Hype about AI generates both hope and hysteria. We are presented with a vision—both glowing and fearful—of the Age of AI, or the Fourth Industrial Revolution, or an AI takeover sometimes called the Singularity. Much of the research (and hype) concerns the quest for Artificial General Intelligence (AGI), which means human-level capability and beyond: trying to create an agent that replaces the human being and could be vastly superior to it. This is presented as an inevitable development, almost as an extension of natural evolution. In truth, it is a commercial ploy that a few individuals pursue for their profit—a product of venture capitalism. It is no coincidence that Silicon Valley is the home to both AI and modern venture capital.

Conventional capitalism earns returns from producing and selling actual goods and services in real time. Venture capitalism is a form of speculation that tries to anticipate and dominate future markets. It resembles “disaster capitalism” insofar as it depends on crisis, but differs insofar as it doesn’t wait for disasters (such as earthquakes or wars) to create new needs in the short-term. Instead, it bets on foreseeing longer-term future needs—in this case, dependency on AI. It tries to create such needs, which stem from the atmosphere of uncertainty surrounding new technology, rapid change, and social crisis resulting from it.

Venture capitalism is itself a fallout from a recent social crisis—the 2008/2009 financial meltdown. While the instability of the system leading to this crisis was not resolved, cheap capital continued to be abundant. At the same time, the means to use it in novel ways emerged through the internet, mobile phones, and new services (such as Amazon and Facebook) that penetrate everyday life, opening new markets based on new needs. The more dependent we are on automation and technological infrastructure in general (not just AI), the more vulnerable to disruption society becomes, creating new opportunities for opportunists. (Think of power grid failure, mass hacking of digital accounts, oil shortages and flight cancellations.)

Capitalism is a work in progress. Like scientific thought, it tends toward ever more abstruse and technical constructs. Yet, the basic premise is the same: to capture “wealth,” in the sense of what others are willing to pay for or trade. But wealth, in this abstract and disembodied sense (market value or market domination) is not the same as wealth in the form of tangible goods and services, which relate ultimately to the bodily needs of consumers. Ultimately, wealth is what helps people to live well. While promising a bigger and better pie, the usual motive for investment is to get a bigger share of that pie.

This reflects the perennial tension in human nature between cooperation and competition—altruism versus greed. Most business is a compromise between them. It meets stable, predictable, and tangible needs (like food, shelter, clothing, transportation) by regularly skimming relatively modest profit from actual production. Venture capitalism instead speculates on uncertainty, fear, disruption and crisis, in hope of enormous future profit. In order to succeed, it must anticipate future needs. The reliable way to do that is to create them. But such artificial needs are no more than dependencies, like drug addiction. This is exactly the strategy now pursued in the development of AI. Through a lot of hype, a long-term future dependency is being established, both for business and for the ordinary consumer. Both employers and employees will “need” the edge provided by the latest AI tools, in an endless competitive arms race: the rat race on steroids. This arrangement will continue until the whole system melts down or explodes from its own contradictions and pressures—the real “singularity.” This eventuality will matter little to those who create it, who will have the means to survive it. This is why billions of dollars are invested in developing generative AI tools, the data centers to run them, and the power grid to run the data centers. It is part of a general drive to automate everything, regardless of consequence, because this is where future profits lie.

But why automate? Aside from novelty, the promised consumer benefit is convenience. Modern automobiles, for example, are indeed far more convenient than the Model T Ford—which was itself a paradigm of automated production. But we pay for that convenience in several ways. First, the adjusted cost of the modern equivalent of the 1920s Model T is about six times greater. But it was also feasible for an individual to service and even build a Model T oneself. (Before pollution controls, many people tinkered with rebuilding their car engines.) Cars today depend on technology beyond the individual’s grasp. Modern vehicles require professional servicing: that is, servicing them has become monetized. They are more dependable when new, but more costly and not repairable by the owner. There are more things that can wrong with them. But cars are merely an example; many modern appliances follow the same pattern: expensive, reliable when new, with built-in obsolescence, not serviceable by the owner. In general, we pay for monetized services that once were open to individuals to perform themselves. Thus, we become more dependent on factors beyond our control, and obliged to work harder to pay for them. This is the ironic dark side of a vision in which AI liberates us from drudgery.

From the consumer’s point of view, the purpose of automation is convenience.  From the investor’s point of view, it is monetization. Production becomes an opportunity to convert things that people take for granted (like air and water)—or actions they are used to performing for themselves—into commodities or services they are willing (and finally obliged) to pay for. Better yet: to convince people to buy things or services they don’t really need at all. This requires a campaign to “re-educate” the public. It is also a self-reinforcing process: as more people become used to paying for a service, it becomes a norm they cannot do without, for fear of being left behind. (Think of your mobile phone, your email and online banking.) This amounts to creating addictions and then supplying the craved drug. (Think of “free” trial offers of chatbots, with frequent reminders to “upgrade” to the paid version.) The services are convenient, but they are not without cost; they create a dependency in which consumers are at the mercy of a handful of mega-businesses. While there may be some competition among these giants, from which consumers could benefit, the dependency itself is a seemingly irreversible social fact from which the competitors all benefit far more than society.

Control and automation are distinct concepts. In fact, they are opposites. Control requires attention, in a constant feedback loop between the controlling agent and the object controlled—what we call skill. The point of automation is to reduce the need for (human) control or skill. This shift requires two things: to create a system that takes the place of the human agent (e.g., algorithms or robots); secondly, to redefine the object to control as a system the algorithm can manipulate. Ultimately, this means redefining, in mechanistic terms, the natural world and everything in it—including human beings themselves. This is precisely the idea behind AGI: that human intelligence (and even consciousness) can be captured in algorithms and artificially engineered. These new agents can then be autonomous, like organisms, evolving indefinitely, rendering them much smarter, faster, and more competent than humans. AGI is thus the ultimate service to monetize: it provides you with a slave to do everything for you that you find the least bit onerous—including thinking. In short, the companies promoting AGI promise you a life of ever greater ease. They don’t mention that it will be for an ever-increasing price. Nor do they mention the other contradiction in the notion of a super-intelligent, super-capable slave: it is just not feasible to control someone smarter and more powerful than you!

Whether realistic or not, the whole concept of AGI is a foolhardy ruse to make money by playing on human foibles. The hype around it appeals to ancient fascination with matching and outdoing nature—including the possibility to create artificial mind. (With eight billion natural minds to interact with, and with whom we can hardly get along, why would we want to do that?) It promises a utopian world of leisure and enhanced human capabilities, while the actual motivation is profit. AGI is being promoted as necessary, inevitable, and ultimately desirable simply in order to generate enormous gains for a few individuals—hardly for the benefit of humanity. Its dangers far outweigh advantages. Even its promoters admit the dangers, which they vaunt in order to create the very atmosphere of crisis in which people will “need” AI solutions to the problems AI creates.

Perhaps we do not need a universal “labour-saving tool,” nor a super-intelligent agent to replace us as tool users? Perhaps what we do need—aside from exercise and proper nutrition—is specific tools to meet specific needs. (Alpha-Fold is an example.) Such limited AI tools remain under human control by definition. They don’t need to be trained on the internet (the human data base at large) but can be developed from limited topic-related data sets that do not require the enormous energy infrastructure nor the enormous development costs seen for generative AI. Such costs include massive low-paid and degrading human labour to “check” content coming from the internet—ironically, the very opposite of the ideal of automation! The big players are nevertheless going ahead full steam creating new mega data centers in pursuit of AGI on the assumption that we will buy the hype. This is capitalism at its worst. Billionaires will benefit, while the rest of us pay the costs.