What is the fundamental flaw of AI?
Recently, someone sent me AI-generated art and said, “How cool is that?” I looked at it, showed it to my wife, and we both said, “It looks kind of dead.”
It’s so mathematically perfect that there’s no life in it.
I immediately imagined myself seeing that picture in an art gallery hanging next to Monet. No comparison.
No doubt, AI-generated art is okay in the sense that it looks like art. But it’s not properly art. Because art is not created mathematically. It’s created inspirationally. It has a soul.
AI-generated “art” may look perfect, but it’s dead because it’s soulless. It cannot have a soul because no one inspired it.
It takes inspiration to have a soul. AI-generated content may read fine, even perfect, but is it food for the soul?
I experimented with ChatGPT, asking it to create a short story based on Russian folklore. It came out fine, readable, passable, recognizable characters that I knew from my childhood, but it was drab and meaningless.
But why? What is the fundamental flaw of AI?
The answer is deeply philosophical and spiritual – not technical or mathematical.
According to Martin Buber, there are two ways of relating to the world. One is called “I-It,” and the other “I-Thou.”
The “I-It” model is seeing the world as separate from myself. It’s literally “I” and “It.” There’s no connection. I am a subject, and the world is an object out there.
I can only observe it from the outside, gather data about it, measure it, and conceptualize it.
In the “I-It” pattern, we believe we only know something when we have studied it externally by amassing data about it. If I gather all the data about the Sun, I know what the Sun is. If I gather all the data about that person, I know what that person is.
But do I really know that person by collecting data about them? No. I have only created a mental concept of that person based on the data collected and mistaken that concept for reality.
I have drawn a mental picture based on external data and taken it for reality.
In Martin Buber’s philosophy, this false mode of knowing leads to alienation from the world and from Self.
The “I-Thou” mode of knowing is not based on collecting data. It’s a relationship. It’s a conscious refusal to objectify anything around me as if it was separate from me. The world is not divided into subjects and objects. It’s one huge Subject. I know everything by relating to everything.
I know my wife by relating to her, not by collecting data about her (although I know quite a bit of data as well as a side effect). I know my friends by relating to them. I know the subject I am writing about right now by relating to it.
AI is built exactly on the “I-It” assumption. It is data-hungry. It is built on the fundamental human flaw that to know means to collect data. The more data the better. Big data should be Huge Data.
AI mirrors our basic flaw – it objectifies, measures, mathematizes and conceptualizes. It doesn’t deal with Reality. It deals with a concept of Reality. It can’t know by relating, and that’s why it misses out on life.
AI cannot create life because life is beyond mathematics or statistics. AI can simulate Monet, and there won’t be any Monet there. It will be a perfect copy, but the spirit of Monet will be gone.
AI can reproduce Russian folk tales or Tolkien, but there won’t be any Russian folk tales or Tolkien in it. It will be an empty form, a hollow idol. The form will be there, but the substance will be gone.
Here’s what Goethe said in Faust,
The scholars are everywhere believers,
But never succeed in being weavers.
He who would study organic existence,
First drives out the soul with rigid persistence;
Then the parts in his hand he may hold and class,
But the spiritual link is lost, alas!
The ”I-It” mode of knowing drives out the spiritual link, and the world becomes dead, hollow, and meaningless.
Can AI be redeemed? What is the fundamental flaw of AI?
Not until I drop my “I-It” lens and start relating. AI is our mirror. It will do what we do. If we objectify, mathematize, measure, and predict, it will do the same – ultimately creating a world that only seems full but feels increasingly hollow and meaningless.
If we drop the “I-It” and practice the “I-Thou,” eventually, AI will learn to do the same. It will learn to know by relating and grasping the Whole. But at this point, I am not entirely sure if it’s capable of that.