AI and the Christian Life

It is wonderful to be together from so many churches on a Saturday morning for kind of a challenging topic. This is fun. One of the things I love when I speak is questions, because that's when you hear how people are thinking, and it's very helpful. So thank you for being here, and thank you for inviting me. I am delighted to be here.

I am Doug Axe. I'm a professor of biology at Biola University. I have been there for seven years. My wife and I met down here in Southern California, though I grew up in the Bay Area. We spent 14 years in England and then 15 years in Seattle, where Gordon is now. Go Hawks! A controversial statement, okay.

I am loving being at Biola. One of the things that the Lord called me to do, after calling me to write books and publish papers, is just to pour into people more intentionally. Being a professor has been an opportunity to do that, so I've absolutely loved it.

The Strange Behavior of AI

Kind of pivoting from what AI is, what's under the hood when you look at it, and demystifying it, we're going to focus now on something that's more personal, connects to us, and is maybe more spiritual as well: man and machine. We're going to be talking about personhood. I'm going to give you the bullet points of what I'm going to cover, and then we'll just start covering it.

The first thing I want to look at is this: if you work with AIs to any extent, you should notice that if there is a person behind this, it is a problematic, strange person. The persons that we know do not act the way AIs act. As Gordon said, basically, a large language model is a "what word should come next" algorithm. It's not thinking at all. It has just been trained to turn a crank, and if you give it words, it's going to spit out the next words according to probabilistic structure.

You've known of this for a long time because of your phone. When you type text messages, you're typing along and it offers your next word. Sometimes, as I've noticed, it feels like there's a bias, or sometimes it offers you a stupid word. I recently typed in, "It was a wonderful worship night. I felt the presence of the Holy Sp..." and it gave me "spit." I'm thinking, "Dude, that does not belong there." You are avoiding the word that you know goes there.

Humans don't operate that way. When we're speaking, we're not just outputting word after word in a chain. We have an idea, and then we are cashing it out in words. Most humans, anyway. Sometimes I'll start a sentence, and I don't even know where it's going; I just hope I find it along the way. That sounds like Michael Scott from The Office, who is an anomaly in that regard. But that is literally what a large language model is doing. Sometimes it starts, has no idea where it's going, and just hopes to find it along the way. And sometimes it does find it.

We're going to look at the source of human personhood, and then I'm going to give a simple description of the human person. What is a person? We think so naturally about personhood because we are persons, but sometimes we need to pause and say, "Well, what exactly is a person? What distinguishes a person from something that isn't a person?" Then I'm going to look briefly at the potency of personhood, and then the fake danger: AI becoming like us. You'll hear people saying that's going to happen, that they're going to surpass us. I'm going to prove to you that that can't happen. I'm not saying it's at least 20 years off; I'm saying never. It won't happen, and I'm going to show you why it will never happen.

Then we will look at the real danger, which is what I'll call "personhood atrophy"—the fact that AIs are capable enough in certain things that we, who are persons, are relinquishing aspects of our personhood because we've built bad habits by using AIs. That is a real danger, and that is happening right now.

Something is very wrong with the AI "person" compared to the human persons you know. Human persons come in varying degrees of intellect. There are people who don't know a whole lot, people who aren't very conversant, people who are brilliant, people who are wise, and people who are less wise. But there's a consistency to that. If you know someone who is very learned, scholarly, and wise, you learn to trust that person. They will answer things in dialogue in a certain way, and you learn to have confidence in it because that's how persons behave.

An AI is not like that at all. An AI will say something that sounds like wisdom one moment, and then the next moment they're being a total idiot. You can find humans anywhere along that spectrum, but what you won't find is a human blinking back and forth between those two extremes, and that's what you get with an AI.

Here's an example. You've noticed that when you do a Google search, typically you'll get an AI result first. My wife was wondering about what the role of a particular person at Biola is, so she did a search. The AI gave this answer: "William Smith is the president of Biola University, a private evangelical Christian university in La Mirada, California. He has been in this role since 2020. Smith previously served as a university senior vice president and provost before becoming president. Smith's background is in Christian education and leadership. Before joining Biola, he worked at other universities, including Western Evangelical Seminary. His leadership at Biola has focused on strengthening the university's commitment to its Christian mission and academic excellence."

That seems like a lot of knowledge. It's a more full answer than you would get if you asked many humans. The only problem is it's completely wrong. There is not a word of truth to this. It's a complete fabrication. This is what we call an extensive hallucination.

Now, if you asked a human who this person was and they didn't know, they might say, "Well, I don't know." That's what you expect. This is a key difference between a large language model and a human: a human has self-awareness. If I'm not competent to answer a question, I usually know that I'm not competent, and I'll tell you, "Well, I think it's this, but I'm not sure." Humans are good at qualifying their knowledge. An AI, because it doesn't have actual knowledge—it's just a trained machine—is notoriously bad at self-assessment.

One person writing in The New York Times Magazine put it this way: "An AI is like a very smart assistant who has a dozen PhDs, but is also high on ketamine like 30% of the time." You can't really rely on someone like that.

I'm going to show you another example now. It's fun to look at the large language models, but the image generation AI is great for talks because you can see the best and the worst of AI. I have a colleague named Harvey. I took his picture straight from the Biola website. He's a great friend and colleague of mine. We tease each other, and one of the things I've teased him about is that because he wears a brown shirt with brown pants, he looks dressed up like the UPS man. He was making brisket the other day, just a few weeks ago. So I took this photo of him from the website, gave it to Gemini, and wrote this prompt: "Please make a photo of the man in this picture wearing a UPS uniform and smoking brisket."

And there you go. Five years ago, this would have been impossible. This looks like a real photo. You can't tell the difference between this and an actual photo, yet it is totally fake. It just took his face and generated the rest, even putting a UPS truck in the background. If someone had tried to do this with Photoshop, it would take days for an expert to do something that looks this good. Now you get it in a minute and a half. It's amazing.

Then I took it up a notch: "For a second version, keeping his face the same, can you make him absolutely ripped, make his six-pack visible underneath his UPS shirt?" It complied, and you get this ripped version of Harvey. Harvey liked this so much that this has become the photo for our group chat thread.

But yesterday, in preparation for today, I gave Gemini this prompt: "Please give me a photo of the church building pictured above, removing the text from the photo and placing farm animals on the lawn. The church building should look exactly as shown." My intent was to show that it would nail this. I thought it would do it really well, like it did with Harvey, and then I would show some things that it doesn't do well at all.

Instead, it came back with this: "I have created a new photo of the church for you, ensuring that the text is removed while keeping the building exactly as it was. I've also added a selection of farm animals, including sheep, goats, and a cow grazing on the lawn to complete the scene you described."

But if you look at the image, the church building is modified, and instead of a cow, it is like a whole herd of cattle. You get something that's totally wacky. This is what I mean by lack of self-awareness. A human artist might say, "Well, I didn't have any stock photos that resemble that, so I've modified the church." It is one thing for a human to give you something that's not what you asked for, but they know it's not what you asked for. The AI doesn't have that kind of knowledge. It answers as though it did give you what you asked for, but it didn't at all.

Here's another example from yesterday: "Please make a picture of seven snails moving in a line with the second, fifth, and sixth ones wearing little hats." It gives this adorable photo. It's extremely powerful that you can get things like this, but in detail, it's completely wrong. It's not seven snails; it's eight. And the wrong ones have hats.

It does some things very well, and some things very poorly. In humans, if you can do advanced things, it presupposes you can do the elementary things. But AI will flub up the elementary things while doing the advanced things well. It is very unlike a human. This peculiar amalgam of mastery and confused self-delusion is very unlike the work of any human. If you work with AIs, the one thing you'll see is it's not humanlike. I'm not saying it's not helpful; at its best, it's very helpful. I'm saying you have to watch it in a way that you wouldn't have to watch a human.

The Source of Human Personhood

As we go along, we will unpack the reality that AI is profoundly not human. What are the deep distinctions between an AI or any algorithm and a human being? First of all, what is the source of human personhood? How did we become persons? If we know that, then we can ask: is Gemini or ChatGPT a person? Well, we'd have to go back to the source. How did we get our personhood?

The big reveal here is found in the book of Genesis:

Then God said, “Let us make man in our image, according to our likeness. They will rule the fish of the sea, the birds of the sky, the livestock, the whole earth, and the creatures that crawl on the earth.” So God created man in his own image; he created him in the image of God; he created them male and female.

— Genesis 1:26-27 (CSB)

That's personhood. We are made in the image of God. God in three persons—a Godhead who knows personhood because for eternity He experiences relationship as Father, Son, and Holy Spirit—creates us to have something of that in us. We are absolutely unique among the living things on planet earth.

Evolutionary biologists will say that the chimpanzee is our closest living relative. I reject that, but that's what they say. I just say: go look at chimpanzees and see if you get personhood. You don't. They are wonderful creatures, and dogs are wonderful creatures. I love animals, and they are conscious and have feelings. But we need to unpack what it is about us that is categorically different from a dog, a chimpanzee, a dolphin, or any of these other wonderfully created creatures.

What is a Human Person?

By God's design, a human being is an eternal and immaterial mind, soul, and spirit knit to a physical body. Theologians and philosophers will unpack the mind, soul, and spirit distinctions slightly differently, and I don't want to labor us with that. The point is we have a physical body that's made of atoms on the periodic table, but we're not just that. We have this immaterial component to us that is beautifully knit to our physical bodies so that they are made to work together.

Our body is made to work with a mind, soul, and spirit, and our mind, soul, and spirit are made to work with a body. We were not made to be disembodied spirits; this is by God's design. It's clear in Scripture that we are not just physical things.

A lot of sciences went in the direction of materialism starting back in the 17th century, and by the time of Darwin's book in 1859, there was a full-blooded materialistic view of everything, including humanity, taking hold of the sciences. Scripture clearly rejects that view of humanity. Listen to the words of Jesus:

Don’t fear those who kill the body but are not able to kill the soul; rather, fear him who is able to destroy both soul and body in hell.

— Matthew 10:28 (CSB)

This is a clear statement that the stuff we can touch through physical action is not the soul. The soul is not accessible through physical action; it is not a physical thing.

Let's look for a minute at the potency of personhood. What is it possible for us to do because we are persons that would not be possible if we were not? What is it that we can do that a dog or a chimp cannot do? Many of us don't spend much time with chimps, but raise your hand if you have a dog. Dogs are great animals. I love my dog, Sherlock. He's a wonderful dog, and I have a relationship with my dog. But I relate, being a person made in God's image, in a very different way than my dog relates.

Like animals, we humans are made to be embodied. We were not made to be disembodied mind, soul, and spirit. I affirm that animals are conscious and they do have an immaterial component. My dog is conscious, he does have a mind, but it's not like our mind because, unlike animals, we are made to be spiritual and relational. A dog is not a spiritual being. He has a consciousness, but he does not have a spiritual nature.

We can think of anger on different planes. Can a dog be angry? Yes, a dog can be angry. Can a human be angry? Yes. But there's a different thing happening. A dog is angry on a different plane than human anger, and the dog doesn't even have access to the plane that describes human anger.

On the carnal plane, where a dog exists, a dog can be agitated and aggressive. It is bothered by something and it acts out, experiencing something in a carnal way. Whereas a human experiences things in a spiritual way. A human can be offended and indignant. There's no such thing as an indignant dog. You can have an angry dog, but it means something different than an angry human. It doesn't have access to this higher plane of spiritual being that we do because we're made in God's image.

There are categories of human activities that don't exist for a chimp or a dog, and they never will. We are contemplative, emotional, and relational on a spiritual plane. We engage in judgment, cognition, insight, and inspiration. We experience joy, grief, and anger on a spiritual level. And we are made to be relational. Dogs have "friends" with other dogs, cats, and humans, but there is a spiritual dimension of relationality in a human because we are made in God's image. We are made for deep communication, empathy, and worship. These are not things shared by any other created kind; they're unique to humans.

Because they're unique to humans, a dog is actually participating in life as a conscious thing with a body, and in that respect, a dog is above an AI. An AI is nothing but a machine that spits out stuff that may or may not be useful. It's not even like a dog in that it's not aware of anything. It is a machine.

The Fake Danger of AI Surpassing Us

I want to talk here about the fake danger. This is something you'll hear being hyped. It is interesting that the people who are hyping that AI might take over and surpass us are the people who have a trillion dollars on the line. They know if they make a statement like Elon Musk saying, "I'm afraid of what AI is going to do," then you'll see the tech stocks rise. People who don't have the biblical grounding to distinguish these things think, "Well, if they're really going to take over, I want to put my money in that." But it's an odd thing, because if they did take over, that would not be a good thing.

Regardless, they will not become like us. The claim that AIs can become humanlike is a false claim. Rather than just asserting that based on what we saw in Scripture—that there's no way an AI can become us because we made AI, whereas God made us in His image, and we don't have the ability to make something in God's image—I want to prove to you that we are not merely physical things.

This will get a little philosophical in how I'm going to ask you to think about this, but you can actually show that we are not just physical things. I'm going to use a method called self-defeating propositions to show this. Specifically, I'm going to show you that our brains, which are material, cannot think. Neurons are made of carbon, nitrogen, oxygen, phosphorus—all things on the periodic table. They cannot think. When I say that to people, they go, "What do you mean your brain can't think?" I'm saying your brain can't think, and I'm going to show you how we know that's true.

First, let me introduce this principle called "the fallacy of self-validation." It's a simple idea: if the validity of X is in question, then X cannot provide the answer. This makes complete sense. If you're hiring someone named John, but you have some questions about whether John is honest, who can you not ask? You can't ask John. "John, are you honest?" "Oh, yes." That doesn't help you. If the validity of X is in question, X cannot provide the answer.

This has some interesting implications. One of the things we take for granted is that we can think, and that thinking is a valid thing. But if this principle is true, what can you not appeal to in order to show that thinking is valid? Thinking. But if we can't use thinking to show that thinking is valid, then we can't prove that thinking is valid.

Human thought is an enterprise that we take on faith to be valid. We're not saying that all human thought is true or that humans always arrive at the truth. We're asking here: is it a complete waste of time for humans to think at all? Is there any hope of getting to something true by thinking? We hope that that's true, so we take on faith that it's valid for humans to think. It's a faith position.

Once we take that to be true on faith, we can do something called science. By the way, scientists are rarely trained on this. Some scientists say, "If it's not shown by science, it's not true." They have never learned how you think about this. Science is another thing like this: for us to do science, it has to be true that we can think, but it also has to be true that this apparent external world actually has objective existence independent of us. We have to trust that there is an objective reality out there with consistency and objectivity so that we can study how it works. We can describe equations, make observations, and they marry up. That's what science is.

But guess what? If we ask, "Is the whole enterprise valid?" you can't do science to validate science. You just have to trust that it's valid. Remember the movie The Matrix, where your entire experience is fabricated. If that were true, you couldn't do science. We're trusting on faith that that's not the case, that there's something reliable out there. As Christians, we believe God made this world, He made us, and He put us here to do things like science, and that's why we believe this is valid. But it is a faith position.

Once you've taken these things to be true on faith, scientists go about trying to come up with explanations in the natural world. Suppose one of those scientific conclusions becomes so standard and widely accepted that it becomes part of textbook orthodoxy. But suppose that explanation, if you take it to be true, has implications that undercut our starting point—the faith position that human thought is valid. If that scientific conclusion undercuts thought itself, then science goes, and the textbook goes. That would be a self-defeating proposition: a proposition that, if taken to be true, undercuts the entire program, and therefore cannot be true. I'm going to show you that "brains think" is exactly this kind of self-defeating proposition.

Imagine yourself in a futuristic brain imaging laboratory that has equipment that doesn't exist today. You are fully conscious in a seat. There are scientists around you who all think that this is a purely material world, that there is nothing but physics. They believe that thought is a physical process that is nothing more than the operation of the neurons in your brain, and they are imaging in real time everything that's happening in your head down to the atomic resolution.

They ask you, "Count to ten, and think about the numbers." You say, "Okay, one, two..." and they say, "Stop right there! When you said 'two,' we captured this image of activation in your frontal lobe. Is that what you mean when you say 'two'?" And you say, "No. I'm glad my frontal lobe is doing what it's doing, but that's not what I mean when I say 'two'." They say, "Well, wait a minute. Let's zero in here. This particular synapse between two neurons fired when you said 'two'. Is that what you mean when you say 'two'?" And you go, "I'm not denying the synapse fired, but that's not what I mean when I say 'two'."

They start to get upset with you. Why? Because in a material world, if there is nothing but atoms and molecules, then if "two" is anything, it has to be made of atoms, because there is nothing but physical stuff. And you're thinking, "Well, no, I have an idea that is 'two'. There's a concept that is 'two', and it isn't on the periodic table. There isn't any picture you can show me of anything happening in my brain that I will say, 'Yes, that is what I mean by two', because it isn't a physical thing."

And yet, you are thinking about it. The materialists get upset and say you must be confused because this is a material world and there isn't anything that isn't physical. If you follow the materialists by doubting your conscious experience, you are forced to doubt absolutely everything. Why? Because you don't have access to anything except through your conscious experience. If the materialist view forces you to accept that you are fundamentally confused about your own consciousness, then you are deeply confused indeed, and thought is delusional.

You are left with these two options:

• A: I am manipulating concepts, which are not physical; they live in a different world when I think.
• B: I am being manipulated by neurons into getting the impression that I'm manipulating concepts, but I'm not.

The materialist scientists are proposing B, and you are saying, "No, it has to be A." But do you see what happens if you go down the route of B? "Manipulating" is a positive thing in A; you are working with concepts and fashioning them. "Manipulated" is a negative thing in B; you are being duped, tricked, and deceived. If you are being tricked, then you lose the whole thing. Human thought is not valid if we are fundamentally deceived. And that means science goes, brains go, neurons go, and you don't exist. All of this is just a load of delusion, including the claim that brains think.

So, you have to affirm A over B because B is self-contradictory. It's incoherent. Yes, concepts exist. No, they're not physical. Yes, I am working with them. Yes, my brain is physical, which means my brain is not actually working with concepts. My brain is an organ that operates in service of my mind, and my mind does work with concepts.

We are left with a view of humans that looks like this: there is a material, physical realm. We do have these marvelously designed bodies that are physical, and all of biology, human anatomy, physiology, and biochemistry studies this physical aspect of the human body. But we are not just bodies. We have an immaterial mind, soul, and spirit that is in a different realm. And yet they're connected somehow.

I come up with the idea of raising my hand—that's something I thought—and my physical hand goes up. It only does that because Someone knows my thoughts before I think them, is in charge of the physical world (including my brain), and when I think this, He knows I'm thinking it and presses that appropriately into the motor cortex. The brain activates, it pushes signals down the spinal cord, my muscles react, and my hand goes up.

Going the other way, I see the light from outside. Photons are impinging on my retina. This is all physical stuff. The rods and cones are being activated by light, producing a signal that goes through the optic nerve to the visual processing center of the brain. All of this is physics, including the processing that the brain does. But "red," "white," or "green" does not appear as a physical thing in the brain. Where is "red" in physics? It doesn't exist. Red exists in the mind. Someone is operating all of this physical stuff, and at the appropriate moment, when your brain has done its physical calculations, it is translated into a mental, conscious experience for you.

We are marvelous creatures. The fact that this is who we are means there is no way that a machine, which doesn't participate at all in that upper sphere, is going to become like us. They can churn out physical results that might look like a human did them because they've been trained on human activity. But they will never become us because they are fundamentally operating in a different way.

The Real Danger of Personhood Atrophy

That leads to the real danger. The real danger is that when machines become highly capable—and without doubt, despite the goofs and the epic fails, AI is very capable, and we will be seeing more of that in the coming years—humans, who are persons, will allow themselves to become dependent on machines in a way that causes us to shrink back from who we are and what we were made to be.

You hear stories of this: people who are desperate for relationship, who find it hard to relate with other humans (which we all experience at times), find that an AI is easier to relate to than a human. They develop a pseudo-relationship with an AI rather than with a human. That's an extreme case, but it's going to become more and more common and tempting. Less extreme versions involve being lazy about exercising what we are as persons because we can get an AI to fill in.

If we look at a diagram of what a human is, even a trivial example of communication between two humans looks like this: a human on the left has an idea, "Oh, I want to ask her if she did the statistics homework." It's a simple, everyday idea that lives in the mind. It gets pushed into the brain and results in action, either typing a text or speaking. That physical stuff conveys the message to another human who receives it, processes it in the brain, and then it becomes a conscious thought and is interpreted.

This is how humans converse and relate. It is two mind-soul-spirits connected through the physical structure of the universe that God designed for us to live and move and have our being. Even if this is not a philosophically deep conversation, what's happening here is profound because two spirits are connecting.

Compare that to a chatbot. The chatbot isn't a person. It doesn't have access to this spirit-concept realm. You can ask a question of the chatbot, the machine cranks through, and something comes out from the machine. But nothing ever understood you. Nothing ever connected with you. There was no meeting of the minds, because there is no mind on the other side of that.

I will stop there, and I will take some questions.

Questions and Answers

Question: You know that picture of the snails with the hats? Was the snail picture already existing somewhere on the internet and then just modified with the hat?

Doug Axe: No, it's actually more remarkable than that. I've spent quite a bit of time trying to understand large language models. If you go back to 2022 or early 2023, the images you were getting from AIs were horrible-looking, crude things, but now they look like a photograph.

It is not the case that the exact image existed. There are millions of photographs of forest floors, lots of pictures of snails, and lots of pictures of hats. The clever thing is these AIs are now pretty good at pulling together the hat pictures, the forest picture, and the snail picture to create combinations that nobody has ever actually photographed. So it isn't just giving you something that exists the way a Google Images search used to do.

But it is also flubbing up something very rudimentary and elementary: it can't even get the right number of snails. I give assignments to my students where they're required to go off to an AI, get it to fail, and turn in the failure. We spend a whole day going over these. Image generation is very bad at getting numbers right and counting. One of the prompts a student turned in was, "Give me a picture of a fat guy eating six hot dogs at once." It gave two attempts. It looked like a person, but there were never six hot dogs, and one of the persons had three arms. It can't get numbers right.

So it's doing something clever in that it's putting things together that there is no exact example of in the world of photography, but it's also strangely messing up relatively simple things.

I actually had a long conversation with an AI when I said, "Give me a simple drawing of eight snails in a row wearing hats." It did sort of a Richard Scarry style figure, but again, I said seven and it gave me eight, or I said eight and it gave me nine. I asked, "How many snails are there in the picture?" It said, "Eight." I said, "Are you sure?" It said, "Yes, I counted them." Then I went through this long process of, "How do you know there are eight? What would you conclude if I tell you you're wrong?" It said, "Well, it must be that it's an ambiguous image."

Then I asked, "Would you like to analyze the picture you gave me?" "Yes, please do." It drew boxes around them. It had actually drawn the snails in two lines, and it drew a box around the top line and the bottom line and said, "I was wrong. There are two snails." It was absolutely hopeless.

I am of the opinion that if you can't actually think, you're severely handicapped when it comes to thought. When we're trying to get machines to act like they're thinking when they can't, we are forever going to be stuck with these things where they just can't do it. Some things will get patched up. When the big companies become aware it's bad at something—like counting letters in a word—they correct that by saying, "If someone asks this, we'll just run this specific code to do it right." So it's no longer the large language model doing it. You can do a lot of patching like that, but you'll be forever patching, and there will always be holes.

Question: Would you consider transhumanism to fall under the false danger of AI becoming like us, or the real danger of personhood atrophy?

Doug Axe: I think the first is not a danger because humans will not be like AI, and AI will not be like us. In transhumanism, you have this unhealthy desire to make sort of a combination of human and machine. I think there is no danger of machines ever actually being like us.

If you think about this, it's actually blasphemous, because the creator is always over the created. We will never be God. Why? Because God created us. We created AI, so AI will never be us. To say that AI will be us or better is like saying we can become better than God. It's just not going to happen; it's fundamentally impossible.

Question: You mentioned extremes, like people having relationships with AI. Could there be a possibility that AI could be involved in discipleship?

Doug Axe: If someone can want to marry an AI, then someone could see themselves as being discipled by an AI. But we could also ask: can a book disciple you? We have books written on discipleship that have a huge amount of wisdom. But if by "discipleship" we mean a connection between two mind-soul-spirits who love the Lord and one is trying to pour into the other, then no, AI will never do that. If by discipleship we mean someone getting the feeling that they're being discipled, you could get that from a book, a podcast, YouTube, or an AI, but it won't be the real thing.

Question: As a landscape architect, I was trained in architectural design and construction engineering. I want to know: can AI understand symbolism—the symbolism behind colors, shapes, forms, and ideology?

Doug Axe: A symbol, properly understood, is the connection between a physical token and something conceptual that it refers to. Language is symbolism. We use physical things, like putting ink on paper, but it is a word. You read the word, and then you go up to this concept world and know what is meant by it.

In that sense, AIs are not capable of symbolic manipulation. They are capable of crunching tokens. Anything you can do with computers can be done, but a computer is not accessing an actual concept. It's not making a conceptual connection.

What's happening with an AI is that when you have a conversation with ChatGPT, you're thinking something, you type in a prompt, it crunches through the machine, and it comes back as physical text on your laptop. It doesn't become an idea until you read that text and say, "Oh, in English, that means this." You came up with the idea, and you attribute it to the machine, which didn't even have the idea. It's not capable of having an idea; it's capable of stringing words together. But when those words come back to you, you go, "Oh, it means this." You have an idea that you generated because of your understanding of language. The bot didn't actually have that idea.

Question: From what I've heard in your lecture, it sounds like it's a mistake to call it "artificial intelligence." Is there a different name or title you would give it?

Doug Axe: You know, that's an interesting thing because there's spin on this. I'm happy with "artificial intelligence" if the emphasis is on artificial—meaning not real. But you'll also hear it called "machine intelligence."

I think a lot of people working in AI think of that term as meaning, "This is actual intelligence that was made by humans, as opposed to human intelligence." The spin is that this is every bit as much intelligence as our intelligence is, it's just that we made it—kind of like an artificial knee is a replacement for a natural knee. So they would say natural intelligence is what humans have, and artificial intelligence is not inferior, it's just that we made it.

I say that's a wrong way to think about it. If you say "artificial" means it's not real—like artificial vanilla is not real vanilla—then I'm happy with the term.

Thank you very much.