Arms Races At The Speed Of Light

Flexible Manufacturing and Weapons Technology in The Information Age

 

Dual-use weapons are as old as mankind, you can bale hay with a pitchfork or plunge it into someone’s chest, a base-ball bat can hit a home run or smash open a skull; everything from bicycle chains to chainsaws and, most recently, cars can be used to injure – and even kill – other human beings. People will always have access to the tools of death. Freedom to use a wide range of tools in a variety of activities necessarily enables individuals, with sinister inclinations, to abuse that freedom and kill others. We can only punish them after the fact.

But what if a dual-use tool could enable an average person to kill thousands, perhaps millions, of people in a short space of time? What if the benefits of this technology were so great that countries which banned it would plunge into recession?

 

Does such a technology exist?

 

Yes. It’s called flexible manufacturing and its future implications are as terrifying as they are unavoidable, with clear solutions all but non-existent.

3-D printing – the general term for manufacturing processes that can convert a digital file into any arbitrary 3 dimensional shape – is the poster child for flexible manufacturing. A rudimentary 3D printer costs a few hundred pounds while big brands, like Makerbot, might cost a few thousand. Although off-the-self 3D printers can make arbitrary shapes, the variety of materials they can use is limited. However, commercial 3D printers work with many more materials and can print aircraft and even spacecraft components! Some 3D printers can even build most of the components that are required to replicate and even upgrade themselves.

Skeptics of 3D printing technology point out that it is still often more expensive than other manufacturing techniques and that 3D printed parts are often lower quality compared to other methods…

…but 3D printing is only the most dramatic example of a general, inexorable trend for all types of manufacturing systems to become cheaper, smaller and more flexible. CNC machines can also manufacture a vast array of components of all shapes and sizes through simply downloading a file with the right information in the right format. And the cost of a rudimentary CNC machine is also £200 or so – about the same as a 3D printer.

On a factory level, 3D printers, CNC machines and other automated processes can be integrated into flexible manufacturing systems  that can rapidly switch between producing completely different products simply by inputting new instructions.

The economic case for flexible manufacturing systems, that can rapidly respond to changes in consumer demand without expensive retooling, is compelling, and, as time goes by, these systems will inevitably become ever cheaper, ever smaller, and ever more flexible.

A universal manufacturing system is the logical conclusion of this trend. A set of automated tools and robots in a small space that could manufacture anything – surgical instruments, lawn mowers, aircrafts, guns, cars, robots, computers, mobile phones, bio-weapon laboratories, tables, chairs, cosmetics, androids, cutlery – and arbitrarily switch from making one product to making another in less than a minute.

Once this technology matures, and becomes affordable to everyone, then everyone will have the capability to manufacture sophisticated, lethal military-grade weapons platforms from the comfort of their own homes.

 

Flexible Manufacturing and The International Balance of Power

 

Automation is not just about job loss. It is also enables extremely rapid shifts in the coordinated behaviour of artificial actors and manufactured goods.

 

A side effect being the distinction between the ability to produce weapons and the ability to produce economic goods of all kinds – will disappear.

In general, it takes a year or two to fully mobilize a military for large scale war. Conscripted civilians must be trained as soldiers, factories must reorganize their workforce to build armaments, dedicated armament factories must be built. And only after manufacturing state of the art weaponry, can you properly train soldiers how to use it.

Once every country on Earth has universal (or highly flexible) manufacturing systems, the switch to a war footing will take minutes. Automated weapons platforms will emerge from factories with optimized battle-software that can overcome and destroy an enemy with maximum efficiency – no training, or generals, required. Any country with universal manufacturing capability and up to date design and battle software – even ones without any military at all – will rapidly be able to create the most formidable military on Earth.

Diplomatic relations between countries depend, among other things, on the implicit knowledge of how total war would pan out – especially for the loser. A nation that anticipates defeat will likely back down as disputes with stronger nations escalate. However, if both nations mistakenly believe they could easily defeat the other, the escalation towards total war is far more likely. World War 1 is what happens when combatants grossly underestimated the cost and duration of conflict and, consequently, make little diplomatic effort to avoid it.

 

Furthermore:

Increasingly flexible manufacturing systems, and automated military units, will eliminate the barrier, and lag time, from the acquisition of a blueprint to deploying the actual weapon in battle.

 

Let us assume, for the sake of argument, that in the 19^th century Brazil acquires the plans for a state-of-the-art British warship. Without the U.K.’s shipyards, a knowledge of how to build them or skilled workers to man them, merely possessing a warship’s blueprint is a far cry from building it. The Brazilian state would have to invest huge amounts of money and resources to attract people from England with the right skills and know-how, train its native population to build and work in the shipyard, acquire the right grade of steal etc., etc.,. This would take decades of concerted effort. Even once Brazilian shipyards were producing British warships, a crew would still have to be trained, British naval tactics studied, etc., etc., before the military capability of a Brazilian navy even approached that of the British. And by the time Brazil got its warships in the water, British warships would be even more advanced.

 

Imitation has always been cheaper than innovation. But, in the past, successful imitation still required a lot of effort. As the pace of globalization has increased, the speed of imitation has also increased – as the meteoric rise of China demonstrates. But in the future, the time required to acquire all the technological advantages (including military technology) possessed by a competitor will approach zero.

 

Technology is information. The time it takes to copy information varies with format:

• Digital information – Instantaneous
• Technical information (skills, knowledge) – several years
• Organizational information (interaction between workers in an organization) – years.

 

Skills must be learnt by human beings over several years. Skills have two components: reading and experience. Stealing books and reports from a competitor can certainly accelerate the training of one’s workforce, but the workforce of the imitator must still learn through trial and error. The same applies to organizational information, a company may have reports that define corporate policy and protocol, but there will always be an unspoken, implicit corporate culture overlaid on top which a competing rival can only develop through trial and error.

 

Automated systems store everything in a digital, instantly transferable format. When the designs of weapons (in the form of software instructions to a universal manufacturing system), the behaviour of weapon’s systems, and the protocols coordinating how different weapons platforms interact with each other as part of a coherent battle strategy, are all stored in digital format, then a single hack by an opponent could neutralize a technological military advantage that cost trillions to develop within hours.

 

For example, if, in a future where universal manufacturing systems are everywhere, the Syrian government hacked all the information possessed by the U.S. military and U.S. weapons companies. Then, within hours, Syria could put their universal manufacturing systems to work making fully-automated U.S. weapons platforms and become the military equal of America in less than a day! A technological military edge that cost trillions to develop could be lost to a team of hackers working for a small government on a budget of less than ten million pounds.

 

Arguably, high-level encryption could be deployed that may cost 100s of billions for a competitor to break through…

 

The problem is that if country A is a large superpower who has invested trillions into developing state-of-the-art automated military software and has heavily encrypted it to make it very hard to hack, country B is a rival superpower who has invested trillions into decrypting and hacking into country A’s military secrets, then if little nations C, D, E, F, G, H, I, J and K can get a copy of country B’s decryption software, they could potentially access all of country A’s military secrets for a fraction of the price that superpower B spent to initially develop the decryption software.

 

So military technology will become very leaky in the coming information age.

 

In this sense, Vladimir Putin’s comment that “the nation that leads in AI will be the ruler of the world”, is not accurate as that leader will likely get hacked, and lose all its hard won advantage to competitors in an instant.

Instead of one “AI superpower”, hundreds of independent sovereign nations, many ruled by shady dictators locked in regional power struggles with their neighbours, will all rapidly gain access to state-of-the-art technologically advanced, fully automated battle systems.

 

This is a recipe for global chaos.

 

The Criteria for Victory

 

The desirable design features of an automated weapons system fall into four categories:

• Victory
• Security
• Safety
• Cost

 

All categories involve trade-offs. A human in the loop may increase safety, but may also increase system response time and get defeated by a rival system. Robot swarms communicating with heavily encrypted messages, that can’t be intercepted, might exchange information at a slower rate than swarms with lighter encryption, enabling the lightly encrypted swarm to outmanoeuvre the heavily encrypted swarm. Similarly, a weapons platform that can only be built by a military-grade manufacturing system will churn out less units than one that can be mass-produced by generic civilian manufacturing systems. A weapons system that never attacks friendly units, or never launches an unprovoked strike against a neighbouring country might also be slow to fire at an attacker and get destroyed by it… and so on and so forth.

So designing a fully automated weapons system that achieves victory against all opponents while simultaneously being safe, secure and cost effective is anything but straightforward.

Nations can subjectively decide to focus on designing a safe AI weapon’s system but cannot ensure it achieves victory in battle.

 

On the whole, I think it’s likely that…

 

…the most dangerous developments in military AI will come from the weakest actors…

 

A powerful nation, confident of victory, will likely invest a lot of money into safe and secure AI weapons systems. A poorer country, on the verge of being invaded by a far more powerful foe, will throw everything into designing AI systems with maximum destructive capability, irrespective of safety and security, in order to prevail in a battle.

The U.S. deputy defence secretary, Robert O. Work tells us “there will always be a man in the loop”, but what if the U.S. decided to invade Iran with drone armies and the Iranians found they could achieve victory over America by taking the man out of the loop? Clearly such desperate, rushed measures taken by the losing side of a war could progressively increase the danger of AI military technology.

For example, what if an opponent hacks, reverse engineers and decrypts a rival nation’s battle software and swarming strategies in a degraded and incomplete form? What if their software designers do a rushed job to fill in the gap? Such imperfect, rushed attempts to replicate rival battle systems could produce weapons systems that are simultaneously highly lethal, highly uncontrollable and highly unsafe. Such rushed cyber-espionage jobs might even produce lethal weapons systems that spontaneously attack peaceful neighbouring countries by accident!

 

Weapons Proliferation to Non-State Actors

 

Imagine a simple battery-powered drone quadcopter, no larger than a dinner plate, with a dagger attached. Imagine this system is equipped with machine vision and manoeuvring software and is programmed to seek out human jugular veins, ram into them and then back out. Imagine it can achieve a kill rate, under favourably crowded conditions, of one person every 40 seconds. Imagine a swarm of 10,000 of these drones, can operate in a coherent manner like sheepdogs and corral and surround masses of terrified people before going in for the kill. Imagine these “flying daggers” massacring the residents of one city after another.

This would be an example of how sophisticated and lethal software could transform basic hardware into weapons of mass destruction. If we assume that a futuristic 3D printer which cost £500 could produce these drones a £3 a pop, then ten people chipping in £3,000 each could manufacture a swarm of 10,000 drones, download appropriate battle-ware into the drone swarm from the dark web, and kill whole cities filled with millions of people.

 

But surely states will keep their battle-ware safely encrypted?

While state battle-ware encryption may be too secure for non-state actor to design codes to penetrate, rival states will hire large teams of hackers to decrypt this battle-ware. These sophisticated decryption software packages will likely leak into the wider web. At this point, cults, companies, terrorists, mercenaries, drug gangs and small time hackers will all be able to access powerful decryption tools and obtain the designs of sophisticated weapons platforms as well as the software to control them.

At this point, everyone will have free-access to some of the most sophisticated weapons systems out there. How will the police cope?

 

Unfortunately, there are no easy solutions.

 

For example, a blanket international ban on flexible manufacturing for personal use, would be impossible to implement. How do you distinguish “personal use” from use by a small business? Any group of terrorists dedicated enough to want to destroy whole cities would also be willing to set up a small business. Or perhaps you could force all manufacturers to apply for a government-issued permit for every file they download onto a 3D printer, or manufacturing system, but what about people who design their own CAD files? Would they need to apply for a permit every time they manufactured something from a CAD file that they designed? If so rapid-prototyping would become a lot less rapid – and if not, then what’s to stop someone downloading something off the internet while claiming that they designed it themselves? The other issue is that flexible manufacturing is a sliding scale with no clear boundary. So all manufacturing would need to be heavily regulated with a hellish degree of red-tape. But this would make countries that don’t regulate flexible manufacturing vastly more wealthy, while the economy and quality of life in countries that did would diminish.

Cody Wilson’s 3D printed gun is the very small tip of a very large iceberg. If anything, he has done humanity a service by raising awareness of this critical issue early on before the shit REALLY hits the fan.

 

John

 

Do You Have a Burning Desire to Leave a Comment?

 

Have you found this article thought provoking? Is there some message you desperately want to communicate to future readers but can’t because my comment section automatically closes 28 days after my posts go live?

If so, you might be interested to know that I reopen any comments section to members of my mailing on request as one of the perks of joining.

If you’d like to leave a comment, simply scroll to the bottom of the page, sign on to my mailing list and them email me with a request to reopen the comments section for this post.

Happy Commenting!

John