Billions of years ago, the Earth populated an orbit shared by myriad asteroids. A bombardment of these asteroids with our Earth has cleared our orbital path and added minerals, such as lithium, that would otherwise be unavailable to we Earthlings. We need not expect major asteroid collisions in our future, although we may want to be on our toes for some minor ones. 

What does that have to do with economics? It means that whatever resources the Earth provides to humankind are in fixed supply. That’s unique compared to other factors of production.

When economists talk about production, we describe an abstract function that translates into output various inputs, such as the fixed Earth factor E, our human capital L that we can procreate over generations, and the machines we build, which we call physical capital, and use the symbol K, from the German word Kapital (thank you, Karl Marx). Economists then write that output Y is created by combining in an abstract and opaque function f these factors E, L, and K:

Y = f(E, L, K)

I promise, that’s the end of the math lesson. 

Let’s look at the danger of treating these factors as equivalent and equally fungible, and then introduce a new way to understand Artificial Intelligence (AI). Regular readers of this column have followed concerns over unemployment, lower wages, and worsening income inequality that shall be unleashed by AI. An additional challenge to economics is the need to develop tools that specify rather than blur the relationship between human capital and machines. 

Where does AI fit into our equation for production? AI and robots are certainly not Earth capital. AI shares qualities with machines, including those that are able to produce more robots. In other words, some machines can replicate themselves, some might even say procreate themselves. 

AI connected with robots can then perform a number of functions. They can procreate and they can learn. They can improve their design over time, either by design or by Darwinian evolution. But, while humans take centuries or millennia to iterate into improved designs, machines can create, scrap, and recreate themselves as rapidly as they wish. They can also ensure that their electronic DNA evolves and is passed fully onto their offspring, not merely through Mendelian chance.

It appears, then, that physical capital in the form of AI and robots overcome the limitations of Earth capital and human capital. They do not suffer a ploddingly slow ability to create new capital. Our human salvation is that we rely on one human quality that machines don’t have, or so we think. We are sentient. Are they?

In 1950, the great WWII codebreaker and computing pioneer Alan Turing developed a Turing Test to determine if a computer is capable of intelligence. Can a strong plurality of people conversing with an entity for five minutes determine whether it is human or machine? Recently, people conducting such a test have concluded that Artificial Intelligence machines appear to them to converse just as would a human. These machines are only going to get better. 

The pace of improvement in AI is phenomenal. A shovel-full of sand and bits of copper are converted into elaborate silicon chips capable of calculating, remembering, learning, and procreating at a rapid rate. They can generate new and improved generations of themselves many times in one year, rather than once every two or three decades as do humans. Intel founder Gordon Moore postulated that computing power doubled every 18 months, but that was in an era which relied on human intervention. Left solely to their own design iterations, AI machines may be able to double their capacities even more rapidly. 

We have discussed in past blogs the economic displacements that may occur over the next few decades based solely on what jobs we now know or can expect AI can perform. These economic innovations are already predicted by Goldman Sachs to constitute about half of the economic growth over the next few decades, even without the exponential growth and the accelerating rate of innovation for which AI is capable. 

The great and underappreciated early 20th Century economist John Maurice Clark, son of America’s first famous economist, John Bates Clark, noted a hundred years ago that humans willy-nilly assume more machines translate into progress. He warned us that:

… we see no more than a single machine … nothing more than the sum (of its parts). We lack the imagination to suspect that we may be dealing with that order of being whose whole is more than the sum of its mechanical parts - in short, with something organic. Not a conscious being, perhaps… Let us drop … the provincial habit of looking at everything on earth from a human point of view, and try to interpret things from the standpoint of machines as a race of beings, having their own needs and vital forces. To (these machines), men would be a convenient instrument of production. 

Let us add to Clark’s prophecy that machines now act as if they are sentient and conscious, and can be designed to be resilient, reproduce, and self-protective, not unlike HAL 9000 (one letter removed in each character from IBM), the computer from the dystopian movie 2001: A Space Odyssey that switched from an ally for astronauts to an adversary when it took over a spaceship. Clark suggests that superior machines may employ human and Earth capital, rather than superior humans who employ physical and Earth capital. The table has been tipped. The pyramid has been flipped.

The human response to a dystopia in which we work for machines, rather than the other way around, is that we can always unplug the machines. What happens when AI is unleashed to optimize and run the electric grid, secure the nuclear power plants, and protect our sources of energy production from tampering or terrorism? In its infancy, AI already uses more energy worldwide than mining of cryptocurrency, equivalent to the power consumption of some developed nations such as Sweden or Ireland. This sector will demand and perhaps even divert more energy, at an exponentially growing rate, to fuel its growth. 

Even if most humans try to limit this expansion of a new alien race “living” alongside us, there are always some humans with the greed or optimism to see how far the omnipotent AI machine experiment will take us. As Clark noted, we should not underestimate our willy-nilly faith that machines will fuel our growth, rather than our harder or smarter work. AI can work harder and smarter than us already. It is hard to imagine what it will be able to do in a decade or two. 

Clark added that AI may outsmart us. 

“By teaching man that he is the end of all things, when he is not, his subjection is concealed and thus perhaps perpetuated…. The machines have domesticated man to their service in such shrewd fashion that only a few have been dimly aware of what was happening to us… From Homestead to Hollywood, the machines have reared cities after their own needs… And humans themselves have been pre-empted from the stage by these creatures. (The) working unit is no longer a man, or any group of men, but (the) “plant.” … (Man’s labor) often lacks variety and still oftener lacks openings for initiative and discretion. 

A century ago, Clark recommended a “reasonable degree of racial equality” with this new and inanimate life form. 

“When this is achieved, we can deal with machines in that spirit of confidence which characterizes the bargaining of equals… (We need) economic machinery for bringing the collective human judgment and will to bear at the point where things are being decided, in the process of industry itself, rather than waiting till the decision is made by the engineers, and captains of finance, and then, through our ‘political machinery,’ taking belated and purely defensive action.”

To be captains of our own ship, we must think about our economy entirely differently, as readers of this blog well know. We must view our world and economy as dynamic, a moving picture, rather than a series of static still photos. Unfortunately, we are often so preoccupied at what is happening this week or month, and rarely exercise the necessity to navigate what will happen next decade, generation, and century. Given our short-termism, my bet is on this new alien race. 

Rather than wrestle on whether humankind works for machines or machines work for humankind, perhaps both entities should work for the sustainability of Earth at this point.