The Inventor of the Thinking Machine Didn't Worry. Neither Should You

We're all worried about artificial intelligence. At my small college, professors fret mostly about the use of AI to cheat. But students have much more profound anxieties about the technology. In my philosophy seminar they recently wished aloud that AI didn't exist. They worry it will rob them of opportunities when they graduate, upend the economy, and throw our society into (further) turmoil.

These are real problems. But beneath this is an underlying fear of being personally replaced, not just as workers, but also as writers, creators, even friends. AI can already write, be a companion, compose music, and make art, so what's the point of learning to do these things? What's the point of being human? What are we for? This kind of worry has a long history in philosophy. Two philosophers in particular show us why the worry is misplaced, and how simulating a human encounter differs from the real thing.

In 1642 the 19-year-old son of a tax collector decided to help out his dad, who was facing a mountain of figures. He did this by inventing an arithmetic machine, a marvel of engineering later described as the first calculator. This invention, his sister wrote, "transformed a science that resides entirely in the mind into a machine." But this helped his dad, it didn't replace him. The machine was the object of the user's will, a mere tool. The young inventor was Blaise Pascal, the 17th-century genius who would go on to make enormous contributions to mathematics, science, and philosophy.

Machines, Pascal later mused, lack heart, which he took to be the foundation of human life. The heart is the feeling part of us that desires, fears, loves, and intuits things that are otherwise invisible. He thought of humans as hearts with some thinking appendages, so we were not diminished by artificially augmenting the appendage.

That may all seem quaint now. Today's AI does a lot more than calculate. Besides creating art and comforting us, it can charm, manipulate, and guide our will; perhaps it makes decisions for us. There are many reports of love affairs with AI. So much for our hearts! Was the inventor of the thinking machine wrong not to worry?

The case of music is especially telling. Friends send me links to AI music. Isn't that amazing? It is. But once I'm over the novelty, I'm bored. If you fool me by telling me only halfway through a piece that it was made by AI, I'm impressed. But now I hear it in a new, flat key, and finish the track only to be polite. This isn't out of some principle: I just don't find it that compelling or moving and can't seem to get into it because I know there's no one there. There's in fact no heart expressed in these sounds. But why does this matter to music?

About a century ago, in his magnificent work I and Thou (written shortly after the mechanized horrors of World War I), the philosopher Martin Buber drew a distinction between two ways to relate to beings in the world: I-Thou and I-It. When you describe, assess, or explain something or someone in your environment, you approach it or them as an it, a thing. When you connect, relate, encounter another person, you meet them as a Thou. These are entirely different modes of relating. Communication, expression, interaction, and responsibility happen only in the I-Thou mode. Without a Thou, there is no complete "I."

Listening to a person's music can be a connection, an I-Thou encounter, because the music is an expression of that other. You may know nothing about the musician; you may not even know their name. But listening to their music is in fact a connection and is experienced as such. Even if AI music is auditorily indistinguishable from human music, in fact it is not the same thing: One is a genuine connection to another human being, the other is not.

Currently, AI at best produces notes as if it were a person without being one. Perhaps someday soon real people will use AI as their instrument, artfully crafting prompts for music or literature, and this will be perceived by others as the prompt-writer's expression (like sampled music as expression in electronic music today). This would merely replace our instruments, not us. And for now, just as the calculator's output doesn't express the user's calculations, AI-produced music doesn't express the prompt-writer's emotion musically.

Why does it matter whether, in fact, the music connects me with another, if I can't tell the difference? Everything in our lives says it matters. Think about when you say, "I'm sorry," and don't mean it, or when you interpret a person's accidental wave as a "hello." These are counterfeit encounters which feel like connections but aren't. A work created by a person enables an encounter with that person, but the product of an algorithm cannot do this. Sometimes we think we are engaged with a Thou, but we are in fact alone.

None of this is to say that you can't be moved by AI-generated music or art. Surely you can, and quite possibly you already have been. The point is that this is not the same thing as art created by a person, even if it can feel the same. It is not a genuine I-Thou encounter. It is not the craft of the hand-drawn line or a sentence written from me to you.

Art, music, writing, and intimacy do not simply reduce to the notes or blotches of color, the sounds or sensations. The connection they embody matters. And, likewise, we are not reducible to the products we make or the impacts we have. We are not merely producers any more than we are merely calculators. We are uniquely capable of being an I and relating to a Thou.

Pascal worried not that we would be replaced but that our hearts are in the wrong place, that we fail to care about the right things. Caring only for the sensation and not the personal connection, mistaking the person for the product, would be a failure, one that AI exposes but does not itself create.