What have a Greek philosopher who lived about 2300 years ago, a Dominican friar, theologian, and philosopher who died in 1274 A. D., and artificial intelligence (AI) got to do with each other? Sounds like the beginning of a bad joke. What I have to say is no joke, however, because if we don't heed the wisdom of these two parties, which has been almost totally forgotten, we are liable to make some mistakes shortly that will be exceedingly costly in terms of missed opportunities, misdirected hopes, and maybe even lives.
The particular wisdom I speak of is known by the technical term of scholastic metaphysics. It was, and still is, a way of viewing the universe, and it dominated the West for over a thousand years. The Scientific Revolution vigorously uprooted it and threw it in the trash pile, to oversimplify a very complicated history. And now it is almost completely forgotten except by a few lucky people such as yourself, if you keep reading.
The aspect of scholastic metaphysics I wish to explain concerns the nature of things. Here's an example, taken from an article by journalist Mary Harrington in the June/July issue of First Things. Take a dog, any dog. It may be black, white, spotted, or bald. But none of those attributes change the fact that it's a dog. The scholastic metaphysical term for the dogness of a dog is its substantial form, also known as essence. The aspects of a thing that can change without changing its substantial form are called accidents. That's a weird use of the word, I know, but I'm not going to argue with Aristotle, who came up with these ideas, and Aquinas, who put them in terms compatible with Christianity. The substantial form of a thing is its organizing principle. The accidents are characteristics that can change without altering the substantial form.
What about people? We have an organizing principle called the soul. No matter what accidental characteristics we have—whether a person is fat, thin, from Teaneck, New Jersey, or in a coma—that substantial form remains.
What about computers, and all the systems that are known by the general term of AI? AI is, of course, artificial—that is, it is a man-made artifact. Unlike living things, an artificial object has no substantial form that is prior to what its human makers say it is. We can cobble together a bunch of transistors, call it a computer, and get it to do certain things for us. But computers, and the AI systems supported by them, are simply assemblies of accidents. This makes AI systems fundamentally different in kind from human beings.
What emerges from looking at humans and AI from the viewpoint of scholastic metaphysics is this: human beings are just a clean different kind of thing from AI systems. And treating AI systems and humans as though they were basically the same kind of thing, simply piles of atoms arranged differently—which is how modern materialistic science encourages us to think—ignores a vital difference between the two. Human beings have souls as their organizing principle, and AI systems do not. No imaginable or unimaginable future progress in AI can change that fact.
This viewpoint flies in the face of the transhumanist agenda so popular in Silicon Valley, which looks forward to the day when human beings can transcend their messy, disease-and-death-prone meat-cage origins and live forever in an eternal bit-world of computer-borne bliss that will take over the universe. (I may have left something out in that description, but I think I got most of it.) Human beings are the rational animals that result from propagation by other human beings. Anything you might upload from a human being to a computer is not going to be that human being. And treating it, or any other AI product, as if it were human is a category error that can lead to heinous consequences.
In Harrington's words, trying to teach sand to think is not only futile, but if we start treating the result as though it had a soul, we are in danger of losing our own souls.
This may sound like just another anti-technology rant, but it's not. AI can do truly amazing and useful things, but only if its human users keep in mind what it is and don't try to endow it with quasi-human agency. A tool should be under the control of its human users at all times. That control may be loose and barely perceptible sometimes, but it has to be there. If we begin to bow down to AI entities as though they were thinking humans like us, we will soon regret it. An automobile provides useful service when under control, but if it gets out of control it becomes a dangerous missile. The same thing is true of AI.
Thinking in terms of scholastic metaphysics seems strange to most people unfamiliar with it, and it takes a while to realize that it's just as useful as it ever was. I have had a passing acquaintance with it ever since I read a book on Thomistic philosophy in the 1990s. The thing that drove me to that book was a statement by a former theoretical physicist turned pastor, who said that the Catholic Church's notion of transubstantiation of the bread and wine of the Eucharist into the body and blood of Christ was based on an outmoded theory of matter. I wanted to know more about this outmoded theory, and it turned out to be scholastic metaphysics.
What modern science did to scholastic metaphysics is a classic case of throwing out the baby with the bathwater. There was a lot of useless bathwater in Aristotle's physics—rocks having a desire to fall downward and so on—but there was also great wisdom in his metaphysics. The wisdom of scholastic metaphysics can save us from making lots of mistakes that could be avoided if we just keep in mind that however clever and human-like AI appears, it's really just words written in the sand. And we wrote the words to start with.
Sources: Mary Harrington's article "Thomophobia" appeared on pp. 19-27 of the June/July 2026 issue of First Things. For those wanting to know more, she refers to Edward Feser's book Scholastic Metaphysics (Editiones Scholasticae, 2014, $34 on Amazon).
No comments:
Post a Comment