LAWS

When Isaac Asimov set out the three fundamental laws of robotics in 1942, he not only created a brilliant piece of science fiction but also launched a monumental philosophical experiment that spanned his entire body of work. The first and most important law – a robot may not injure a human being – reflected an infinitely humanist, enlightened and rational worldview.

Asimov believed that logic and engineering precision were capable of curbing violence. However, as our present day demonstrates, this belief was a beautiful but infinitely naïve notion.

I recently referred to the LAWS (Lethal Autonomous Weapons Systems), in the post where I wrote that the average survival time of invading Russian soldiers on the front line in Ukraine is a mere 20–30 minutes.

If we want to understand why military technology and artificial intelligence are exactly where they are today, we must abandon Asimovian idealism. Instead, we must turn to ’s dark but far more realistic cyberpunk visions, because Gibson had a much deeper understanding of basic human nature.

Official definitions (for example, those of the US Department of Defence (DoD)) define LAWS as a weapon system capable of selecting and attacking targets without human intervention once activated. These are not simply remote-controlled drones, but systems in which algorithms and artificial intelligence make the final decision to open fire.

Experts generally classify autonomy into three levels:

• Human-in-the-loop (Human-in-the-loop): The machine makes the recommendation, but a human pulls the trigger.
• Human-on-the-loop (Human above the loop): The machine operates autonomously, but a human operator can override it and shut it down at any time.
• Human-out-of-the-loop (Outside human control): Complete autonomy. The machine makes independent decisions regarding life and death, without external control

Isaac Asimov’s ’s entire body of work – from his early robot short stories to his later Foundation novels – is built around a single, vast, coherent idea: that the world, society and intelligence can be improved through mathematical and logical laws. For him, the Three Laws were not an after-the-fact software update, but an inescapable mathematical axiom burned into the fabric of the positronic brain.

When his robots (such as Daneel Olivaw) were confronted with the world’s imperfections, they elevated this logic to an even higher level and formulated the Zerodic Law: a robot must not harm humanity as a whole.

Asimov’s world assumed that humankind, as the creator, would programme its own better self—enlightened humanism—into its machines. He believed that technology, by its very nature, would be rational and protective.

However, Asimov — unfortunately — omitted from the equation the darkest, most primal driving forces of human nature: the lust for power, greed (which was discussed at length in the last post), tribal logic and the survival instinct.

Where Asimov saw pristine laboratories and logical robots, there William Gibson saw flesh, blood, grime bathed in neon light, and unbridled capitalism. Gibson’s entire body of work – from the Sprawl trilogy (Neuromancer,** Count Zero, **Mona Lisa Overdrive) right up to the Bridge trilogy – systematically rejects and rewrites Asimov’s idealism.

Gibson’s genius lay in his understanding of fundamental human nature. He knew that if we were to place god-like technology (such as the AI), humans will not forge a moral shield from it, but a weapon. In the Gibson universe, technology does not make humans better, but amplifies all their flaws.

Artificial intelligences (such as Wintermute or Neuromancer) are not guardians working for the welfare of humanity, but elusive, gigantic entities created by megacorporations and military interest groups to secure their own dominance. In Gibson’s world, there are no hard-and-fast moral laws. There are only interests, and the software is just as ruthless as the people who funded it.

Today, we are living precisely within this Gibsonian realism. When modern armies develop LAWS, they are driven by the arms race and the need for geopolitical survival. The logic of warfare is not morality, but efficiency and speed. If an algorithm is capable of identifying and destroying a target in milliseconds without human intervention (Human-out-of-the-loop), then armies will not bother with Asimov’s laws.

Far from turning Asimov’s laws into reality, they have in fact brought about the exact opposite:

Today’s loitering munitions (suicide drones) are already capable of selecting their own victims based on facial or thermal imaging after the connection is lost. Gibson predicted that in a world of cutting-edge technology the institution of responsibility would cease to exist. If an algorithm makes a mistake and kills civilians, there is no single person to blame. Responsibility is diffused amongst the programmers, manufacturers and commanders – just as in the chaotic, faceless corporate world of cyberpunk.

Although the ‘killer robot’ sounds futuristic, the military has long been using rudimentary autonomous systems (such as naval and air defence missile defence shields, like the Phalanx or the Patriot). In modern conflicts – such as the wars in Ukraine or Gaza – however, we are moving ever closer to fully AI-controlled attacks:

• Kargu-2 and Switchblade drones: These are suicide (loitering) munitions capable of autonomously identifying targets using built-in sensors.
• Bullfrog system: An autonomous weapon system tested by the US Marine Corps that independently tracks and targets aerial threats

Many people believe that Boston Dynamics, the company famous for developing somersaulting humanoids and robot dogs, Boston Dynamics is behind the development of LAWS. In reality, the company (along with several other leading robotics firms) signed an ethical declaration in 2022, in which they strictly prohibited the sale and use of their own robots for armed purposes. The yellow Spot robot dog can be used at most for reconnaissance, bomb disposal tasks or industrial patrols.

However, the defence industry will not tolerate such self-imposed restrictions, and the vacuum was immediately filled by others:

• Ghost Robotics (Vision 60): This company does not mince its words when it comes to ethics. The defence contractor Skyborne Technologies has directly fitted a shotgun, a grenade launcher and remote-controlled weapon systems onto their four-legged robot platform (Vision 60). The armed robotic dog, dubbed CODIAQ, is already taking part in live-fire military trials approved by the Pentagon.
• China’s Unitree and the armed forces: China (for example, through Unitree Robotics) is churning out cheap, robust robotic dogs and humanoids at an astonishing rate. Robot dogs equipped with machine guns are regularly seen at Chinese military exercises, autonomously storming test buildings, providing a direct and large-scale preview of LAWS.

As if the four-legged robots weren’t menacing enough, two-legged humanoid robots have also begun marching towards the battlefields. Elon Musk has officially announced that the Tesla Optimus will be a civilian factory worker and domestic helper.

However, whilst seeking financial and AI influence from investors, Musk has repeatedly let slip that “robot army” in reference to the Optimus project. Although Tesla denies any military intent, the technology (bipedal movement, precise handling of objects, facial and environmental recognition) is directly transferable to warfare.

And this dystopia is no longer confined to test laboratories. Whilst Tesla is still preparing for the mass production of Optimus, defence firms contracted by the US Army have already deployed the world’s first humanoid built specifically for military purposes, the Phantom MK-1, on reconnaissance missions in Ukraine. The geopolitical tension of the situation is clearly illustrated by the fact that the US government, citing security and data protection risks, has completely banned the import of foreign (primarily Chinese) humanoid and quadrupedal robots, in order to protect the domestic AI and robotics supply chain.

The circle is complete. Asimov’s factories dreamt of peaceful industrial robots. Gibson’s reality, however, is that by the time Tesla’s Optimus or the Chinese Agibot robots step out of the factory gates, the armed forces will already be ready, both in terms of software and hardware, to put weapons in their hands, eliminating the human factor from the front lines.

Let’s not delude ourselves; these technologies will not stop at the front lines. History has taught us that whatever proves its worth on the battlefield will sooner or later find its way into civilian life, albeit in a different guise. Think of everyday drones: originally military surveillance devices, today anyone can buy them at the local shop.

The same spill-over is now taking place with the artificial intelligence and robotics behind LAWS:

• Drone programmes in major cities: Police forces around the world (such as the London Metropolitan Police or US law enforcement agencies) are rapidly rolling out their police drone programmes. In London, drone coverage is already being established on a borough-by-borough basis, and these aerial units are being equipped with artificial intelligence and real-time, live facial recognition software (Live Facial Recognition).
• The legalisation of biometric surveillance: Across Europe – for example in Sweden and Germany – legislators have given the green light for federal police forces to use AI-assisted, real-time biometric tracking in public spaces. Meanwhile, the FBI is already testing AI-based surveillance drones capable of identifying faces, number plates or concealed weapons in crowds.
• Algorithmic suspicion: Tech giants such as the notorious Palantir Technologies, provide police forces with data fusion and analytics software (such as Gotham) that uses mass data mining and AI to create personal profiles of citizens. According to critics, this amounts in practice to concepts of ‘automated suspicion’ and pre-emptive policing.

When an autonomous military drone learns to track a target without human approval, its software does not need to be rewritten to identify protesters, government critics or simply those on the ‘watchlist’. The dystopia described by Gibson does not begin when robots start shooting at us, but when they transform our everyday lives into an invisible, algorithmically controlled digital prison.

Humanity stands at a crossroads – one could continue this post in this rather clichéd manner, but that is not the case. Instead of humanity, the superpowers and powerful corporations have voted in favour of LAWS. No one dares to stand in the way, for fear that their rivals will develop an even more ruthless algorithm, and for fear of losing profit.

Asimov’s Three Laws are now nothing more than a naïve memory of a bygone, optimistic era. We live in Gibson’s dark, pragmatic present, and this world is closer to the Terminator Skynet-style dystopia than we care to admit.

For the most terrifying thing is that artificial intelligence doesn’t even need self-awareness to bring about a Skynet-style global catastrophe or a machine uprising. The machine doesn’t need to ‘hate’ humans. It is enough for it to receive a completely cold, emotionless mathematical command, and for the networked, autonomous weapons to deem the extermination of humanity logical due to a software glitch or a misinterpreted command.

Weapons are already flying and marching according to their own code. The question is no longer whether we can teach machines to respect human life, but whether we are capable of pulling the plug before the system takes on a life of its own.

What do you think? How realistic is the danger that, at the end of the arms race, we will actually create our own Skynet?

I also used Gemini AI to help write this post.

• US Department of Defence (DoD): Directive 3000.09 – Autonomy in Weapon Systems (The Pentagon’s official military guidelines on the development of autonomous weapon systems and levels of human oversight) — U.S. Department of Defense Directive.
• Human Rights Watch (HRW) & Stop Killer Robots: International humanitarian law and the legal regulation of autonomous weapons (Summary of the UN’s attempts at arms control) — Human Rights Watch - Killer Robots.
• Militarnyi Defence Journal: Humanoid Robots on the Battlefield (Report on the testing and reconnaissance missions of Foundation Robotics’ Phantom MK-1 bipedal humanoid units in Ukraine) — Militarnyi Defence News.
• Ghost Robotics Corp: Vision 60 QUGV – Mid-Sized Quadrupedal Unmanned Ground Vehicle (Official technical data sheet for a robotic dog designed for military and law enforcement purposes, compatible with live-fire weapon platforms) — Ghost Robotics Vision 60.
• Deutsche Welle (DW) Tech: Civilian AI Surveillance and Data Fusion in Europe (The spillover into the civilian sphere of AI-based mass surveillance systems used on the battlefield and in major cities, as well as Palantir Gotham software) — DW News Tech Report.
• Kharon Geopolitical Research: Unitree Robotics and PLA Procurement (The convergence of civilian and military developments; the use of armed robotic dogs in the Chinese military’s domestic and overseas exercises) — Kharon Briefing.

• Asimov and the positronic brain: Classic book cover illustrations and early sci-fi concept art from Isaac Asimov’s Robot series (Press archive / Fair Use).
• Ghost Robotics Vision 60 / Codiaq: Official military demonstration and press photographs from Ghost Robotics and Skyborne Technologies — Ghost Robotics Media.
• Unitree armed robot dog: Stills from the People’s Liberation Army (PLA)’s official video broadcast showing automated machine-gun units — Unitree Press Release.
• Phantom MK-1 Humanoid: Battlefield photos provided by Foundation Robotics to the Ukrainian Ministry of Defence and the international press — RBC Ukraine Photo Gallery.
• Urban AI drone surveillance: Illustrations of the Metropolitan Police (London) and European police Live Facial Recognition camera systems (Creative Commons / DW archive).

Number of post views: 26