(or: Yes, AI is Coming For Your Job)

The industrial revolution is not over. It just has a shiny new(ish) face as artificial intelligence. Despite what you may think AI is, and despite the disappearance of smokestacks and export surpluses in places that used to be the image of industrialization, an industry is still very active. It continues to set about remaking societies through technologies that reduce or eliminate human labor and to carry out the interests of consumption and international domination. And in so doing, is presented to us shrouded in narratives of misleading nonsense about innovation and progress while doing tremendous damage to the natural world.

There is a great deal of hyperbole about artificial intelligence. On the one end of the hyperbolic spectrum are dystopic predictions of systems or machines that are so much smarter than humans that, upon becoming ‘self-aware,’ they will decide that humanity is dispensible and set about enslaving or destroying us. We can call this the ‘Terminator Doctrine.’ On the other side are grand hopes that AI will solve our knottiest problems, from racist policing to environmental catastrophe by making important decisions using pure rationality that is freed from human passions. We can call this the ‘Perfectionist Doctrine.’ Both of these doctrines are appealing for different reasons, but they are bullshit for the same reason, which is that they fail to account for what AI is actually for.

The appeal of hyperbolic AI is that it conjures a future of gleaming machines embodied with an intelligence that is ruthlessly emotionless and capable of seeing a bigger picture than pathetically limited humans. Such images tempt us through our insecurities and frustrations about the messiness of human relations and individual desires for perfection. A mix of science fiction and sober (if naïve) academics have laid the groundwork for these fantasies, and this what they are; fantasies. Despite what researchers, boosters, and lazy journalists like to assert, AI is not here to save us. AI exists for one reason and one reason only; it exists to reduce labor costs. Virtually every type of AI that attracts investment or a customer base automates some process that humans currently do with the promise of requiring fewer of them to do it. This is what makes AI just the latest wave industrialization.

For every heartening story about AI that detects tumors or provides way-finding for visually-impaired people, there is the real mission of AI, which is to carry out the whim and will of corporations and militaries. If there wasn’t the promise of incredible riches or world domination in the adoption of AI, there would not be AI to support accessibility or cure disease. There’s not enough money in that.

Sure, AI research and development is monstrously expensive and AI systems consume eye-watering amounts of electricity and other precious resources, which makes the up-front investment significant. But employing humans is far more expensive overall. The math of AI is not just the computation required to determine if here is an image of a pumpkin or a bicycle, it is the math of industrialists looking to bring down the most significant cost they face – employing humans; a cost made far costlier when those humans demand safe working conditions and a fair share of company profits. And, as with the manufacturing dimension of industry, AI is a planet-killer. The incredible amount of electricity and raw materials required for sophisticated AI systems is staggering. And the environmental costs to extract, process, and later dispose of, the toxic ingredients in digital equipment falls mainly on poorer nations.

The appeal of AI – its selling point – is that it performs ‘better’ than humans at various tasks, like handing out prison sentences that are not explicitly racist. Unfortunately, time and again we find that technology doesn’t eliminate our social problems but instead repackages them. The only algorithm that could be free of racism or other fundamental flaws would have to be so exquisitely designed that the cost and effort to produce and maintain it would probably defeat the efficiencies it is meant to create. Arguably we’d be better off continuing our efforts to attend to the core social problems we face as human beings than trying to replace ourselves with something ‘better.’

One way to understand what AI is for is to consider who it is for. The simple answer is that AI is for the people and entities who invest the most money into its development and who stand to earn the most from its acceptance. Despite utopian visions of an apolitical AI that frees us from drudgery and enlightens us with wisdom, virtually all AI arrives in our lives with an agenda; one set by its producers to serve either themselves or a target customer who is probably not you. Consider Siri. That helpful bot in your phone or laptop. Siri does what a giant corporation wants it to do. Quite often you’ll notice that it carries our ‘your’ wishes by employing products and services made by the same company. Curious that. If Siri happens to help you in the process, well that is only because there is potential profit for the company in doing that. It is not a strings-free gift to you; it is a marketing mechanism. Much more directly profitable AI includes systems for selecting job candidates, setting insurance rates, or producing robot soldiers. The target customers for these systems have enormous budgets making them extremely profitable to serve. The most exciting and innovative AI is not designed for you and me. We are paltry customers for AI by comparison.

If you are uncertain about this argument, I suggest you conduct a simple exercise. Every time you hear about an AI, be it a military drone, a grammar checker, or a companion for elderly people, ask yourself how many human hours would be required to perform the same task. Couple that with the question about how much money could possibly be made from AI that only exists to relieve some form of human suffering, however defined. Once you consider the economics of AI, the answers become clear. And those answers will most likely reveal what AI is for: it’s here to take your job.