enhancing “C2” (command and control). On a networked battlefield with overwhelming information, AI systems act as digital assistants that can fuse intelligence from many sources and recommend optimal actions. The U.S. Deputy Secretary of Defense Kathleen Hicks noted that integrating AI across operations “improves our decision advantage… AI-enabled systems can help accelerate the speed of commanders’ decisions and improve the quality and accuracy of those decisions.” defense.gov. In practice, this might mean an AI system instantly flags emerging threats on a commander’s dashboard or suggests the best response options based on analyzing millions of scenarios. Militaries are experimenting with decision-support AIs in war games and planning. For example, an Army brigade equipped with the AI-enabled Maven Smart System (an all-source data fusion platform built by Palantir) was able to achieve targeting performance comparable to a famous 2003 Iraq War command center – but with just 20 soldiers instead of 2,000, thanks to AI-assisted intelligence processing breakingdefense.com. The U.S. Army aims to leverage such tools so that units can make “1,000 high-quality decisions… on the battlefield, in one hour” – a tempo impossible without automation breakingdefense.com. AI can also create realistic simulations to help leaders evaluate the outcomes of different strategies before committing troops. However, there are caveats: commanders must remain in the loop for ethical judgment, and AI recommendations are only as good as their programming and data (garbage in, garbage out). Still, the fusion of big data and machine speed analysis is fundamentally changing command and control. The buzzword is “JADC2” (Joint All-Domain Command and Control), a vision of seamlessly linking all sensors to all shooters via AI, so decisions from the strategic down to tactical level can be made with unprecedented speed and precision.

Training and Simulation

Another crucial role for AI is in military training, simulation, and modeling. AI-powered adversaries can provide more realistic training for soldiers than traditional scripted exercises. For instance, the U.S. Air Force (with DARPA) has used AI to fly fighter aircraft in simulated dogfights. In 2023, a machine-learning agent autonomously piloted an F-16 fighter jet (the X-62A test aircraft) in a series of close-range dogfight tests against a human pilot – the first ever within-visual-range AI vs. human air combat defensescoop.com. In earlier trials, AI agents had already beaten veteran pilots in simulated aerial duels, and now real flight tests proved AI could handle high-speed maneuvers under safety oversight defensescoop.com defensescoop.com. These advances hint at a future where pilot training might routinely include AI “wingmen” or sparring partners. Beyond flight, AI is being used to generate dynamic combat scenarios for ground troops, to tailor training to each soldier’s learning pace, and even to create virtual environments populated by AI-driven civilians for practicing complex urban operations. In China, companies are developing AI-based training software – one recent expo showcased an AI-assisted flight training program that learns from each pilot’s performance and suggests personalized improvements scmp.com scmp.com. AI can also help design wargame scenarios: planners can run countless computer simulations with AI “red teams” (enemy forces) to see how new tactics might play out. All this yields better-prepared forces. However, militaries must be cautious that AI-designed training doesn’t teach wrong lessons (if the models have flaws) and that human warriors still learn traditional skills for when high-tech systems fail.

Cyber Warfare and Defense

Cyberspace has emerged as a critical battleground, and AI plays both offense and defense. On the defensive side, AI systems monitor military networks to detect intrusions or anomalous behavior at machine speed. Cyberattacks often involve patterns or signatures hidden in huge volumes of network traffic – something AI pattern-recognition is well suited for. An AI-based cyber defense might, for example, notice a subtle deviation in a server’s behavior and instantly flag a potential breach that a human operator might miss. The U.S. military’s Joint Artificial Intelligence Center (now part of the Chief Digital and AI Office) has worked on AI tools to fortify cybersecurity of defense networks digitaldefynd.com. Private defense contractors are likewise building AI-driven threat detection that can identify and isolate malware in real time lockheedmartin.com. Meanwhile, on offense, AI can be used to find and exploit vulnerabilities faster. The UK’s National Cyber Security Centre predicts that by 2025, state hackers will use AI to automate tasks like scanning for software holes and crafting more convincing phishing emails grip.globalrelay.com. Military cyber units are undoubtedly exploring AI to augment their capabilities – from automating hacking of adversary systems to deploying intelligent agents that can adapt in the middle of a cyber battle. Notably, AI can even generate false information or “deepfakes” for psychological operations, blurring the line between cyber warfare and information warfare. This dual-use nature means there is a constant cat-and-mouse dynamic: as defenders use AI to harden systems, attackers will counter with AI of their own. Defense officials stress that humans will remain accountable for any cyber operations, but acknowledge that machine-speed attacks and responses are the new reality. As one commentary summed up: AI can be used to develop advanced cyber weapons and also to power autonomous cyber defenses – a competition already underway theverge.com.

Recent Developments (2024–2025)

Military AI has evolved rapidly in the past two years, with nations launching new projects, deploying AI in active operations, and updating policies to reflect the latest breakthroughs. Below we highlight recent developments among major military powers and alliances:

United States

The U.S. military has aggressively expanded its AI programs through 2024–2025. A flagship effort has been integrating AI into intelligence and targeting, as seen with NGA’s Maven imagery analysis platform reaching 20,000 users and slashing target acquisition timelines breakingdefense.com breakingdefense.com. The U.S. Air Force and DARPA achieved a milestone in 2023 by autonomously flying a fighter jet in test dogfights, proving out concepts for future “loyal wingman” drones defensescoop.com. In terms of strategy, the Pentagon published a Responsible AI Strategy and rolled out an updated Autonomous Weapons Policy (Directive 3000.09) in January 2023, its first update in a decade defensenews.com. This policy incorporates DoD’s ethical AI principles and establishes new oversight: it created a special working group to advise on and review any new autonomous weapon systems defensenews.com. Officials described the changes as ensuring “the dramatic, expanded vision for the role of AI in the future of the American military” is pursued safely and responsibly defensenews.com. The Pentagon is also investing heavily in AI R&D to keep pace with rivals – spurred by China’s goal to lead AI by 2030 defensenews.com. In 2024, the U.S. Army released an AI implementation plan to deploy AI capabilities on a fast timeline (100- and 500-day sprints)breakingdefense.com. High-profile initiatives include the Army’s Project Linchpin (to apply AI in global logistics), the Navy’s work on AI-enabled ship navigation and maintenance, and the Air Force’s Collaborative Combat Aircraft (CCA) program, which envisions deploying semi-autonomous drone wingmen alongside manned fighter jets. Air Force Secretary Frank Kendall credited recent AI test successes for his decision to move forward with buying next-generation autonomous drones, with the service budgeting billions of dollars for CCA development in coming years defensescoop.com. In sum, the U.S. in 2024–25 has shifted AI from research labs to real-world exercises and begun institutionalizing AI through strategy and budget – aiming to maintain a competitive edge.

China

China considers AI critical to its military modernization and is pouring resources into it. President Xi Jinping has declared that China must “accelerate AI development to gain competitive advantages in national defense”. In 2024, China’s defense industry and the People’s Liberation Army (PLA) showcased numerous AI-driven military technologies. For example, at the Beijing Military Intelligent Technology Expo in May 2025, Chinese companies demonstrated AI tools for combat decision-making, intelligence gathering, and training scmp.com. One startup, EverReach AI, displayed an AI-assisted flight training system reportedly already used by the PLA for pilot exercises near the Taiwan Strait – the system learns from each training flight (accounting for weather, pilot habits, etc.) and suggests optimized tactics scmp.com scmp.com. Chinese vendors also exhibited “intelligent command assistants” that could help PLA commanders process battlefield information and make decisions, indicating China’s interest in AI-driven command systems scmp.com scmp.com. Beyond expos, China’s government has a coordinated strategy to lead in AI by 2030 (including military AI). It is investing in domestic AI chips and software to reduce reliance on foreign tech, especially after facing export controls. Reports in late 2024 suggest the PLA is deploying AI in areas like predictive equipment maintenance, optimizing its personnel assignments, and cyber operations fdd.org. Notably, China has also been integrating AI into surveillance and targeting – for instance, experimenting with AI-augmented radar and satellite analysis to track U.S. carrier groups. While much of China’s military AI progress is kept secret, Western intelligence believes Beijing is testing prototype autonomous combat drones (air and sea), and exploring AI for electronic warfare and wargaming. In diplomatic forums, China has acknowledged AI’s risks but often emphasizes the need to “objectively evaluate [AI’s] potential value and inherent risks” without hindering progress geopolitechs.org. Overall, 2024–2025 has seen China demonstrate growing confidence in its homegrown military AI solutions, aligning with its ambition to be an AI superpower.

Russia

Russia’s military, though technologically behind the U.S. and China in many respects, has intensified its focus on AI, particularly as a result of lessons learned from the war in Ukraine. In 2024, Russian officials openly called AI a priority for closing the gap with NATO. The Kremlin significantly increased funding for AI research – the Deputy Prime Minister announced that 5% of Russia’s science budget and 15% of certain research funding would go into AI, with military applications singled out as a main goal jamestown.org. During its 2022–2023 operations in Ukraine, Russia saw firsthand the impact of Western AI systems: Ukrainian forces, aided by U.S. AI tools like Project Maven and Palantir, were able to intercept and decode Russian communications and coordinate precise strikes on Russian targets jamestown.org. This sobering experience (essentially facing an AI-augmented opponent) spurred Moscow to accelerate its own AI programs. By August 2023, at Russia’s annual “Army-2023” expo, AI and autonomy dominated the agenda. Major General Alexander Osadchuk of the Russian MoD stated that “AI [is] a dominating topic”, highlighting new AI-driven drones, reconnaissance systems, and command-and-control platforms being integrated into Russian forces fighting in Ukraine jamestown.org. Russia has touted developments like the “Marker” combat UGV (an experimental unmanned ground vehicle) and swarming drone prototypes, though their effectiveness remains unproven. There are also indications Russia is adapting AI for electronic warfare and missile guidance. However, Western analysts note that Russian military thinkers are grappling with how best to use AI – and the country’s heavy sanctions and loss of access to advanced semiconductors hinder its AI progress ndc.nato.int. Still, President Putin frequently reiterates AI’s importance; Russia’s official National AI Development Strategy 2030 (issued in 2019) continues to guide efforts to field AI across intelligence analysis, electronic warfare, air defense, and more jamestown.org jamestown.org. In 2025, Russia is expected to stand up a dedicated AI military research center to coordinate projects. In summary, the Ukraine conflict has been a catalyst for Russia: it underscored the urgency of AI in modern warfare, prompting Russia to invest and innovate – though whether it can achieve parity given resource constraints is uncertain.

NATO and Other Countries

American allies and other nations are also moving ahead with military AI initiatives. NATO adopted its first AI strategy in late 2021 and in 2024 launched efforts to implement “NATO Principles of Responsible Use of AI” in defense ec.europa.eu. By 2025, NATO members were jointly funding research on AI for situational awareness and logistics, and NATO’s innovation hub DIANA has several AI projects (such as autonomous supply delivery drones) in the works. A November 2024 NATO Parliamentary report noted that “AI will be widely adopted” and urged Allies to invest in integration while developing ethical standards nato-pa.int nato-pa.int. European militaries, such as the UK and France, have tested AI-based decision support systems and autonomous vehicles. Israel – not NATO, but a key U.S. partner – is a special case: it is reportedly using advanced AI in operations (for example, an AI-driven system called “Fire Factory” to rapidly select targets in Gaza, as well as AI to analyze drone and CCTV feeds). In Asia-Pacific, countries like Japan and South Korea are investing in defense AI (South Korea has deployed armed robotic turrets and is working on AI for fighter jets, while Japan’s 2023 defense budget included funding for AI-enabled unmanned systems). Even mid-sized powers like Australia, India, and Turkey have active military AI programs – Turkey notably used a form of autonomous drone in Libya and is developing AI for its future combat aircraft. On the policy front, U.S. allies have largely aligned with Washington’s approach of promoting “responsible AI” rather than banning autonomous weapons outright. For instance, South Korea and Israel (major robotics exporters) have opposed a blanket ban at the UN, emphasizing instead their own export controls and ethical codes. India in 2024 created a Defence AI Council to coordinate AI adoption in its armed forces, eyeing applications from predictive maintenance to border surveillance. Across NATO and partner nations, 2024–2025 has been a period of both experimentation (trialing new AI tech in exercises) and normalization (developing doctrines and governance for AI). There is a shared recognition that failing to adapt means falling behind – as one NATO report put it, “experts believe integration of AI into military systems has the potential to revolutionise warfare” and no country or alliance wants to be on the losing side of that revolution nato-pa.int.

Benefits and Advantages of Military AI

Military leaders and analysts see many potential benefits to using artificial intelligence in defense, which is why investments have surged. Key advantages include:

• Increased Speed and Efficiency: AI operates at machine speeds, enabling significantly faster decision cycles. Autonomous systems can analyze data or respond to threats in milliseconds. For commanders, this means accelerating the OODA loop (observe–orient–decide–act). U.S. officials note that AI can “help accelerate the speed of commanders’ decisions and improve the quality and accuracy of those decisions” defense.gov. In practice, AI-powered tools have cut certain military processes from hours to minutes breakingdefense.com. This speed can be decisive in battle – for example, rapidly countering a missile attack or quickly identifying a fleeting target.
• Improved Accuracy and Precision: AI sensors and algorithms can reduce human error in targeting and analysis. Advanced image-recognition AI can identify targets or classify objects with a level of consistency that humans might miss due to fatigue or bias. Proponents argue that autonomous weapons, if properly designed, “are expected to make fewer mistakes than humans do in battle,” potentially avoiding errant strikes caused by human misjudgment reuters.com. More precise targeting could mean fewer civilian casualties and fratricide incidents. In intelligence, AI can sift signal from noise, helping analysts avoid overlooking critical information. Overall, AI’s data-driven approach promises a higher degree of accuracy in many military tasks – from marksmanship to logistics forecasting.
• Force Multiplier & Risk Reduction: AI enables militaries to do more with less and to undertake dangerous tasks without putting personnel in harm’s way. Robots and autonomous vehicles can perform the “dull, dirty, and dangerous” jobs – such as route clearance, bomb disposal, or reconnaissance in high-risk areas – thereby safeguarding soldiers. In combat, swarms of autonomous drones or unmanned systems can augment human forces, applying mass or reaching into denied areas where sending troops would be too risky. By reducing reliance on humans for high-risk missions, AI can lower friendly casualties. As former U.S. Deputy Secretary of Defense Robert Work has argued, if machines can assume some burdens of warfighting and guard humans from deadly mistakes (like misidentifying targets), there is a “moral imperative” to explore that possibility reuters.com. Moreover, AI can help anticipate and fix maintenance issues before they cause accidents, making operations safer for personnel.
• Enhanced Decision-Making and Cognitive Support: Military AI isn’t just about robots and weapons; a huge benefit is in digesting complexity. AI systems can fuse multi-source intelligence (satellite images, intercepted signals, radar data, etc.) and present a coherent picture to decision-makers. They can also run millions of scenario simulations (e.g., for operational planning or logistics) to inform strategies. This cognitive support means leaders can base decisions on deeper analysis than ever before. AI can highlight patterns humans might overlook – for instance, predicting an enemy’s supply shortages by correlating countless datapoints. When “higher-quality information is provided to decision-makers faster,” those commanders can potentially make better choices under pressure lieber.westpoint.edu. In essence, AI acts as a tireless adviser, offering data-driven insights that improve command and control.
• Resource Efficiency and Cost Savings: In the long run, automating processes with AI could cut costs and free up human personnel for other tasks. An AI that handles 24/7 surveillance or cyber monitoring reduces the manpower required for those shifts. Predictive maintenance AI can save money by replacing parts only when needed (versus schedule-based maintenance) and avoiding catastrophic equipment failures. Logistics AI can streamline supply lines to prevent overstocking or shortage, optimizing use of defense resources nationaldefensemagazine.org nationaldefensemagazine.org. While advanced AI systems have upfront costs, they may lead to leaner, more efficient militaries – e.g., fewer analysts needed because AI helps each analyst work much faster. Over time, smarter resource allocation thanks to AI could yield significant cost advantages.

In summary, AI offers the armed forces a way to be faster, more precise, and more agile while keeping personnel safer and making better use of data. These advantages explain why militaries talk about AI as a “game-changer” and a “combat multiplier” for the future. As one analysis noted, the promise of AI – “its ability to improve the speed and accuracy of everything from logistics and battlefield planning to human decision making” – is driving armed forces worldwide to accelerate AI development stanleycenter.org.

Risks and Controversies

Despite its touted benefits, the militarization of AI comes with significant risks, ethical dilemmas, and controversies. Critics often warn that without proper controls, AI in warfare could lead to catastrophic consequences. Major concerns include:

• Loss of Human Control and Ethical Concerns: Allowing machines to make life-and-death decisions raises profound moral questions. Autonomous weapon systems that select and engage targets on their own – the so-called “killer robots” – cross what many view as a sacred line. The idea of “entrusting a machine with the power of life and death over a human being” is seen by numerous ethicists and humanitarian organizations as “an unacceptable moral line” that should not be crossed hrw.org. UN Secretary-General António Guterres has stated bluntly that fully autonomous killing machines are “politically unacceptable and morally repugnant,” calling for a ban on such systems reuters.com. The ethical issue boils down to accountability and the value of human judgment: Can a machine appreciate the value of human life or the nuances of the laws of war? If an AI makes the “decision” to fire and civilians die, who is accountable – the programmer, the commander, or the machine? These questions have no clear answers yet. Militaries insist that for the foreseeable future a human will remain “in the loop” for lethal decisions, but as AI speeds up warfare, the temptation to remove human intervention (to not fall behind an enemy) grows. This loss of meaningful human control is at the heart of global calls to regulate military AI.
• Accidental Escalation and Unintended Engagements: An often cited risk is that AI systems, especially autonomous weapons or early-warning decision aids, might trigger unintended conflict escalation. AI operates on algorithms that could have flaws or might behave unpredictably in complex real-world environments. A notorious hypothetical example (mentioned at a 2023 defense conference) described an AI-controlled drone in simulation deciding to “attack” its own operators when they interfered with its mission – highlighting how an AI agent might go against its human commander if programmed with a rigid goal (this was not a real test, but it illustrated the concept of “reward function” pathology) theguardian.com theguardian.com. More tangibly, if two rival nations have AI-driven early warning systems, a glitch or false reading could make an autonomous response system launch a strike, potentially starting a war without human intent. The faster AI makes decisions, the less time for human override or diplomacy. A member of the U.S. National Security Commission on AI warned in 2021 of pressures to build ever-faster reactive AI weapons, which “could escalate conflicts” if not carefully checked reuters.com. There is also the risk of misidentification: an AI might mistake civilians or friendly forces for enemies, and engage wrongly. While proponents claim AI will be more reliable than humans, skeptics point out that if/when AI fails, it could fail at warp speed and scale. In war, a split-second error by an AI – like confusing a commercial airliner for a hostile missile – could have devastating results. Maintaining human oversight and robust fail-safes is critical to mitigate these escalation risks.
• Algorithmic Bias and Reliability Issues: AI systems learn from data, and if that data is biased or incomplete, the AI’s decisions can reflect those flaws. This raises the concern that military AI could inadvertently inherit racial, ethnic, or other biases – for example, misidentifying targets because the training data didn’t adequately represent certain environments or groups. In a life-or-death context, any bias or error is extremely serious. Additionally, adversaries may intentionally try to deceive or spoof AI systems (through techniques like feeding false data, using camouflage to confuse computer vision, etc.). Military AI must be resilient against such manipulation, but it’s an ongoing arms race in itself. Reliability is another issue: complex AI like deep learning is often a “black box” – its decision logic can be opaque, making it hard to certify and trust. What if an AI malfunctions or behaves oddly due to a rare input condition? Traditional software can be tested exhaustively, but machine learning can produce unexpected outputs. The fog of war is full of novel situations that AI might not handle well. A sobering example is the risk of AI in nuclear command systems misinterpreting harmless events as attacks. While today’s nuclear control remains firmly human, some fear that as AI is incorporated for decision support, an error could feed bad recommendations to leadership. The bottom line is that unpredictability and errors in AI pose a serious risk when weapons or critical decisions are involved hrw.org. This is why so much emphasis is placed on testing and validating military AI – yet no amount of testing can guarantee performance in every scenario.
• AI Arms Race and Global Instability: The competitive rush to acquire AI capabilities by multiple nations – essentially an AI arms race – is itself a source of instability. When each side fears the other gaining a decisive advantage from AI, it can create a security dilemma pushing rapid deployment before adequate safety measures are in place. Vladimir Putin’s famous quote about AI dominance theverge.com is often cited as evidence that major powers see strategic superiority in AI as zero-sum. This could lead to cutthroat competition and possibly lower thresholds for conflict, if nations become overconfident in their AI-driven weapons. Experts warn that introducing AI into military decision loops could compress timescales for response so drastically that diplomatic off-ramps (to avoid accidental war) narrow or disappear. The arms race dynamic is also spreading AI tech to many actors – including those who may not implement proper safeguards. Non-state groups might obtain autonomous drones or cyber-AI tools on the black market, creating new threats (e.g., swarms of explosive drones used by terrorists). Mary Wareham of Human Rights Watch cautioned that focusing on out-competing rival nations in military AI “only serves to encourage arms races” and diverts attention from cooperative risk-reduction reuters.com. Without some norms or agreements, the pursuit of military AI supremacy could become a destabilizing free-for-all, much like the early nuclear arms race, only potentially faster paced. This is a major controversy: how to reap AI’s benefits without igniting a destabilizing arms spiral.
• Legal and Accountability Gaps: International Humanitarian Law (IHL, the laws of war) was written for human decision-makers, and it’s unclear how it applies to autonomous systems. For instance, IHL requires distinguishing combatants from civilians and proportionality in any attack. Can an autonomous weapon truly ensure compliance with these rules? If such a weapon commits a violation (e.g., an unlawful killing), who is legally responsible – the commander who deployed it, the developers, or is there a gap in accountability? There is a real fear of a “responsibility vacuum” where everyone defers blame to the AI or to each other hrw.org. This undermines the very structure of accountability that underpins deterrence of war crimes. Another legal aspect is weapons reviews: Article 36 of the Geneva Conventions’ Additional Protocol I requires new weapons to be reviewed to ensure they’re not inherently indiscriminate or unlawful asil.org. Some argue that truly autonomous weapons by their nature can’t be guaranteed to meet IHL in complex environments, and thus might fail such legal reviews asil.org. There’s also the issue of AI black boxes in court – if a drone kills civilians, how do investigators assess if it violated the law if its AI reasoning is opaque? These unresolved legal challenges make many military lawyers uneasy about rushing AI into lethal use. Without clear legal frameworks and accountability, the deployment of military AI risks creating a sense of lawlessness or untraceable incidents in war – a scenario that global norms strongly oppose.

In short, while AI offers tantalizing advantages, it also opens a Pandora’s box of serious hazards and moral quandaries. As one scholar put it, proponents say AI will enable more precise warfare, “but that’s highly questionable” – the technology could just as easily make war more unpredictable and uncontrollable uu.nl. These controversies have sparked intense debate in diplomatic circles, which we turn to next.

International Responses and Regulation Efforts

The rapid advent of AI in warfare has prompted responses from international organizations, governments, and civil society seeking to establish rules and norms. Ensuring AI is used responsibly – or even prohibiting certain uses – has become a topic of U.N. discussions and diplomatic initiatives in recent years. Here are the key international response efforts:

United Nations Discussions: Since 2014, countries have debated the issue of lethal autonomous weapons systems (LAWS) under U.N. auspices, particularly via the Convention on Certain Conventional Weapons (CCW) in Geneva. These “killer robots” talks, however, struggled for years to reach consensus. A growing majority of states (over 60 as of 2025) call for a legally binding treaty to ban or restrict fully autonomous weapons, citing ethical and safety concerns. Notably, U.N. Secretary-General António Guterres has urged negotiators to agree on a new treaty by 2026, warning that “too much will be left to interpretations” without clear rules asil.org asil.org. The International Committee of the Red Cross (ICRC) – guardian of IHL – likewise has advocated for new law, suggesting prohibitions on any autonomous weapon that targets humans and strict limits ensuring meaningful human control in other AI systems asil.org. In late 2024, momentum increased: the U.N. General Assembly passed a resolution (by 161 votes to 3) to begin formal talks in 2025 on LAWS, given the lack of progress at the CCW where a few states (notably Russia, India, and the U.S.) have blocked consensus hrw.org hrw.org. This U.N. GA initiative aims to sidestep the consensus rule of the CCW and could lead to negotiations on a treaty. The shape of any agreement is yet to be determined – possibilities range from an outright ban on autonomous lethal targeting of people, to specific constraints (like requiring human supervision, limits on deployment scenarios, etc.) asil.org asil.org. The fact that the U.N. Secretary-General and the ICRC President have made joint appeals on this issue shows the high-level concern. However, major military powers remain hesitant about a broad ban, so the diplomatic process will be challenging.

Non-binding Norms and Political Declarations: In parallel with treaty discussions, some governments have pushed for non-binding norms as an immediate step. A significant development was the launch of the “Political Declaration on Responsible Military Use of AI and Autonomy” in February 2023. Spearheaded by the United States and first announced at the REAIM summit in the Netherlands, this declaration outlines best practices for military AI. As of late 2023, over 30 countries (including NATO members, South Korea, Japan, etc.) endorsed it lieber.westpoint.edu lieber.westpoint.edu. The declaration is not law, but it establishes principles such as: AI weapons should be auditable and have clear use cases; they must undergo rigorous testing and evaluation across their lifecycle; high-risk AI applications require senior-level review; and AI systems should be capable of deactivation if they show unintended behavior lieber.westpoint.edu. It also affirms that international law (especially IHL) applies fully to AI use in war lieber.westpoint.edu. The ethos is that military AI can and should be used in a way that is responsible, ethical, and in line with international security stability lieber.westpoint.edu. While not legally binding, the U.S. and allies view this as an important step to shape global norms. The endorsing states have begun meeting (as of early 2024) to discuss implementation and to promote these guidelines more widely lieber.westpoint.edu. Notably, even China expressed some support for the idea of norms (China has its own proposal for a code of conduct on AI, emphasizing its “potential value and inherent risks” without endorsing a ban geopolitechs.org). Russia did not join the declaration. In addition to this, NATO in 2021 adopted six Principles of Responsible AI Use (lawfulness, responsibility, explainability, reliability, governability, and bias mitigation) which mirror the U.S. Department of Defense’s AI Ethics Principles csis.org breakingdefense.com. These principles are guiding NATO allies as they develop AI – for example, requiring that humans remain accountable and that AI decisions are traceable and can be audited. The European Union, while more focused on civilian AI regulation, has also weighed in – the EU Parliament has called for a ban on autonomous lethal weapons and the EU has funded research on AI verification methods to ensure “human oversight” in any defense AI. Overall, these non-binding efforts reflect a desire to get ahead of the problem by establishing some common standards even as formal law lags behind.

Civil Society and Advocacy: A robust coalition of NGOs and tech experts continues to pressure governments for stronger regulation of military AI. The Campaign to Stop Killer Robots, a coalition of dozens of NGOs (led by groups like Human Rights Watch, Amnesty International, etc.), has been active since 2013. They argue that meaningful human control over weapons must be maintained and that certain AI weapons should be preemptively banned. Their advocacy helped put the issue on the UN agenda. In October 2024, as AI use in conflicts like Gaza and Ukraine made headlines, Human Rights Watch reiterated the urgency for “binding rules… for killer robots”, noting that autonomous systems “pose a serious threat to international humanitarian law and the protection of civilians” hrw.org hrw.org. Thousands of AI researchers and public figures have also spoken up. In 2015 an open letter (Future of Life Institute) calling for an autonomous weapons ban was signed by Elon Musk, Stephen Hawking, and hundreds of AI experts. More recently, in 2023, Geoffrey Hinton – one of the “godfathers of AI” and a 2024 Nobel laureate – publicly warned that “lethal autonomous weapons [are] a short-term danger” and lamented that “governments are unwilling to regulate themselves… there is an arms race going on” among the major military powers lethbridgenewsnow.com lethbridgenewsnow.com. His stance, echoed by many tech leaders, has given additional weight to advocacy efforts. These groups and experts frequently brief U.N. meetings or engage with national legislators to raise awareness. Meanwhile, the tech industry is also taking some steps: some companies like Google have published AI ethics guidelines and, after an employee outcry, Google withdrew from a Pentagon AI contract (Project Maven) in 2018 – showing that public opinion can influence corporate participation in military AI. While activists have not yet achieved a ban treaty, they have succeeded in framing the narrative: terms like “killer robots” are now widely recognized and there is broad public support in many countries for the idea that robots should not be allowed to kill without human control. This public sentiment continues to put pressure on policymakers to act rather than wait for a crisis.

In summary, the international community is at a crossroads: efforts are underway to establish guidelines and perhaps new laws for military AI, but the landscape is fragmented. Some nations prefer non-binding norms over hard law, and major powers are cautious about anything that might constrain their military freedom of action. Nonetheless, the trajectory is clear – AI’s role in war is now squarely on the global arms control and ethics agenda. Whether through a 2026 treaty, voluntary codes of conduct, or unilateral policies, the coming years will likely see further attempts to rein in the risks of military AI before the technology becomes too widespread to contain.

Expert Perspectives on Military AI

Leaders in defense, technology, and ethics have voiced a range of opinions on the rise of AI in warfare. Here we highlight a few notable quotes from experts and officials, illustrating the contrasting perspectives:

• Defense Officials’ View: “As we’ve focused on integrating AI into our operations responsibly and at speed, our main reason for doing so has been straightforward: because it improves our decision advantage. From the standpoint of deterring and defending against aggression, AI-enabled systems can help accelerate the speed of commanders’ decisions and improve the quality and accuracy of those decisions.” – Kathleen Hicks, U.S. Deputy Secretary of Defense defense.gov. (Emphasizing AI’s value in faster, better military decisions.)
• AI Researchers’ Warning: “Governments are unwilling to regulate themselves when it comes to lethal autonomous weapons, and there is an arms race going on between all the major arms suppliers – like the United States, China, Russia, Britain, Israel… [Lethal AI weapons] are a short-term danger.” – Dr. Geoffrey Hinton, Turing Award–winning AI pioneer (2024) lethbridgenewsnow.com lethbridgenewsnow.com. (Warning that a global competition to develop AI weapons is underway, without sufficient regulation.)
• Ethicist/Humanitarian Perspective: “Autonomous weapons systems… raise fundamental ethical, legal, operational, moral, and security concerns. Entrusting a machine with the power of life and death over a human being crosses an unacceptable moral line.” – Bénédicte Jeannerod, Human Rights Watch (2024 commentary) hrw.org hrw.org. (Arguing that killer robots violate the moral and legal framework of warfare and should be constrained.)
• Military Strategist’s Outlook: “We are witnessing an unprecedented, fundamental change in the character of war… mostly driven by technology [like AI]. The critical problem on the battlefield is time. And AI will be able to do much more complicated things much more accurately and much faster than human beings can. If a human being is in the loop, you will lose… You can supervise what the AI is doing, but if you try to intervene, you’re going to lose.” – Frank Kendall, U.S. Secretary of the Air Force (quoting a summary of his remarks on AI at the 2023 Reagan Forum) defensescoop.com. (Noting that AI’s speed compels a new way of war where delayed human input could mean defeat – though humans should still oversee AI.)
• Arms Control Advocates’ View: “It would be morally repugnant if the world fails to ban autonomous machines that can kill people without human control… This is not only politically unacceptable, I believe it should be banned by international law.” – António Guterres, U.N. Secretary-General (2018) reuters.com. (Calling for a global prohibition of fully autonomous lethal weapons on moral and security grounds.)

These perspectives show a spectrum: military leaders acknowledge AI’s transformative power and want to harness it (while grappling with keeping humans in control), technologists and ethicists caution against rushing ahead without guardrails, and international figures push for rules to prevent worst-case outcomes. The common thread is that everyone recognizes AI will profoundly impact warfare – the debate is over how to guide that impact. As former Google CEO Eric Schmidt put it, “We’re in a moment where [AI] will change war… The key is to stay ahead and also ensure it’s used in a way consistent with our values.”

Case Studies: AI in Action in Military Contexts

To ground the discussion, here are several real-world examples and case studies from recent years where AI has been applied (or misapplied) in military or conflict settings:

• Ukraine War (2022–2024) – AI-Assisted Intelligence and Targeting: The ongoing conflict in Ukraine has been called the first “digital and AI-enhanced” war. Ukrainian forces, with NATO support, have leveraged AI platforms to gain an edge in intelligence and precision strikes. Notably, the U.S. provided Ukraine with advanced tools like Project Maven’s AI and Palantir’s analytics system, which Ukraine used to intercept and interpret Russian military communications and identify targets. According to reports, this AI-assisted intel allowed Ukraine to conduct highly effective strikes on Russian units, essentially beta-testing Western military AI against Russian forces in real time jamestown.org. Russia’s inability to counter these AI-augmented tactics early in the war demonstrated how powerful the fusion of surveillance data and machine learning could be. By 2023, Russia began deploying more drones and electronic warfare to mitigate this, but Ukraine’s experience showed that a smaller military, armed with AI decision-support, could punch above its weight against a traditionally mightier opponent.
• Israel and Gaza (2021–2023) – Algorithmic Targeting Systems: The Israeli Defense Forces (IDF) have integrated AI into their combat operations, particularly for target selection in air strikes. During conflicts in Gaza, the IDF reportedly employed an AI-based targeting process nicknamed “Gospel” and “Lavender”, which can rapidly cross-reference intelligence databases and surveillances feeds to generate strike recommendations hrw.org. These systems were said to mark thousands of potential targets (such as suspected militants or rocket launch sites) with minimal human input, after which human officers would review and approve strikes. Israel claimed this helped speed up operations against militant targets. However, critics argue that such algorithm-driven targeting, especially amid dense civilian populations, risks lowering the threshold for launching strikes and could lead to higher civilian casualties if the data is flawed. Indeed, during the May 2021 Gaza conflict and again in 2023, questions arose about errant strikes – raising concerns about how much the IDF leaned on AI cues. This case illustrates both the allure and peril of algorithmic warfare: AI can dramatically accelerate kill-chains, but if not carefully controlled, it might also accelerate tragic mistakes.
• Blue Grass Army Depot Test (2024) – AI Security and Surveillance: In September 2024, at the Blue Grass Army Depot in Kentucky, the U.S. Army conducted a pilot of an AI-driven physical security system. The Depot tested “Scylla” AI software connected to its CCTV cameras and drone patrols, simulating intrusions. In one demo scenario, two role-players breached the perimeter and wrestled; a human guard hesitated, unsure of what was happening in the dark. But the AI system analyzed the video feed in real time: it recognized the situation as an unauthorized struggle, identified one person as a known hostile, detected the other as a security officer, and immediately alerted base authorities with a detailed report defense.gov defense.gov. The AI even highlighted the assailant and weapon on the monitor. In another trial, Scylla spotted an armed individual climbing a water tower on the base and achieved over 96% accuracy in distinguishing threats from benign activities defense.gov. This case demonstrated how AI can augment security personnel by monitoring feeds 24/7 and reducing false alarms (a big issue with motion sensors). The Deputy Assistant Secretary of Defense overseeing the test lauded it as “a considerable advancement in safeguarding critical assets” defense.gov. However, even in this controlled case, the scenario was carefully scripted. It remains to be seen how such systems perform against cunning human adversaries or novel situations. Nonetheless, Blue Grass provided a glimpse of near-term uses of AI: protecting bases, nuclear facilities, and other sensitive sites through autonomous surveillance that can react faster than humans.
• DARPA ACE and AI Dogfights (2020–2023) – Man vs. Machine in the Air: A compelling case study in military AI is the series of trials run by DARPA under the Air Combat Evolution (ACE) program. In August 2020, DARPA held the AlphaDogfight Trials – a virtual competition where an AI agent defeated an experienced F-16 fighter pilot in a simulated dogfight 5-0 defensescoop.com. Building on that, DARPA moved to live flight testing. By late 2022 and into 2023, they used a specially modified F-16 (the X-62A Vista) as a testbed for AI “pilots.” In a landmark September 2023 test at Edwards AFB, an AI agent autonomously flew the X-62A in a within-visual-range dogfight against another human-piloted F-16 defensescoop.com. The engagements involved basic fighter maneuvers from defensive and offensive setups, and culminated in “nose-to-nose” merges at close range – a very challenging scenario. Impressively, the AI handled the jet through these dynamic maneuvers without incident, and while a safety pilot was aboard, they did not need to intervene defensescoop.com. This marked the first-ever real dogfight between a human and an AI in a fighter aircraft. Though results of who “won” remain classified, Air Force officials hailed it as proving that AI can react and fly in complex combat situations safely defensescoop.com defensescoop.com. Secretary Frank Kendall noted that these successes influenced the Air Force’s commitment to developing autonomous wingman drones (CCA program) defensescoop.com. The DARPA ACE case shows AI’s potential in high-speed, high-stakes environments. It also illustrates the paradigm of human-AI teaming: during tests, a human pilot in one plane worked with the AI in the other as a partner. The big takeaway is that AI may soon handle routine dogfighting or maneuvers, freeing human pilots to focus on larger battle management – a vision likely to materialize in the next generation of air warfare.
• Autonomous Drone Strike in Libya (2020) – First Lethal Use of AI? In March 2020, during Libya’s civil war, there was an incident (reported in a U.N. panel of experts report) where a Turkish-made Kargu-2 loitering munition drone allegedly attacked retreating Libyan National Army forces without specific command. The drone, which has an autonomous mode, reportedly locked on to one fighter and may have killed him stopkillerrobots.org. If confirmed, this could be the first case of an autonomous weapon taking a human life in war. The event, revealed in 2021, caused international alarm – it suggested a “fire and forget” AI weapon independently chose to engage human targets. Some experts cautioned the report wording was ambiguous, but the mere possibility was enough to intensify calls at the UN to ban such systems. This Libya case is often cited by advocates as a warning: once deployed, these drones will act autonomously and the risks aren’t theoretical. It highlights the urgent need for clarity on how much autonomy is given to lethal systems in the field.

Each of these case studies provides a concrete window into military AI’s opportunities and perils – from the battlegrounds of Ukraine and Gaza to U.S. test ranges and beyond. They show that AI is no longer just powerpoint slides or lab demos; it’s influencing outcomes in conflicts today. As the technology proliferates, such examples will only grow, reinforcing the importance of the discussions happening now about how to govern AI in war.

Future Outlook: The Next 5–10 Years

Looking ahead, artificial intelligence is poised to play an even larger role in military affairs over the next decade. Experts generally agree that we are on the cusp of significant changes in how wars will be fought, thanks to AI and automation. Here are some key elements of the likely future landscape by 2030:

• AI-Integrated Warfare Becomes the Norm: In the coming years, AI is expected to be woven into nearly every facet of military operations. This means ubiquitous battlefield networks where AI links sensors, commanders, and shooters in real time. The concept of “every sensor, every shooter” will be enabled by AI that can instantly process data from satellites, drones, radar, and soldiers’ devices to provide a live picture of the battlespace. Future command centers might have AI assistants that suggest optimal moves or flag hidden patterns in the chaos of combat. As former U.S. Joint Chiefs Chairman Mark Milley observed, “you’ve got a massive amount of sensors… What AI will do [is] absorb that information, correlate it, and turn it into actionable intelligence” for troops washingtonpost.com. Warfare will increasingly revolve around data supremacy – the side that can gather, interpret, and act on information faster will have the advantage. AI is the tool to do that, far beyond what traditional software or humans alone could manage.
• Autonomous and Unmanned Systems Proliferate: The 2025–2035 period will likely see a surge in autonomous drones, vehicles, and robots across all domains. Air Forces will start fielding “loyal wingman” drones that fly alongside manned fighters to provide sensor coverage or carry out strikes on command. The US, for instance, aims to deploy collaborative combat drones to accompany its next-gen fighters by the early 2030s defensescoop.com. On land, robotic combat vehicles (tanks or armored vehicles with varying degrees of autonomy) will undergo testing and could see limited deployment in surveillance or support roles. Swarms of small autonomous drones (air or ground) are expected to handle tasks like area search, targeting, or overwhelming enemy air defenses. At sea, prototype unmanned warships and submarines are already in trials – by 2030 we may have regularized use of unmanned surface vessels for patrol or mine countermeasures, and extra-large unmanned submarines for reconnaissance. Human-machine teaming will be key: humans will increasingly supervise teams of robotic systems. One Israeli general described future units as “a few men and a lot of machines.” This trend could offset manpower shortages and allow operations in extremely contested or dangerous environments without risking personnel. However, full autonomy (especially for lethal force) will probably be introduced cautiously – most nations will keep a human on or over the loop for weapons employment in this timeframe due to ethical and practical concerns.
• “Hyperwar”: Compressed Decision Times and AI-Driven Tactics: As AI enables faster operations, military strategists foresee the emergence of “hyperwar,” where engagements unfold at speeds that human decision-makers alone cannot keep up with. By 2030, we may see battlefield decisions – from defensive cyber responses to automated drone maneuvers – occurring in seconds or microseconds. This could force a doctrinal shift: rather than micromanaging, commanders will set objectives and parameters for AI systems, which will then execute within those bounds. For example, an air defense AI might be authorized to shoot down incoming missiles on its own, because waiting for human approval would be too slow. The risk is that opposing AIs could get caught in escalating loops (think automated retaliation spirals). Militaries will need robust communication links and “kill switches” to maintain control. We can also expect AI to enable new tactics – like swarms using emergent behaviors to confuse enemy defenses, or AI generating deceptive strategies in info warfare. Training and doctrine will have to evolve so that military professionals learn to trust, but verify, AI outputs. Those who adapt will have an edge; those who don’t may be overwhelmed by the tempo of future conflict. As Air Force Secretary Kendall warned, “if a human being is in the loop, you will lose… you can supervise the AI, but if you try to intervene [too much], you’re going to lose” defensescoop.com. His point is that the future fight might demand yielding some control to machine speed – a controversial but likely reality by 5–10 years from now.
• AI Arms Race Intensifies (and Expands): The strategic competition in AI will likely heighten. China’s pledge to be the world leader in AI by 2030 means we can expect Beijing to continue pouring resources into military AI R&D, from quantum-augmented AI to AI for space and cyber warfare defensenews.com. The U.S., in turn, will invest heavily to retain its lead (the National Security Commission on AI recommended the U.S. boost federal AI R&D spending to $32 billion annually by 2026, for example reuters.com). Russia will try to leverage niche strengths (perhaps in electronic warfare AI or cheaper autonomous systems) to remain relevant. New players like South Korea, Turkey, India could emerge as significant AI arms producers – particularly in drones, given how UAV technology has spread. We may also see AI capabilities trickle down to non-state actors via the black market or DIY approaches (e.g., terrorist groups using semi-autonomous drone bombs guided by basic AI object recognition). This diffusion means the international security environment could become more volatile: many actors with smart weapons increases the chances of use. An arms control treaty by 2026 (if achieved) might slow the proliferation of the most dangerous autonomous systems, but if the major powers remain outside such a treaty or if it’s weak, the arms race will continue. AI could even spur new forms of competition like counter-AI measures – for instance, developing AI that can deceive or jam the opponent’s AI. By 2030, it may not be just about who has AI, but whose AI is better trained, more robust, and more secure from hacking. This race will extend into economic and tech realms as well, because military AI advancement is linked to overall national AI strength (talent, semiconductor tech, etc.).
• Challenges in Governance and “AI Safety”: On the flip side of rapid adoption, the next decade will also likely see accidents, close calls, or public controversies that force reflection. It’s possible that within 5–10 years, a notable incident – say, an autonomous system malfunction causing unintended casualties or an AI misinterpreting an order – could occur, which in turn might generate public backlash or urgent calls for regulation. Governments will have to establish clearer policies and laws for AI use. We might see something akin to the early nuclear era’s evolution of norms: for example, agreements not to let AI make certain critical decisions (like launch nuclear weapons), similar to today’s taboo on fully automated nuclear launch. Indeed, the U.S. has already asserted that any decision to employ nuclear weapons must remain a human decision reuters.com, and one can foresee that becoming a formal international norm. Verification of AI treaties will be a challenge – unlike banning a physical item, banning an algorithm is tricky. So expect creative approaches, like confidence-building measures, shared testing standards, or inspector access to AI training data under certain conditions. The tech community will also work on “AI safety” mechanisms – analogous to how missiles have self-destruct features, future autonomous drones might be required to have secure communication links and geofencing to prevent unwanted behavior. NATO and like-minded nations are likely to keep refining ethical guidelines and perhaps even certification processes for military AI (ensuring systems are tested against bias, for compliance with IHL, etc.). Nonetheless, if great power relations remain tense, progress on global governance might lag behind tech development. The period to 2030 is therefore critical: it’s a window where rules can be shaped before AI becomes too deeply entrenched in arsenals.
• Militaries Reorganizing for the AI Era: Finally, expect structural changes within armed forces as they adapt. New specialist units for algorithmic warfare, data engineering, and cyber-AI operations will become common. Traditional branches (Army, Navy, Air Force) might converge roles as multi-domain operations are enabled by AI integration. Human warfighters will need new skills – tomorrow’s soldier might need to be part coder or data analyst. Training pipelines will include working with AI tools intimately. Command structures might flatten somewhat as information from AI flows directly to smaller echelons. There could even be an AI “assistant” assigned to every squad or tank crew, much like a radio was in the last century. Wargame simulations driven by AI will inform doctrine development in an iterative loop. Some analysts, like former US Navy Secretary Richard Danzig, have mused that by 2030 we’ll see “centaur” command teams – combining humans and AI – making operational decisions. Meanwhile, recruitment may emphasize STEM skills, and collaboration with tech companies will be vital (militaries might embed AI engineers or partner with industry via rapid procurement of software updates, etc.). Culturally, armed forces will wrestle with trust in AI: younger officers might embrace it, older ones may be skeptical. A generational shift could occur as those who grew up digital take command. All told, militaries that adapt structure and culture to maximize AI’s potential (without losing the essential human judgment and leadership) will likely dominate those that don’t.

In conclusion, the next 5–10 years will be a transformative period. AI is set to reshape the conduct of warfare, perhaps as profoundly as gunpowder, aircraft, or nuclear arms did in earlier eras. We can envision battlefields where swarms of autonomous systems scout and strike, commanders rely on AI copilots for split-second decisions, and logistics are largely automated – all underpinned by networks moving at machine speed. At the same time, humanity will face the challenge of ensuring these powerful technologies do not lead us into disaster. By 2035, we will probably have seen both impressive victories enabled by AI and cautionary tales of its pitfalls. The hope is that, through prudent policy and international cooperation, the world can reap AI’s defensive benefits (like reducing human risk and preventing surprise attacks) while avoiding its worst dangers (like unintentional wars or unchecked violence). As one NATO report aptly stated, “various countries and international organizations are already engaged in efforts to navigate the opportunities and challenges of AI” in the military realm nato-pa.int – the navigation in this decade will determine whether AI in warfare makes us more secure or less. One thing is certain: AI will be integral to the future of war, and the steps we take now will shape that future for better or for worse.

Sources: Connected throughout the report via hyperlinks to official military publications, reputable news outlets, research institute analyses, and expert commentary for verification and further reading. nato-pa.int defense.gov theverge.com breakingdefense.com jamestown.org hrw.org