All our historical inventions – from the wheel to a microprocessor or a PC operating system –  were designed to execute specific tasks. Artificial intelligence (AI)  is different. In theory, we expect it to execute any imaginable task.

Technology-based business ventures generally seek to simplify or even replace existing human activities. Whether you are seeking to launch a screwdriver or a word processor, to test the logic of the business you must first ask a simple question: “Will people use it?” You then look carefully at existing human activities to get a clear idea of how your product is going to be used.

In the 15th century, someone invented screwdrivers. This was a true innovation since screws do not exist in the natural world. AI is different. Intelligence already exists. It is both plentiful and reproducible. Humans appear to possess the most sophisticated version of it. It’s a tool they don’t have to pay for and it’s conveniently located inside its natural protective casing, our skull.

Fortunately for the promoters of AI, the science of marketing contains a second principle, beyond the criterion of utility. People are never satisfied with what they already have, even their own intelligence. Moreover, our years at school teach each of us to believe that our intelligence is defective and inadequate.

The screwdriver’s inventors didn’t know what specific structures people would build with the tool, but they could imagine what kinds of things: houses, furniture, vehicles and machines. More significantly, they could visualize people using a screwdriver. The same can be said about an operating system or a word processor. You can literally see in your mind’s eye the novelist, coder or secretary sitting in front of a keyboard making your system function.

Artificial intelligence is different. We cannot visualize its use. This should help us to understand the burning questions that have contributed to the raging debate over AI in recent years. Recent headlines have featured public dramas spawned by the uncertainty concerning where all this is headed.

Gizmodo reveals that, after November’s attempted coup d’état at OpenAI, a new case may be brewing inside an emerging startup. “Anthropic is reportedly in talks to raise $750 million from Menlo Ventures, which values the startup at $15 billion, according to The Information. Like OpenAI, Anthropic also has a complicated board structure with misaligned values, and this recent influx of cash begs the question: is Anthropic bound to be the next AI board showdown?”

Today’s Weekly Devil’s Dictionary definition:

Misaligned values:

1. Contradictory moral or strategic orientations within a group expected to agree and act upon achieving a common goal.
2. The natural state of the culture found in high-tech enterprises that claim to be dedicated to providing tools to meet the needs of the public, but whose reasoning will inevitably be guided by an overriding law: seeking maximizing profit, even at the public’s expense.

Contextual note

I mentioned earlier that throughout history, when launching a new technology, the inventors see in their mind’s eye the person using the product. We know that Thomas Edison visualized an entire nation illuminated by his invention of the light bulb. He wittily proclaimed his vision of every dwelling in the US bathed in artificial light: “We will make electricity so cheap that only the rich will burn candles.” Artificial light had tangible meaning. Its utility was visible in ways that AI can never aspire to.

Nikola Tesla, Edison’s contemporary and rival, saw even further into the future. He predicted an age of robots: “In the twenty-first century the robot will take the place which slave labor occupied in ancient civilization.” Tesla could say this because he and everyone else could visualize what slaves did. His imagination literally figured robots replacing the tasks of slaves. He also presciently believed intelligent machines would exist. But he saw them executing known tasks, not thinking randomly. Elon Musk has consistently promoted the vision of his cars executing all the tasks human drivers are required to master. That vision may have been a little too optimistic. Musk is now being sued for failing to deliver his promised “full self-driving” vehicles.

The website Pressfarm offers this description of Sam Altman’s vision of ChatGPT’s ultimate goal: Artificial General Intelligence. “AGI, often referred to as ‘strong AI’ or ‘full AI,’ represents AI systems that can perform any intellectual task that a human can do. It is the point at which AI becomes capable of outperforming humans in nearly every economically valuable work.”

This vision of AI in the future appears comprehensible so long as it is stated as a “logical” principle. The logic often dissolves as soon as the inventor attempts to move from the abstract to the concrete. Before the screwdriver existed, its actual use had to be imagined. But as soon as the inventor imagined it, he could envision people physically turning the screw. He had no idea how pervasive its use would become, but he knew some people use it. The same is true concerning the invention of the word processor.

As of today, we cannot see what AGI will do, even less how humans will act in its presence. We can only hope it does whatever it does well and within acceptable boundaries. But we have no idea what those boundaries are. We are even faced with the uncomfortable hypothesis that it may be up to AI to define those boundaries.

Historical note

Because we have no clue as to what AI will do, misalignment among those responsible for creating and promoting it is not only possible, but likely. The boardroom drama of OpenAI illustrates very clearly how that misalignment will predictably work. It will follow the simple — if not simplistic — logic of modern capitalism as defined by its champion promoters, Friedrich Hayek and Milton Friedman.

Hayek was a great theoretician, meaning that he preferred to live and think in a world of pure abstraction. He invented and marketed ideas. As so often with such people, he was an inveterate optimist. That meant that he could remain blind to any reality that failed to conform to his theories. Or if he perceived it, he would be ready to dismiss its manifestations as an exception rather than the rule. “Our faith in freedom,” he proclaimed, “does not rest on the foreseeable results in particular circumstances but on the belief that it will, on balance, release more forces for the good than for the bad.” 

Just as Karl Marx supposed that rationality and a belief in the value of equity would lead a new proletarian order to seek the good of all rather than the obsession with power of an elite, Hayek supposed that commitment to freedom would dispel the abuse of monopoly. Both made a good case in the abstract for the efficacy of their theories, but reality failed to conform to their wishes.

Friedman doggedly adhered to the same abstractions as Hayek but at least recognized the problem of misalignment when he observed: “Businessmen are in favor of freedom for everybody else but not for themselves.” In other words, his ideology of free enterprise in reality spawned businesses that betrayed the theory’s pure principles. 

In reality, the sacred principle of maximizing profit for shareholders would always win out over the magical workings of Adam Smith’s invisible hand, so long as the resources exist to back it up. Monopoly not only works, it has become a dominant ethos in the business world. Hayek warned against monopoly as a perversion of good economics. But a system that holds as its fundamental mechanical principle that monetary value (shareholding) should grow through profit will always see monopoly as an ideal. Two principles end up overriding everything else: growth is good and cash is the sole reliably objective measure of value. In other words, to quote Gordon Gekko, “greed is good.”

If a self-driving car runs off the road or crushes a bicyclist we can say that its artificial vision was somehow misaligned. When an inventor creates a new product without envisioning how it will work in concrete terms, the goals and values they imagine but fail to visualize will most likely be misaligned. When, to make a new invention work, billions of dollars are required and those billions promise produce more billions, misalignment – as Milton Friedman himself appears to warn – is going to be inevitable. On that scale, ethical value simply cannot compete with monetary value.

*[In the age of Oscar Wilde and Mark Twain, another American wit, the journalist Ambrose Bierce produced a series of satirical definitions of commonly used terms, throwing light on their hidden meanings in real discourse. Bierce eventually collected and published them as a book, The Devil’s Dictionary, in 1911. We have shamelessly appropriated his title in the interest of continuing his wholesome pedagogical effort to enlighten generations of readers of the news. Read more of Fair Observer Devil’s Dictionary.]

The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.