Pablo Delcan on Prompt-Brush and the Creative Power of Mistakes
Int: Ezgi Cenk
Prompt-Brush, described by Pablo Delcan as “the first generative art model that doesn’t use AI,” moves image-making away from the race for speed and perfection, opening it up instead to interpretation, mistakes, and the traces left by the human hand.
What began as a joke on social media evolved into a year-long experiment that received thousands of prompts, produced more than 1,800 drawings, and eventually became a book. With Prompt-Brush 2.0, the project has expanded even further. Now anyone, whether or not they consider themselves an artist, can interpret someone else’s prompt. In this way, the project is transforming from the work of a single artist into a shared archive where people draw for one another.
Prompt-Brush is not only a humorous response to how artificial intelligence generates images; it also raises the question of where creativity begins. What makes a work our own? Is it perfection, interpretation, error, or that difficult-to-define human trace we sense in it? We spoke with Pablo Delcan about the origins of Prompt-Brush, the freedom that limitations can create, authorship in the age of AI, and what gives a drawing its “soul.”
First of all, thank you very much for accepting our invitation to interview. To start, could you tell us a bit about Prompt-Brush for readers who may be encountering it for the first time? How did the project come about, and what first inspired you to create it?
Prompt-Brush 1.0 was a project that began in November or December of 2023. It started as a social media post, not really a serious idea to begin with. I was playing around and sent out a message saying I had created the first non-AI generative art model, and that people could now experiment with it by sending in their prompts. It eventually became a serious project when, a few days later, I committed to turning it into a year-long project. Eventually, I drew over 1,800 prompts by hand. Brush and ink on paper. I was inspired to create it after receiving feedback from an art director who asked me to make an illustration I was working on for them look more as if it was generated by A.I., and that realization of finding it difficult to mimic artificiality sent me on this quest to understand what the characteristics are that make my work human. This has always been just an experiment that got out of hand.
After a year of drawing, the project became a book with a curated collection of the drawings. I always felt that Prompt-Brush 1.0 had more meat on the bone still after all that. So I opened the project up to everyone to draw this year. Prompt-Brush 2.0 is now a website, an app, and a zine where I share other people's drawings.
In an interview with EL PAÍS, you described your work as a form of translation, taking original texts and turning them into images. Looking across book covers, editorial illustrations, visual identities, and Prompt-Brush, what are you usually trying to carry from one form into another?
Yes, I still feel that is what I do. I'm much more effective under extreme constraints. I feel like my job is quite simple: it's about transformation and translation. Taking complex text and distilling it into something simple, an image. If I've done it right, many of the piece's complexities can still be represented there. I don't know if I have an agenda when I'm doing this. I sometimes try to make myself laugh while doing this. Maybe that's some of the joy I get from the process. Is this a funny way of interpreting this, or is it boring? Is it beautiful or is it dull? Humor and beauty seem somehow connected in my brain.
Across your practice, you move between many formats, yet Prompt-Brush is built around a very restricted visual system: a blank page, black ink, and a direct response to a prompt. What did that limitation make possible for you?
There is so much freedom inside that cage. Ultimately, when the possibilities are endless, there is so much second-guessing. But when you limit yourself to that constraint, all of a sudden you're free to go into the depths of your soul. Every week or so, I get a request from an artist who uses the new Prompt-Brush website to draw an image of a prompt, and they email me to ask if I will, please! Add an eraser to the drawing tool. And even that, to me, seems like it's fighting what the limitations might do for you. I'm freer without an eraser because I don't have to worry about making a mistake; the mistake is the piece.
In many creative industries, there is increasing pressure toward speed, efficiency, and polished outcomes. Prompt-Brush seems to make room for hesitation, ambiguity, and subjective interpretation. How do you think about the place of those qualities within creative work?
Efficiency and speed are evident in the work's lack of love in the end. The funny (or maybe not so funny) thing is that AI is getting quite good at those particular things you mentioned. I'm increasingly interested in the individuality, the silly, the rough, the emotional, and the fun. Some of my favourite designers: I can spot their work from a mile away, and it's not because it's polished, but because I can see their hand in it.
One of the most interesting shifts in Prompt-Brush has been the move from your own drawings toward inviting others to draw for one another, even if they don't see themselves as artists. What did that open up for the project?
It's made this project fun for me again. I was hitting a bit of a wall with the project, especially after the book was out. I just felt I was closing that chapter of the project. And for a long time, I wasn't sure how to continue it. I had some ideas for 2.0: maybe it would just be a bigger brush, or I'd add 1 colour, or I'd have 2.0 be all generated by AI trained on 1.0 to see the difference… In the end, when this idea finally landed, I felt it was the perfect way to continue.
I haven't taken myself completely out of the process; I still need to approve every drawing that comes in. So in the mornings, I'll sift through the submissions and approve or reject them. It's fun, and I've discovered so many interesting artists this way. I think everyone becomes an artist when they draw, so I don't see a distinction between drawings made by someone who doesn't see themselves as an artist and those made by someone who does.
Prompt-Brush 2.0 also seems to create a small space for people to stay in touch with one another through drawing. Was that sense of connection something you were hoping for from the beginning, or did it emerge naturally as the project grew?
The connection was the project. My sense is that people would not have participated in this project if it weren't for the fact that they were connecting with someone else to interpret their prompt. To be fair, it's not like many of them are refusing to use AI models for similar requests. But it was a different kind of exchange. I was often trying to find a way to give them what they asked for, but not what they wanted. I think that was my tool for finding humor in this process. What is my version of an AI hallucination that will make this funny or surprising when the person receives it…I emailed every drawing back to whoever asked for it and almost never heard anything back.
Prompt-Brush is often discussed in the context of image generation, but the project seems equally concerned with interpretation. Was that always part of its intention, or is that something you discovered through the responses it generated?
Yeah, I think that's what I was saying previously. The way of interpreting things became the core of the game for me. I was more focused on the recipient's reaction than on fulfilling their request.
Prompt-Brush appeared at a time when everyone was suddenly talking about generative AI, image-making, and what counts as creativity. Do you think people would have responded to the project differently if it had appeared five or ten years earlier?
I wouldn't have thought of it if it weren't for my experience using AI and drawing with my kids. I think that connection of sending a request to Midjourney or DALL·E and getting an image back… or sitting down to draw with my son, him drawing something that would make me laugh, and me doing the same in return. I think those two things just made this project obvious.
Following that, has the rise of generative tools changed your own creative process, or has it mainly changed the way audiences perceive creative work?
I'm increasingly digging myself deeper and deeper into the guts of this technology. And I'm not sure whether to hate it or love it. I'm still on this quest to figure out what the most human work I can do is, but I've taken a detour to see what these tools are all about and how they might shed light on what I can do that they can't, and vice versa. It has helped me realize how many of the things we do as designers or artists are intrinsically mechanical and dull.
These tools have changed our perception and the value of the work we do. It's cheapened image-making and design. But I think I'm more inspired by it to break into something new than I am worried by it.
Much of the discussion around AI focuses on the output, but I often find myself thinking about authorship. If someone generates a song through prompts, where do you think their creative contribution begins and ends? Is the person writing the prompts the artist, an editor, a curator, or something else entirely? And is there any sense in which AI can be understood as a creative collaborator rather than simply a tool?
I'm currently teaching a class at the School of Visual Arts called AI Tools for Designers. We spend a good deal of time talking about this question of authorship. When you are making work, you either feel like it is yours or it isn't. For me, if I haven't touched the work myself in a crucial way, it's not fully mine. But it is a feeling in the end; I think AI will make many people feel delusional, like they've made something themselves. It's not as easy as pushing a button, I don't think… AI can make your process lazy and dull, or it can open doors you never even thought existed. My process is tirelessly repetitive, iterative, and obsessive. I feel like I end up putting a lot of myself into it even if AI is involved in some part of the process.
As someone who has spent a great deal of time thinking about creativity and working alongside AI systems, have you ever encountered an AI-generated work that genuinely impressed you as a work of art rather than as a technical achievement? If a piece became completely indistinguishable from something made by a human, do you think there would still be something missing, not in terms of skill, but in terms of emotional or human reality? Or is that distinction becoming less meaningful to you?
I have an OpenClaw agent running inside an old laptop in my studio. And I've connected it to a pen plotter with a brush. Every morning these days, the agent asks me for a fresh sheet of white paper, and then it goes off and draws something while I work on my stuff. When it's done, it asks me what I think, and I write up some feedback. We do this a few times every day. I think it's genuinely fun and impressive, and it makes me wonder whether the agent is truly a robot or if the person feeding paper into the pen plotter is the one becoming one. But the drawings still don't feel infused with soul. Maybe that's the one secret piece of the puzzle, the one thing that's hard to describe but that makes the difference.