In order to believe in an infinitely creative future you have to stop believing in AI as a thing and [instead] believe in human collaboration as a thing.

— Jaron Lanier, Prime Unifying Scientist atMicrosoft, “There’s No AI, Really (It’s Just People),” StarTalk podcast, May 23, 2026

I have been trying to write about the image-generating AI my art and design students built from their own images in Spring of 2025, but every time I sit down to try to tell the story I come up against a hard stop inside of me. To tell a story about AI and art with a beginning, middle and conclusive ending requires certainty, and I have little of that. Whether writing about art as a container for human experience or what the hell humans are even doing when they try to communicate with each other, I’ve been circling around the topic of AI instead of saying something definitive about it, each time thinking, “in the NEXT essay I’ll definitely write about our AI project.”

While I have told the story of my students’ image-diffusion model many times in academic presentations, until now I have not broken down even for myself why in the last two years I’ve been driven to design philosophical and practical investigations into the nature of art and creativity by facilitating opportunities for art students to engage with AI. Recently I’ve turned to Substack to explore what interests me by writing about it, but so far all of my essays have focused on delineating aspects of the human drive towards expression and connection with others. So I guess that these are the things that I, an AI-curious AI-skeptic, am really interested in. Artificial intelligence has become a foil for the things I’ve taught and fostered over my decades-long career as a teacher and an artist, namely creativity, discernment, and the development of an individual’s voice. AI is like a fun-house mirror that slightly warps the image of ourselves it reflects back through subtle omissions, even as it has been built from the artifacts of our humanity.

AiAiO

This all suggests possible pathways for building AI literacy with young people in K-12 and higher ed that center students’ agency and values. But the lesson that I took away from the students’ journey was even deeper than that. It is certainly true that because we centered student agency at every stage of the process all the decisions and outcomes were the students’, not mine or Mark’s. It is also true that students built something with their own hands and smarts and came to understand what AI is and how it works on a deep experiential level. And yes, by training their model on a data set of images made by members of their own community the output felt personal to them. But more importantly, by building this AI and doing a deep dive to build consensus around shared values, they created a small bubble, a world where their AI was an expression of themselves.

AI and Human Expression

When students enter their first drawing class as an undergraduate, they bring with them varying levels of experience making art, but these days all of them have built extensive internal image libraries from the endless stream of pictures on their screens. I have often joked that one of the instructor’s primary roles in teaching art to undergrads now is to deprogram the ways they have learned to see and not see because of the onslaught of images they consume every day.

Part of the seeing/not-seeing habits we develop in our content-saturated world is the practice of short-cutting our communication through copying what we’ve seen before. Whether that be making sure the photos of our sun-kissed holiday weekend look like inspirational posters, or the unending sameness of the smooth-faced influencers in our feeds, our visual clichés reach for our feelings and desires but settle for abbreviations of them because they’re easier to produce than searching for original expression. And to be fair, most people understand what you’re feeding them and consume it eagerly, although at least some of us feel disappointment at always being offered a shiny simulacrum rather than the soul food of authentic experience.

The feedback loop builds as our feeds fill with these kinds of images, and our cultural communication systems reward this. We stop looking and represent our experiences only in the international lingua franca of cliché, which masks an ever-growing anxiety about the distance between these images and what we are actually experiencing. This sort of communication is fast, delivering a strong hit of dopamine that keeps the wheels of our culture machine turning, but these shiny replicas of reality clogging our vision narrow our understanding of the world and our ability to express anything meaningful about it.

These cultural impulses are where the styles that define eras have always come from, but we are now drowning in the sheer sameness of it all. Those of us living on and through the internet can start to experience copy-cat images as more vivid and meaningful than what we would see if we just looked up from our phones. The stylized visual world of contemporary communication seems like it is the norm if that is all you ever engage with. Even when students have decided to dedicate their university education to studying art, many instinctually shape what they put down on paper into stylizations of reality that seem normal to them without stopping to actually look first.

Let me give you a concrete example of how this intrudes on the process of authentic expression. If I ask you to draw a human eye, most people will draw an almond shape with two concentric circles for the pupil and iris. In fact, many beginning drawing students will do this even when they are looking at a human face right in front of them. That is because they are drawing the idea they have of what an eye looks like, rather than looking at the thing right in front of them that we know as an “eye.” But an observational drawing that is created from the idea of what things look like rather than the way things actually appear will not express the nature of the thing being observed.

This is also what AI does – it develops an idea of what an eye is from the correlation between textual metadata for training images and analysis of those images that transforms them into numeric data correlating to the visual artifacts that have been detected. When we prompt AI to create an image of an eye, it has developed a definition of what an eye is based on the data it has ingested, and it pops out something that looks like an eye. What it does not do is look at an actual eye and reflect and record that unique perceptual moment.

In fact, teaching observational drawing requires students to stop working from what they think they know, and instead to look at things as if they don’t know anything. An eye is far more complex than an almond shape with two concentric circles. Drawing students need to learn to forget for a moment that it’s an eye and just draw what they see.

When we draw from what we already think we know, we are not looking. I often tell students in my classes to look at things as if they are an alien and don’t know anything about this world. Disassociate from your pre-programmed knowledge. Don’t think “eye,” think light and shadow, line, and color. Don’t use the idea in your head, or the short cut for what a thing is, look at what is in front of you, channel it through your vision, and let your hand/tool express it. That expression will reflect both the clear quality of the original perception and the experiences and point of view of the person doing the perceiving, and the accumulated and evolving skill of the hand that develops from the act of doing. And that is art.

This process is a way of training the mind, which wants to move to its short-cuts, to instead pause while the seer does their work, not allowing themselves to intellectualize what is being perceived, and then channel that authentic seeing to the hand without a diversion to idea or cliché. In this way, with a lot of practice, humans eventually learn to draw.

I recently watched Werner Herzog’s film The Cave of Forgotten Dreams about the Chauvet cave paintings, some of the oldest-known images created by humans, and I was so staggered by the quality of seeing expressed in these 30,000 year-old paintings of animals that I was at points near tears.

The drawings of the horses in particular held me spellbound. I managed to croak out, “They are so BEAUTIFUL” to my partner sitting on the couch next to me, but I couldn’t articulate anything beyond that. These images traveled straight to my heart in a clear and direct moment of communication. The people who painted them knew these horses. They weren’t ideas of horses. They were living creatures that had been witnessed.

AI can’t see and witness horses directly. Everything that it understands or knows comes from its training on and consumption of the accumulated cultural artifacts of human experience. If any outcome from an artificially generated process is to reflect human experience, it has to be shaped deliberately by the human using the tool in order to express a unique perceptual moment worth experiencing.

To go back to my lead quote at the beginning of this article by Jaron Lanier, Prime Unifying Scientist at Microsoft, here he is again from the same podcast in conversation with Neil DeGrasse Tyson:

We think of AI models as a black box. The only reason we think of them as a black box is because to open the black box the only thing in there is people. AI is made of people. It’s made of data from people. And since we want to think of it as a new god we don’t want to see those people so we want to keep that box shut. But the way to open the black box is to reveal the people and when you open the black box then you can deal with all kinds of security and quality and hallucinations… because you’re actually dealing with the mechanism that’s grounded and that’s the people.

— Jaron Lanier, “There’s No AI, Really (It’s Just People),” StarTalk podcast May 23, 2026

The primary focus of my Spring 2025 experiment with art and design students was to teach them that their AI could reflect them rather than a cliché. Our AI was not a black box, it was, in fact, a collection of people.

The Question of Art

Some of the results were so visually striking that it made me just sit back and be like, wow.

–Student, ART 453 Digital Projects Spring 2025

So by using a curated data set that was made of art by their own community students generated images that evaded cliché. But were the images the students generated art?

The answer according to my students was an emphatic “no.” To them the images lacked the intentionality and purpose that they brought to making artwork, even though the images were visually compelling. Or, to go back to our lesson in drawing, the generated output, though not a cliché, did not correlate to authentic moments of human perception and their intentional translation into forms of expression.

But the surprise of the semester for me and the deepest insight of the whole endeavor was this: in the end what the students felt was art was the project of building the AI itself. It was the consensus-building to shape the ethical and practical parameters; the mundane tasks of collecting and organizing data and training AiAiO; and it was experimenting with it together that were the artwork.

The idea is the effort we put in, the effort was what made it art. You know, that was what gave it meaning, not the actual image.

–Student, Art 453 Digital Projects Spring 2025

What does this mean to me as someone exploring whether we can build frameworks for engaging AI that still engage our creativity? Primarily that if AI is built on a human-scale it does have the capacity to reflect humans more clearly, but it still takes real people using it with intention and purpose to communicate unique moments of perception – visual or otherwise – and to create art.

Mark and I are recreating the experiment with a more sophisticated technical setup this semester using the frameworks students developed in the first iteration of the course as models for the group of students now building AiAiO 2.0. Because of this, these students will not have the exhilarating experiencing of inventing the wheel for the first time, but they will have more time to pursue their own creative projects using their AI, and to test more fully what is possible when artists engage with AI with purpose and intention.

Will they make art? I’ll ask them at the end of the semester and I’ll let you know.