How AI Disrupts the Dopamine of Creatives
Is using generative AI bad for the brains of creative people?
This week’s topic is the role of dopamine in the creative process and how AI messes with the chemicals in the heads of artists. I made a short video recently where I said,
“Forgive me for my lack of interest in the jpgs that get 💩 out by AI prompts.”
I thought I would add some long-form nuance to that flippant statement.
Before I bash AI, let me reinforce my reputation as a pro-technology artist. I’ve always been an early adopter of new tech. I’ve been online since dialup. I played with the first digital cameras as soon as I could get my hands on them. I filled my hard drives with MP3s from Napster. I had a blog before “blog” was in the vernacular. My GitHub account is older than most of the developers I work with. I was early to crypto. When there is new tech, I want to learn about it and play with it. I’m just saying that when you hear me rant about AI, it means something. Because AI is exactly the type of thing that somebody like me would be promoting.
But I’m not.
Here’s my experience with AI. First, there’s the professional side. I work as a user experience designer for a big company that makes tools for IT professionals. On that front it’s complicated. Like everybody, we are figuring out how AI fits into the puzzle. There might be a beneficial revolution coming for tech workers, time will tell. In other words, I am invested in AI, at least on the career side of my life.
But on the personal side, as a creator, that’s where I haven’t been able to get on board. With an open mind I have tested generative AI tools (mostly chatGPT and Midjourney) and it’s impressive. My experimentation came in the form of challenges/themes that I used to explore the possibilities of generative AI. I tried to make movie posters for Atari games, muppet presidents, street photography, dream cars, obscure punk art styles, and more. It was fun, but also deeply unsatisfying.
My resistance isn’t that it’s bad technology. My observation is that it messes up the psychology of the artist. Let me explain…
When we talk about the creative process, one thing that rarely gets mentioned is dopamine. That’s because talking about dopamine makes art seem less magical. We prefer myths about talent and genius over theories about chemical reactions happening in the mind. We’d rather say that artists are following their muse instead of chasing a dopamine hit. But I’d argue that much of the driving force (if not the main driver) behind creative output is the pleasure that is found in the flow state. It feels good. Dopamine. Listen to how Paul Simon describes it:
“How do you get there? How do you make yourself feel that chemical high that you feel when you make something that you like? And we don't really know the answer to that… something goes on in the brain and you get a really big reward when you make something that you like… And sometimes people call it flow or ‘in the zone’ or things like that. It’s the mystery of why that happens, but that mystery, and when that does occur to you, the reward is so great, you want that for your whole life.”
This quote comes from Malcolm Gladwell’s interviews with Paul Simon in “Miracle and Wonder,” that I highly recommend. We can put whatever mystical terms we want to on top of it, but dopamine is in the mix when you are trying to understand what makes an artist tick.
So how does AI mess with the dopamine process of artists?
Generative AI starts with the user writing a prompt. It feels like you are “talking” to the AI, but underneath you are kicking off a process that converts your inputs into a formula that the computer can use to query the large language models. You receive the output of the query in the form of words, images, or code depending on the specialty of the AI.
The quality of what AI generates varies widely. Sometimes it is garbage, requiring you to refine your prompt or start over. But when you get it right it seems like magic. You forget the failures and indulge in the dopamine reward from the big payoff. It might seem counterintuitive, but when rewards are unpredictable we actually enjoy them more. The dopamine hit is bigger when we are surprised. There is a name for the psychological mechanism, it’s called “variable rewards.”
Variable rewards are the tool that addicting products use to hook their users. This isn’t a secret, marketers openly refer to this as the “Hook Model” and there is an entire methodology developed to exploit it.
Let’s uncover how variable rewards are delivered by the popular generative art product, Midjourney. Before they had a website or app, Midjourney rose to prominence as a chatbot in Discord. This resulted in a clunky workflow that prevented many non-Discord users from even getting started. This “flaw” unintentionally filtered their first users to be tech-savvy advocates and influencers, a blessing in disguise.
If you made it past the Discord hurdle, the process of generating art was equally clunky. Learning to write prompts involves learning a new language of nouns, verbs, modifiers, and variables. This would seem like a design flaw, however it makes sense from a variable reward perspective. This friction is what makes Midjourney addictive. You have to “work” to make it work. By the time you figure it out, you are invested. And when you experience the variable rewards that Midjourney dishes out it’s easy to get hooked. It’s not uncommon for dedicated Midjourney users to produce thousands and thousands of prompts.
Now let’s play this out in the creative process. This is where I am going to make a distinction between the dopamine hit you get from traditional/analog processes and the new dopamine hit people receive from AI prompts. For lack of better terms I call the old-school hit “deep dopamine” and the artificial hit “quick dopamine.”
Deep Dopamine
Serious artists are doing something hard. If you are really pushing your craft you don’t know exactly what you are doing because you are pushing into the unknown. You are solving a problem. Dealing with uncertainty. Experimenting. Failing. And when you’re done (if you’re lucky) you have something new and original. But there is no guarantee. You could work your entire life without ever having a big payoff. Intuitively we know this. And it prevents many people from even trying. It’s easier to say, “I’m just not creative” or “I don’t have talent.”
Attaining mastery is so hard that even the people who have the guts to try eventually give up. Closets get cluttered with unused paints, unplayed guitars, and empty journals. Most artistic adventures end in silent shame by people who give up. Sad but true.
The people who stick with it, are the ones who can lose themselves in the work. They know how to trigger the flow state and stay there. This isn’t the tiny trickle of instant dopamine you get from your phone. This is deep dopamine, earned not from reliable variable rewards, but long extended time spend without results.
Quick Dopamine
When we lose ourselves in our phones, we are surfing on instant dopamine. It kind of feels like flow, in that we lose sense of time, space, and self. Unfortunately, when we come back to consciousness we don’t have much to show for our lost time. Unlike doom scrolling, however, using generative AI tools is indeed a creative process. You are trying to make something. It’s not as instant as the dopamine from a TikTok feed, but it’s quicker than the method used by analog artists.
I’ve already pointed out how variable rewards are the mechanism that generative AI tools use to delivers dopamine. You write a prompt and wait for a response. That’s your lottery ticket. If it fails you re-roll the prompt dice. Eventually you hit the jackpot and you’ve “created” something that resembles art. I’m not going to get caught up on whether or not the output is or isn’t art (but you can guess where I would land), rather I want to drill in on the psychology.
Post-Creation Psychology
After you’ve used AI to generate a pretty picture, poem, video or whatever, how do you feel? You feel great, actually. Something existed only in your mind. Then you did some work and now it exists in reality. Boom.
Now what? You push it out into the world. You post it on social media, maybe you publish it, print it, put it in frames. Maybe somebody even buys it. More dopamine! Feels great.
But you can’t avoid the question of how to attribute the role that AI played in your creation. If you hide the AI assistance you are going to feel like a fraud. If the truth came out that you hid the AI assist you would be embarrassed and discredited. So you openly disclaim that AI is a part of your process. You aren’t alone, there are vast communities of AI creators who encourage you. More dopamine. Maybe you’ve found your people.
But there will be inner voices. They will say, “I like it, but the AI is the thing that did all the work.” You can’t argue with the observation that, “With AI anyone can do it.” The credit that you “deserve” is deflected, disregarded, and devalued. This might be fine for some, but I believe that for most creative people this is unbearable.
Even if you are able to maintain artistic integrity, if you are able to use AI in a completely original and un-copyable way, you are still going to have that unpleasant little asterisk hanging over your head. Your art was “created with AI.” Is there a way to maintain authentic agency and ownership when there is a whiff of AI in the mix? What are you contributing that nobody else could add? This is the moral/existential impasse that all artists will wrestle with as AI tools embed themselves in our tools.
It’s tempting to end this essay with a warning of impending doom. What happens to a society that unknowingly loses our best artists to quick dopamine addiction? What is lost when creative people trade hard-earned deep flow states for simple prompt-induced dopamine hits? It’s easy to be discouraged as an analog artist in the age of AI. Don’t give up.
I’m optimistic, actually. That nagging feeling of inauthenticity that accompanies AI-generated art is more powerful than it may seem at first glance. Humans need to be a part of something that has meaning. When we are working with a sense of purpose we don’t need tiny dopamine hits, in fact we can persevere in the face of pain, discomfort, and overcome nearly any obstacle that gets in our way.
It hasn’t happened yet, but a recommitment to deep work will return to the arts. Artists are still in the first wave of AI-generated dopamine highs. There is a backlash coming. I predict there will be a resurgence of interest in analog processes and non-digital creations—not just by the artist but by consumers who will want to support the people behind the creations. And that’s where there’s hope. Take heart. After the quick dopamine hit wears off, humanity will hit back. Stay creative.
Your friend,
Ade
P.S. If you’ve made it to the end, you may be interested in my ideas about AI attribution. Check out this repo of assets that you can use to identify your original artwork as AI-free.
