\u201cLa guerra en Ucrania est\u00e1 provocando una revoluci\u00f3n en la guerra con aviones no tripulados que utilizan inteligencia artificial\u201d, dec\u00eda a todo volumen un titular del Washington Post en julio pasado. Luego, en oto\u00f1o, una avalancha de informes dec\u00eda que tanto Rusia como Ucrania hab\u00edan desplegado peque\u00f1os drones que utilizaban inteligencia artificial para identificar y localizar objetivos. Tener IA a bordo significaba que los drones, versiones del Russian Lancet y el ucraniano Saker Scout , no necesitar\u00edan un operador humano para guiarlos hasta el impacto. Si esta IA\u00a0 hubiera demostrado su eficacia en la batalla, realmente habr\u00eda sido una revoluci\u00f3n.<\/p>\n
WASHINGTON \u2014 \u201cThe war in Ukraine is spurring a revolution in drone warfare using AI,\u201d blared a\u00a0Washington Post headline<\/a>\u00a0last July. Then, in the fall, a flurry of reports said that both\u00a0Russia<\/a>\u00a0and\u00a0Ukraine<\/a>\u00a0had deployed small drones that used artificial intelligence to identify and home in on targets. Having on-board AI meant that the drones, versions of the Russian\u00a0Lancet\u00a0<\/a>and the Ukrainian\u00a0Saker Scout<\/a>, wouldn\u2019t need a human operator to guide them all the way to impact.<\/p>\n If this AI had proved itself in battle, it really would have been a revolution. Electronic warfare systems designed to disrupt the operator\u2019s control link \u2014 or worse, trace the transmission to its source for a precision strike \u2014\u00a0would have been largely useless against self-guided drones. Skilled and scarce drone jockeys could have been replaced by thousands of conscripts quickly trained to point-and-click on potential targets. And instead of every drone requiring an operator staring at its video feed full-time, a single human could have overseen a swarm of lethal machines.<\/p>\n All told, military AI would have taken a technically impressive and slightly terrifying step towards independence from human control,\u00a0like Marvel\u2019s Ultron singing Pinnochio\u2019s<\/a>\u00a0\u201cI\u2019ve got no strings on me.\u201d\u00a0Instead, after more than four months of frontline field-testing, neither sides\u2019 AI-augmented drones seem to have made a measurable impact.<\/p>\n In early February, a detailed\u00a0report<\/a>\u00a0from the Center for a New American Security dismissed the AI drones in a few lines. \u201cThe Lancet-3 was advertised as having autonomous target identification and engagement, although these claims are unverified,\u201d wrote CNAS\u2019s defense program director,\u00a0Stacie Pettyjohn<\/a>. \u201cBoth parties claim to be using artificial intelligence to improve the drone\u2019s ability to hit its target, but likely its use is limited.\u201d<\/p>\n Then, on February 14, an independent analysis suggested that the Russians, at least, had turned their Lancet\u2019s AI-guidance feature off. Videos of Lancet operators\u2019 screens, posted online since the fall, often included a box around the target, one capable of moving as the target moved, and a notification saying \u201ctarget locked,\u201d freelance journalist\u00a0David Hambling<\/a>\u00a0posted on\u00a0Forbes<\/a>. Those features would require some form of algorithmic object-recognition, although it\u2019s impossible to tell from video alone whether it was merely highlighting the target for the human operator or actively guiding the drone to hit it.\u00a0<\/strong><\/p>\n However, \u201cnone of the Lancet videos from the last two weeks or so seem to have the \u2018Target Locked\u2019 or the accompanying bounding box,\u201d Hambling continued. \u201cThe obvious conclusion is that automated target recognition software was rolled out prematurely and there has been product recall.\u201d<\/p>\n Don\u2019t Believe The (AI) Hype<\/strong><\/p>\n It\u2019s impossible to confirm Hambling\u2019s analysis without access to Russian military documents or the drone\u2019s software code. But Pettyjohn and two other drone experts \u2014 both fluent Russian-speakers who are normally enthusiastic about the technology \u2014 agreed that Hambling\u2019s interpretation was not only plausible but probable.<\/p>\n \u201cThis is a fairly detailed analysis, looks about right to me,\u201d said\u00a0Alexander Kott<\/a>, former chief scientist at the Army Research Laboratory, in an email calling Breaking Defense\u2019s attention to the Forbes piece. \u201cIt is difficult to know for sure\u2026I have not seen an independent confirmation, and I don\u2019t think one can even exist.\u201d<\/p>\n \u201cI think it\u2019s accurate,\u201d said\u00a0Sam Bendett<\/a>\u00a0of CNA, a think tank with\u00a0close ties<\/a>\u00a0to the Pentagon, in an email exchange with Breaking Defense. (Bendett also spoke to Hambling for his story).<\/p>\n \u201cThis technology needs a lot of testing and evaluation, this technology needs a lot of iteration, [and] many times the technology isn\u2019t ready,\u201d he had told Breaking Defense\u00a0before the Forbes story was published. \u201cI think it\u2019s a slow roll because both sides want to get it right. Once they get it right, they\u2019re going to scale it up.<\/p>\n \u201cThis is in fact technologically possible,\u201d Bendett said. \u201cWhoever gains a breakthrough in drone technology and quickly scales it up gains a huge advantage.\u201d<\/p>\n But that breakthrough clearly hasn\u2019t happened here, Pettyjohn told Breaking Defense. \u201cRussian industry often makes pretty outlandish claims about its weapons\u2019 capabilities, and in practice we find that their performance is much less than promised\u00a0\u2026 This has been most prominent with autonomous systems, as Sam Bendett and Jeff Edmonds found in their\u00a0CNA report<\/a>\u00a0on uncrewed systems in Ukraine.\u201d<\/p>\n The Ukrainians don\u2019t seem to have done better, despite similar media hype.<\/p>\n \u201cThere are lots of really exciting reports out there about the Saker Scout and the autonomous target recognition software that the Ukrainians have been developing,\u201d Pettyjohn said. \u201cIf Saker Scout does what it\u2019s supposed to \u2026. it could go off, find a target, and decide to kill it all on its own without a human intervening.\u201d<\/p>\n \u201cWhether it can actually do this\u2026 it\u2019s hard to sift through,\u201d she continued. \u201cI am definitely on the skeptical side.\u201d<\/p>\n The Real AI Revolution \u2013 Date TBD<\/strong><\/p>\n So what would it really take for Russia and Ukraine \u2014 or for that matter, the US or China \u2014 to replace a human operator with AI? After all, the brain is a biological neural network, honed over millions of years of evolution to take in a dazzling array of sensory data (visual, audio, smell, vibration), update an internal 3D model of the external world, then formulate and execute complex plans of action in near-real time.<\/p>\n For AI to match that capability, it needs what theorists of combat call \u201csituational awareness,\u201d Kott told Breaking Defense. \u201c[Like] any soldier\u2026 they need to see what\u2019s happening around them.\u201d That requires not just object recognition \u2014 which AI finds hard enough \u2014 but the ability to observe an object in motion and deduce what action it is in the middle of performing, Kott argues.<\/p>\n That\u2019s a task that humans do from infancy. Think of a baby saying \u201cmmmm\u201d when put in their high chair, even before any food is visible: That\u2019s actually a complex process of observing, turning those sensory inputs into intelligible data about the world, matching that new data with old patterns in memory, and making inferences about the future. One of the most famous maxims in AI,\u00a0Moravec\u2019s Paradox<\/a>, is that tasks humans take for granted can be confoundingly difficult for a machine.<\/p>\n Even humans struggle to understand what\u2019s going on when under stress, in danger, and facing deliberate deception. Ukrainian decoys \u2014 fake HIMARS rocket launchers, anti-aircraft radars, and so on \u2014 routinely trick Russian drone operators and artillery officers into wasting ordnance on fakes while leaving the well-camouflaged decoys alone, and\u00a0machine-vision algorithms have proven even easier to deceive<\/a>. Combatants\u00a0must also keep watch for danger, from obviously visible ones the human brain\u2019s evolved to recognize \u2014 someone charging at you, screaming \u2014 to high-tech threats unaided human senses can\u2019t\u00a0perceive, like electronic warfare or targeting lasers locking on. A properly equipped machine can detect radio waves and laser beams, but its AI still needs to make sense of that incoming data, assess which threats are most dangerous, and decide how to defend itself, in seconds.<\/p>\n But the difficulty doesn\u2019t stop there: Combatants must fight together as a team, the way human have since the first Stone Age tribe ambushed another. Compared to rifle marksmanship and other individual skills, collective \u201cbattle drills,\u201d team-building, and protocols for clear communication under fire consume a tremendous amount of time in training. So great-power projects for military AI \u2014 both America\u2019s\u00a0Joint All-Domain Command & Control<\/a>\u00a0and\u00a0China\u2019s \u201cinformatized warfare\u201d<\/a>\u00a0\u2014 focus not just on firepower but on coordination, using algorithms to share battle data directly from one robotic system to another without need for a human intermediary.<\/p>\n So the next step towards effective warfighting AIs, Pettyjohn said, \u201cis really networking it together and thinking about how they\u2019re sharing that information [and]\u00a0who\u2019s actually authorized to shoot. Is it the drone?<\/a>\u201d<\/p>\n Such complex digital decision-making requires sophisticated software, which needs to run on high-speed chips, which in turn need power, cooling, protection from vibration and electronic interference, and more. None of that is easy for engineers to cram into the kind of small drones being used widely by both sides in Ukraine. Even the upgraded\u00a0Lancet-3<\/a>\u00a0fits less than seven pounds (3 kg) of explosive warhead, leaving little room for a big computer brain.<\/p>\n The requisite engineering \u2014 and the cost \u2014 may prove too much for Russia or, especially, Ukraine, many of whose drones are hand-built from mail-order parts. \u201cGiven the very low cost of current FPV [First-Person View<\/a>] drones, and the fact that many of them are assembled by volunteers literally on their kitchen table\u2026 thee cost-benefit tradeoffs likely remain uncertain,\u201d Kott told Breaking Defense.<\/p>\n \u201cThe reason you\u2019re seeing so many drones [is] that they\u2019re cheap,\u201d Pettyjohn agreed. \u201cOn both sides\u2026they\u2019re not investing in increased defenses against jamming\u2026 because it would make them too expensive to afford in the numbers that they\u2019re needed. They\u2019d rather just buy lots of them and count on some of them making it through.\u201d<\/p>\n So even if Russia or Ukraine can implement on-board AI, she said, \u201cit\u2019s not clear to me it will scale in\u00a0this\u00a0<\/em>conflict, because it depends a lot on the cost.\u201d<\/p>\n However, that doesn\u2019t mean AI won\u2019t scale up in other conflicts with other combatants, especially high-tech nations with big defense budgets like the US and China. But even for those superpowers miniaturizing AI to fit on drones is daunting: There\u2019s good reason headline-grabbing AIs like ChatGPT run on massive server farms.<\/p>\n