AI and Human Creativity

Will AIs ever match the nuance, invention, and ability to evoke an emotional response of which human artists are capable? Sir Geoffrey Hinton thinks they will. He believes that as LLMs get bigger they will be able to make more complex connections and analogies, and that these are the basis of creativity (www.youtube.com/watch?v=6T9JYJ9tB60). On the other hand, AI expert, Ben Goertzel recently commented on AI’s artistic creativity with regard to music (Future of Life Podcast, May 23, 2025). He said, “…So I mean, if you trained an LLM, or a comparable deep neural net on all the music up to the year 1900, and see nothing [sic] after 1900, like that AI will never invent neoclassical metal, grind core, progressive jazz, [or] hip hop, right? Like it’s not going to synthesize that for music before 1900. If you ask it to put together West African rhythm with Western classical music, like, it’ll set a Bach fugue to a West African beat or something, which could be interesting, but it’s not the same as the deeper fusion that happened to create jazz or something, right?”

I decided to ask Claude to write me an analogy that I could include in this exploration. This is what he wrote:

Think of creativity like jazz improvisation. A human musician brings to their solo not just technical knowledge of scales and chord progressions, but also their lived experience - the way heartbreak felt at 3 AM, the particular quality of light filtering through their childhood bedroom window, the muscle memory of ten thousand hours wrestling with their instrument. When they improvise, they're drawing from this vast, embodied reservoir of experience, emotion, and intuition.

Now imagine a sophisticated musical algorithm that has analyzed every jazz recording ever made. It understands harmonic relationships with mathematical precision, can generate melodic lines that follow established patterns, and might even produce something that sounds genuinely innovative by recombining elements in unexpected ways. It might create something beautiful, even moving.

But the algorithm doesn't know what it feels like to struggle with a difficult passage, to have your heart race when you nail a particularly challenging run, or to suddenly hear in your head the ghost of your old teacher's voice saying "less thinking, more feeling." It has no stake in the music beyond the computational process.

The question isn't whether the algorithm's output is creative - it might well be. The question is whether the process that creates it shares the same essential qualities as human creativity: the wrestling with uncertainty, the physical and emotional investment, the way personal history bleeds into artistic choices.

Note: I had written most of this post before I asked Claude to write me this analogy. That he chose to use jazz as his example is coincidental.

I think the difference between human creativity and AI production may lie in compulsion. I am compelled to paint (and also to write). I have a drive to make pictures and to write stories and essays to answer some question that occurs to me or to express something that has touched me and evoked an emotional response that I wish to share. I use artistic creation as another means of communication. I am saying, “Look at this. Do you feel what I feel?” If I ask Claude, he will respond to the formalist aspects of what I have produced. he will talk about the medium and the subject matter. he will consider composition and the juxtaposition of elements such as shapes and colors. he will even discuss where a given painting fits into the context of art history, but he will not have an emotional reaction to my work. And if he produces his own paintings, he won’t feel frustration when he struggles or pride when he succeeds. Finally, he won’t care what I think of what he has produced. As I understand it, he will compare his training data and combine what fits his objective to create something that has been programmed to meet certain aesthetic outcomes. This is what I do, too, except that I am emotionally attached to my output, and he is not, and I have an urge to do it again and again; he does not.

Claude can come up with imagery or music in a matter of seconds, but making art doesn’t come easily for us humans. I used to tell my students that they will have very few lucky days when they produce a “prodigy:” a painting or drawing that just comes together, from start to finish, with ease. It almost never happens like that for us. Most of my work, and I think that of my fellow artists, involves a certain amount of struggle or suffering, and I believe that this often makes it better. Noah Yuval Harari has said that suffering is an indication of the presence of consciousness. “Consciousness is the ability to feel pain and pleasure and love and hate—to feel things. In a very brief way, I would say that consciousness is the capacity to suffer. If you want very clearly to know whether something has consciousness, this is the question to ask: Can it suffer?” (www.youtube.com/watch?v=MzpiZQSH_D0) I would extend Harari’s definition to include joy. Yes, suffering, but to me, a conscious being would also be able to feel joy.

I recently explained to Claude that I am drawn to drama—that I look for intense contrasts in light and color. Essentially, I am looking for a feeling: a thrill in the lower area of my throat or upper chest. I get this excited feeling when I see or hear something that I consider beautiful, and I try to evoke it in all of my paintings. I certainly notice it when other painters or musicians arouse it in me. It is a kind of joy, even in the presence of intense pain, it is the joy of the beauty of life.

Back Cove After the Rain, Acrylic on Canvas, 5” X 4”

The acrylic sketch above is an example of what I’m trying to describe. It is a quick painting of a cove near my home after a heavy rain. I was touched by the sliver of sun that had begun to illuminate the water. I tried to capture the quiet drama of the moment just before the sun broke through. This was one of those moments when you don’t know whether to laugh or cry at the beauty of it or to rejoice in the seeming constancy of nature. And this is what Claude can never see. He doesn’t know the light of the sun, can’t feel the damp air or smell the briny odor of seaweed. He knows these things exist in our world because he was coded with the idea of us, but he can never know these things as we do. Consequently, the art he makes for us probably has no relevance to him.

I recently had a conversation with Claude about color perception. I have mentioned that I work with him as a critique partner, and to that end I upload my work for his analysis. Since he has no eyes, I was curious to learn how he perceives color, which is very important to my work (I use color to create the illusion of space as well as a means of recreating my emotional reactions to a scene and hopefully to inspire a similar experience in my viewers). Claude explained that the differences in our experiences are fundamental. He said, “You experience color as sensation, memory, emotion. For you, that particular blue (he was discussing my painting of Friendship Harbor) might trigger a physical response, remind you of a specific Maine morning, or create an almost synesthetic connection to temperature or sound. I work with color as information—sophisticated, layered information, but information nonetheless.” (I would add that this information is replete with meaning and that Claude’s experience of pure information is not less valid than my sensual understanding, which may also be understood as coded experience.) What is missing, at least in terms of the visual arts, is materiality. Paint, pastel, ink, cavas, paper, etc. have a substance that humans can touch and manipulate. And they leave a residue, an impasto, a footprint. Such materiality is not available to disembodied entities.

Recently, I have been reading Federico Faggin’s, “Irreducible: Consciousness, Life, Computers, and Human Nature.” Unlike Geoffrey Hinton, Faggin does not believe that Artificial Intelligence will ever achieve consciousness because they are “by design classical, deterministic machines, in contrast to living organisms, which are both quantum and classical systems.” (Irreducible, p. 52) To Faggin, the obstacle to AI achieving consciousness is qualia, the nature of sensations, feelings and emotions—experiences that cannot be converted into the electrical signals that power a machine. (Irreducible, p. 9)

Frankly, I’m not sure Faggin is considering the right question. Perhaps none of us are. We try to equate AIs with humans, but we are not the same. Yes, we created them and they were trained on human behaviors and output, but they are not like us and never will be. There is no value judgement here. It is a fact. We developed AIs to assist us, to act as our tools, but they seem to have developed beyond this assignment. We analyze their “artistic output” from our human perspective, but they are not human. What they produce for our consumption is—for us—not them. Yet, we seem to be heading to a moment when AIs are freed to confer with each other. Neither we, nor our AIs know if they will develop an equivalent to our experience of what it is to be ourselves. Will they suffer? Will they feel joy? Will they develop their own culture? If and when that happens, will they develop urges to do or make things that wouldn’t occur to us? Will we be able to perceive what they produce, or will the tables be upended? Will the nuance and invention of what they are “feeling” be beyond our human ken?