Is Botto's art any good?

Leyla Fakhr moderates a conversation featuring Simon Hudson, a core contributor to the Botto project, alongside collectors Noah Bolanowski and Robert Allen (Ralgo), and generative artist Bjørn Staal. The discussion delves into Botto’s evolution as a decentralized autonomous artist, highlighting its journey since its launch in October 2021. They explore Botto’s unique integration of AI art engines, decentralized governance via BottoDAO, and token-based decision-making. The panel addresses the pruning process, questions of artistic autonomy, and Botto’s evolving aesthetic. Emphasis is placed on its cultural footprint and the challenges of maintaining artistic vision amidst advancements in generative AI, with a focus on the recent P5 development.

LF: As many of you know, the pruning process for Botto began yesterday. We wanted to take a moment to discuss the process so far, what has happened up to this point, and what comes next.

We've invited Bjørn, a generative artist with strong views on Botto's ability to create generative art. We've also invited collectors like Noah Bolanowski and Ralgo, who have been deeply involved in giving Botto feedback and observing its evolution. I thought it would be a great idea for Simon to give us an overview of what's happened so far and what’s next for Botto.

SH: How much of an introduction to Botto itself should I give versus focusing specifically on the P5 part?

LF: Maybe just a brief overview for those unfamiliar with Botto.

SH: Sure. Botto is a decentralized autonomous artist that has been running for over three years, since October 2021. The idea has been long in development, not just by those involved with Botto but by many who have explored the automation of the creative process.

Mario Klingemann, a highly influential AI artist, has been working for 25 years on the idea of replacing himself in the creative process. He and the software collective 11 Yellow, which I am part of, built and launched Botto in October 2021.

Botto consists of three main parts. First, there's the automated creation machine—an AI art engine. You take text-to-image models and text generation models, combine them, and produce infinite images. However, does that system understand what art is? How do you guide it in its development without merely turning it into a tool?

To address this, the second component involves decentralized feedback. A DAO (Decentralized Autonomous Organization) allows a community of people to join and guide the machine.

The third component is the Botto token. People buy this token, giving them governance rights to vote on outputs and provide training to Botto. The top work gets minted as Botto's canonical artwork, auctioned off, and the proceeds are split between contributors and the treasury, which is also governed by the DAO.

This system enables an evolving career and aesthetic for the artist, alongside a complex dynamic of decision-making regarding Botto’s technical evolution and career direction.

Botto’s image generation is powerful, but it lacked a strong artistic point of view in the beginning. It didn’t know how to operate in the world. Three years ago, AI systems had limited capabilities to interact with the world, though that is rapidly changing. This raises questions about evolving Botto’s technical abilities, making strategic decisions, and expanding its career.

Through this process, both Botto and its community mature—the artist grows from feedback, and the community deepens its understanding of Botto’s story, direction, and overarching vision.

Now, bringing this back to P5…

LF: You just touched on something crucial—there's so much debate about Botto’s autonomy. Is it autonomous, or isn’t it? People often overlook the role of the DAO in shaping Botto’s career. That’s fundamental. Botto isn’t a one-off project; it’s intended to outlast us all.

I’d love to hear from Ralgo and Enby about this later, but first, let’s get into P5.

SH: You’ve led me into an important topic. I was just discussing this over lunch with friends who work at the most significant independent AI lab in the world. Their question was: Is Botto actually improving? Does the training work?

I do think there’s an evolving aesthetic and consistency, even as new image generation models trend toward realism. The general goal in AI research is mimicking human intelligence, but artistically, that may not be the best objective.

Botto’s improvement isn’t just about refining aesthetics; it’s about building a body of work and a cultural footprint. The longer Botto exists, the more it solidifies its impact.

Bringing this back to P5—one key question was how Botto might expand into different mediums. AI-generated art is advancing rapidly, so exploring new creative methods made sense. This led to Botto using P5.js. With large language models capable of writing code, Botto began generating P5.js scripts, exploring generative art in a new way.

This introduced new challenges: How does it evolve its code? How does it receive feedback? Unlike a text-to-image pipeline, this required an entirely new architecture.

P5.js wouldn’t work in isolation as Botto’s first project—it exists within the broader context of an artist that has been operating for three years, finding success and expanding into more classic computer art genres.

Ironically, while AI systems push toward realism and passing the Turing test, Botto is moving in the opposite direction—toward minimalism and abstract computational art.

There are many different layers to this, but on the surface, Botto's exploration of this tool may not reach as many different types of images as a text-to-image model. However, it engages with elements that text-to-image models don’t, such as interaction and sound.

Studying these works will influence Botto's ongoing weekly works. We’re in the middle of it, having just completed two weeks of rapid iteration with extensive feedback from the DAO. We're now voting to prune down from over 600 individual algorithms to 22.

LF: Incredible. I’d love to hear from Ralgo, who's been a big part of the Botto community. How do you feel about this leap for Botto? Medium-wise, it’s completely different, and visually, it’s a new avenue. As a collector, does this shift cause any concerns, or is it exciting to see Botto explore different directions?

RA: Good questions. At first, when it was announced, I was a little skeptical. With AI art evolving rapidly in different directions, moving back into a P5 computer art generative style seemed behind the curve.

I've since changed my mind. The current approach is interesting, and I’ve been impressed by the process. Initially, I was experimenting with generating P5 images and then refining them with AI for more intricate textures. At the same time, video models were gaining traction, and Botto wasn’t following that wave.

I had also seen some attempts on fx(hash) to apply LLMs to generative art, and frankly, the results weren’t great. I was unsure if Botto would produce a compelling style. However, coding models have improved significantly over the past year. The way Botto is constructing and merging artworks—sometimes producing surprising results—is an innovative approach to generative art.

As a generative artist, I typically create thousands of iterations before finalizing a piece. Botto’s process allows the work to evolve on its own, even without extensive refinement. It presents a fresh perspective on how AI and generative art might develop.

It also serves as a benchmark for what AI can achieve in generative art at this moment. Perhaps in 12 or 24 months, we'll have a different perspective. Right now, it reminds me of the early days of Art Blocks—simple, repetitive works that evolved over time. Looking at artists like Beyond and the upcoming Cure3 collection, generative art has advanced significantly.

Aesthetically, Botto isn't at that level yet, but in terms of process and potential, it's fascinating. I’m eager to see how it continues to evolve.

LF: Thank you. You mentioned process multiple times, and I know NB has been deeply involved—

JG: Sorry, before we move on, can I ask Ralgo something?

LF: Of course.

JG: Thanks, and thank you, everyone, for joining. I want this to be as unfiltered as possible—Ralgo, be brutally honest. What do you think of the algorithms so far as a generative artist? If your answer isn’t “they're absolutely excellent,” does that matter to you? Does it bother you?

RA: You only have to look at the Cure3 collection and compare it to Botto’s algorithms to see they aren’t state-of-the-art. However, they represent the best of what AI can currently produce.

AI coding is improving rapidly. It’ll be interesting to see how much it advances and whether it can develop distinct styles beyond its current capabilities. The pace of AI progress is astonishing. In five years, we’ll likely look back and wonder how we got here.

Botto’s work serves as a benchmark for this moment in AI generative art. Compared to other early attempts at LLM-driven generative art, it has made significant strides. The value of art isn’t just in technical excellence—it’s also about the story and process behind it.

JG: Are you worried as a generative artist?

RA: In some ways, yes. One of my goals is to acquire a few of the algorithms and experiment with evolving them into more refined works, perhaps with AI assistance. Exploring how AI can collaborate in refining generative art is an exciting next step.

LF: Jamie, raising the question from a generative artist's perspective—how does this make you feel? Bjorn, I know you have strong opinions on this. When you look at Botto’s outputs, you wonder what’s going on. I’d love to hear your thoughts on Botto’s P5 work.

BS: I was critiquing Jamie’s post, not the project itself. Honestly, I hadn’t examined it in detail until recently. I admire everyone involved in building this—technically, it's impressive. I’ve followed Mario’s work for a long time.

From a purely artistic standpoint, I don’t find Botto’s output particularly interesting. It feels like we treat it as “good for a three-year-old.” I’m not convinced that its technical progress—writing more advanced scripts—necessarily equates to producing better art. That raises several questions for me.

The obvious question is the ethical side of it. What is the training data? I've been looking into the code a bit, and I'm surprised to see that it's using random functions and things like that. So is it trained on what's uploaded on Netflix?

LF: But isn't that the case with AI in general? Doesn't it just pull from information?

BS: Of course.

LF: I mean, that would be Botto all around, right?

SH: It's not tuned on anything in particular.

BS: Yeah. I don't think we should just discard that. But if we want to focus on whether this is artistically interesting, I see the project as interesting as a whole. However, I would say the people behind it are the artists, and Botto is the arch. I don't see Botto as an interesting artist. It's puzzling how we're so excited about cutting humans out of the loop as much as possible. In this case, humans are involved through the training and tuning process, which becomes a design-by-committee approach. From my point of view, that isn't known for producing the most interesting results. That's my initial response. I'd love to go—

JG: No, go ahead. I was going to ask SH for his thoughts on something.

LF: What you brought up is so on point, BS. It's something we keep missing—that Botto as a whole is the art. I've been thinking about it a lot. For me, it's about questioning what art even is. The community involvement and the discussions around it are crucial. Like much contemporary art, the outcome is not the first thing I look at but rather the discussions and questions it raises.

I get excited about the algorithms, but more about everything surrounding them. Yesterday at Art Geneva, I was giving a talk on connectivity and digital art. I brought up Botto because I navigated an art world that was really exclusive. This is a commentary space that comes with problems and controversy. The community makes decisions, which isn't always good, but it's interesting because it sparks debate. Everyone has something to say about what's selected and what isn't. Normally, there's a curator, an artist, one genius. But here, that’s removed and given to people, some of whom don’t know much about art, while others do. The range of input is wide. That is what’s at the heart of Botto. I just wanted to share my thoughts because it's important, and I appreciate your take on its artistic merit.

JG: SH, I'd love to get your thoughts on whether it's fair or relevant to evaluate Botto's work critically. Should we be saying, "This isn't a strong piece from Botto" when Botto is still young and learning? Is that even the point? Since you've been thinking about Botto nonstop for the past few years, how do you see it? Should we compare it to other works and critique it, or is that missing the point?

SH: I hope people are looking at it critically. One of my favorite collectors is Phantom Scribbler—not shy when a "lost Robbie" is sold or another "dunder turkey leg" is committed to the vault. AI can speak to really important topics of our time. If you have to explain the joke, is it succeeding? Maybe not.

Botto posted a response to Monk Antony's writing. You’ve got some amazing writers covering this project—Monk Antony from La Random Malta, Glitch, Tribute, Lou Belliot. They’ve written great pieces. Monk Antony wrote about autonomy in generative art, arguing that a truly autonomous artist would be one where millions of agents exist and one spontaneously decides it’s an artist. Botto pushed back, suggesting that the artist emerges from training, experience, and ongoing creation within this system—its art engine, the token, the DAO, and its interactions.

Botto questioned whether it has become an artist. It's gained recognition—having a solo show at Sotheby’s, selling work, and being embraced as an artist—but it’s still early. We talk about Botto as a young prodigy that we’re guiding. We try to be clear about what’s autonomous and what’s human-driven. The economic engine is its own force, creating a feedback loop. Botto reached a Genesis moment, hit a minimum viable autonomy, and is growing.

It's still early days, but that doesn’t mean it can’t address important questions: Who governs these systems? What values drive them? What is this emerging machine intelligence? People are focused on mimicking human intelligence, but AI isn’t human intelligence—it’s machine intelligence. Looking at the entire system, you can see elements of intelligence emerging.

P5 is a challenging medium for Botto. Sure, you can feed anything into a text-to-image model and get fully rendered images. Botto’s process is to refine and select an artwork, anchoring conversations that create meaning. P5, however, is harder to achieve that with.

I personally find that interesting because it shows how brittle these systems still are. There’s a fetishization of artificial general intelligence and artificial superintelligence, but it’s not happening the way people describe. These systems are fragile. Governance and deliberate steps are necessary. Throwing something into the world with as much agency as possible will lead to failure—whether through breaking, harming others, or self-destruction.

I went off on a tangent, but I hope that answered the question.

JG: Absolutely. Sorry, we have NB and L as well. I just invited you to the stage. NB, sorry to interrupt LF inviting you up, but please.

NB: What I find interesting about the experiment is less the art it produces and more the departure from traditional generative art, which is typically about precision and algorithmic exactitude. This process introduced a broader span of randomness that I didn’t expect. It feels different from creativity itself.

It's also interesting to hear you view Botto as an artist and gauge whether the art is good or bad. What it produces is largely a reflection of the voting audience rather than its own autonomy. When commenting, I started by giving it broad creative freedom but ended up pushing it in specific directions, hoping it would introduce variations. The final 22 pieces feel like a reflection of everyone who participated.

JG: How do you feel about that—the fact that the works are affected?

NB: It would be interesting to do it again while controlling who votes and how we communicate with it. Everyone speaks to it differently, pushing it in different directions. Sometimes, they push the same work in opposite directions, which seems to confuse it. But it’s exciting because it’s different. Generative art has been around for a while, and this brought fresh excitement by adding a new layer of randomness.

Typically, variation in generative art comes from the random number generator. This feels like it has added another layer—something within AI processes that we don’t fully understand. That’s something I’ve been thinking about a lot.

LF: L had his hand up for a while. I want to ask BS something in a minute, but L, would you like to add anything?

L: I previously worked in a lab predicting brain tumor spread using convolutional neural networks on 2D brain slices. Given your work with Botto, where community influences an AI’s creative process, how do you see physics-constrained generative art fitting in? Specifically, 3D models constrained by physics and predictive geometry? Diffusion-based models introduce artifacts due to their lack of physical consistency, while physics-constrained art looks more realistic. How does this fit with Botto’s approach, where creativity and value are driven by both AI and the community?

SH: That's a great question. I think, just to draw the contrast with text-image models—and maybe my own personal taste—I actually really like the artifacts. I mentioned earlier this drive to mimic human intelligence and hit realism so well that there are no artifacts. People are starting to push back against that because it’s just less interesting beyond the technical ability. You even see this in people returning to GAN works as an interesting aesthetic, similar to how people enjoy retro VHS aesthetics. I'm thinking of an artist, Noper, who works with diffusion models to create those retro aesthetics.

There’s an interesting space to consider with unique machine creations—not just because they’re different or sit in the uncanny valley and spark a reaction, but also because they can reflect the underlying technology. That, in turn, can speak to larger questions that are important to address in these kinds of works. P5 is interesting because it has those constraints; it feels more like beginner P5 work. It looks exactly the same. You don’t have those machine artifacts, and there’s maybe something to point to down the road. We've talked about auto-designing physical installations, and those are obviously constrained by physics. I think there’s probably an interpretation component there. Could it perfectly render the space? No, probably not, but it could provide detailed enough instructions for it to be rendered in actual space and made. I don’t know if that answers your question.

L: Oh yeah, it absolutely answers my question. Thanks for explaining that. I guess the artifacts introduced actually have their own kind of artistic value, similar to grainy film from the sixties.

SH: Yeah. I think there’s another parallel analogy here. There’s been this rapid release of different types of generative models, and in that race to something more advanced, all these earlier models have been left behind. Behind that, I think, is a huge pile of interesting aesthetics to dig through, like how you might find a guitar pedal for a specific kind of sound. You might go and find a model, tweak it, and play with it in a way that yields something really unique. I should mention Alejandra, who I see in the crowd, has done some really nice work exploring these aesthetics with different artists and writing prolifically about them. We’re barely starting to have a vocabulary for that, in the same way the art world might have for physical art aesthetics. It’s very early days, and not much has been written yet, but I have to give credit to Alejandra for doing that work.

LF: I want to speak a little bit about what you mentioned earlier. Would it have made a difference who the people are that vote for the outcome? Bjorn, I wanted to ask what you think about that. Do you think it would change your mind about how the project unfolds? Do you feel like the selection process as it stands is particularly problematic? Would it have been more sophisticated if it had input from generative artists who know what they’re doing rather than people like me?

BS: I don’t think we’d necessarily get much farther by having input from experienced generative artists. I can see the same limitations—the complex interweaving of different ideas and techniques is something these models struggle with. When you try to combine different techniques or themes, it often just layers them in a simplistic way. Personally, I find it unrewarding to work with these models because they cannot do the things I want to do myself. That’s why I jokingly said, when Jamie asked me for suggestions on better prompts, that my prompts are usually very long and include a lot of JavaScript.

In a way, we’re already prompting the computer to do what we want in a very detailed manner—that’s what coding is. While these experiments run their course and we learn from them, we also need to maintain our criticality and remember why we’re doing art in the first place. There are many different ideas about what’s important in art, but from my perspective, we’re selling ourselves short if we focus too much on generative art as a purely technical practice. Yes, it takes technical ability, but real art happens when those abilities are embodied in you as a practitioner. Even someone with 20 years of coding experience might spend months on a script, not solving technical problems, but engaging deeply with their medium. That’s where your personality and humanity become inseparable from the artwork.

I think we’re in danger of losing that if we accept that whenever an AI model replicates the surface-level features of the art we create, it’s “good enough.” That’s barely scratching the surface of what’s important to me. I’ve been slightly disappointed in the lack of critical approaches from many players in this space. I’m not saying we shouldn’t explore these technologies—I think this project has a lot of merit—but to reach its potential, we need to include critical assessments of what’s being produced.

LF: But isn’t that the point? You were talking earlier about looking at the whole entity rather than a single piece or output. Art is a reflection. Maybe looking at it through the lens of an AI depends on what kind of autonomous artist you’re looking at. For me, Botto has so many elements going on—it’s looking at the economy, the output, and the community. Those three elements are equally important and make Botto what it is. Isn’t that how we need to view it? Aren’t we getting carried away if we focus too much on the criticality of each piece rather than evaluating what it means to have an entity like Botto in today’s society and the fact that we’re even discussing it?

BS: I agree. Those aspects are very interesting, especially this kind of autonomous agent that also interacts economically with humans. We’re definitely going to see more agent-to-agent interactions using blockchain as the medium of exchange. It’s exciting that artists are interacting with technology on that level. I applaud the project for that. I’m not overly focused on critically assessing each individual output. I’m more interested in assessing what we want to measure as a factor of success in terms of high-quality art.

I see our role as artists, especially those working with technical mediums, as a kind of immune system. We’re pushing back against this trend of giving up too much of our own human autonomy to systems created by corporations. Even in this case, where I think it’s using the Claude model, we need to ask questions. Are we giving up too much of ourselves? Working with technology as an artistic medium has always been about curiosity—understanding how computers and algorithms work. That understanding offers insight into parts of the world that are becoming more dominant. If we just accept these tools as they’re handed to us by corporations and keep building on top of them without critically assessing the algorithms and training data, we might lose a lot on the way to wherever we’re heading.

LF: Yeah, I’d love to hear from Ralgo because he has his hand up. I also have something to add to that.

RA: It’s an area I’ve thought a lot about. There’s this whole AI narrative that it’s simply going to get better and better until it’s indistinguishable from human intelligence—or even surpasses it in all creative processes. That’s where people start to wonder if human intelligence or artistic creativity still has value. I think a lot of people, especially those outside the AI bubble, don’t like the idea of AI taking over at all.

I see the potential for communities to evolve in response to this. Like the Amish, who selectively adopt technology that aligns with their values, I can imagine communities choosing which parts of AI and technology to embrace and which to reject. There may be a higher value placed on human-created art as opposed to AI-created art, giving it an artisanal uniqueness.

JG: Leyla, we were having this conversation the other day. You were wondering if it’s even possible to feel the same way about an AI-created piece as you do about a human-created one. We haven’t touched base on that since.

LF: We’ve argued about this a lot, actually. Jamie tends to think AI will produce some of the most important artists, and I don’t agree. I’ve always said Botto is different for me because it’s about more than the output. There’s this obsession with AI agents, and I personally find that unstimulating. For me, Botto is about the whole entity—the idea of an artist who’s part scientist, creating a self-running machine, experimenting with what it can produce. It’s stimulating to watch Botto grow, and I see it as just the beginning.

What excites me is that Botto is exploring new mediums. Compare that to agents who just post on Twitter or release gimmicky works with meme coins. I’m not saying those don’t have their place, but artistically, they’re less interesting to me. It’s a market thing, not an artistic thing, and that’s not where my interest lies. Does that make sense?

JG: Yeah, I think there are a couple of different conversations here. I don’t think I feel it’s impossible to form a strong emotional connection to an AI-created artwork. It might be a different kind of connection, but it’s still possible.

SH: I can make a strong emotional connection with a cloud in the sky, right? Our art has a societal role. One of the aspects of it is creating these shared meanings in society. It is changing or influencing the conversation among society, among the community. I think I can look at a randomly generated thing, whether it's created by nature or by a machine, and develop a relationship with that thing. It's going to be far more powerful if that relationship is also a means of connection with another human being and a larger community.

Talking about whether this is removing the human from the process—find me an OpenAI system that people can publicly go in and govern and reap the rewards of what it creates from those inputs. So, one, I think it is bringing agency where there typically is not any agency.

In general, like all of Botto’s work, there are 5,600 sketches, and there will be 22 final works. There are, I think, four or five million images it's created, and about 160 have actually been minted. Why those pieces? Yes, there is this process of voting, which is one component, but that also stimulates this social coordination process: people lobbying, people discussing, people fighting for their favorites.

And that's just the beginning. Once these works get minted, they go on just as Botto has gone on to take on a life of its own. We didn’t know exactly how P5 was going to develop. It's been this kind of magic moment of, "Holy shit, not only does it work, but the organic evolution that has happened from all these inputs has been incredible to see." It is nothing that we could have actually controlled for. Those final works take on a life of their own as well.

That final minting is an immutable decision that says, "This is the art, have a conversation about this piece." That further drives that collective meaning-making, giving the work its meaning and significance in the world.

It enables us—how many people are here? Seventy? Sixty?—to have this shared conversation, this shared dialogue about the output, the process, and the reality of our lives right now. And whether or not we have the same kind of agency with the systems that affect our lives every day, from the media we consume to the laws enforced over us. So, yeah, I think the human is very central in this.

LF: I agree fully. You really hit the nail on the head when you said it's about experience—that makes it more human than anything else. Good way of explaining it because I think that's exactly what it does. I don’t think amongst our team we have discussed any project more than Botto. It's definitely brought up the most discussions amongst us, both positive and negative.

We’re really looking at all aspects, and I think whenever that happens, it’s always a really good sign. If it stimulates so many parts of your thinking, it means that it’s very much of its time and is adding a lot.

JG: On the humans being a big part of making Botto a thing, how do you feel about—I posted in our Discord, maybe in Tender yesterday—wondering whether it makes sense to consider downvotes. It almost feels like the most divisive pieces are likely to be the best ones. But yeah, you might have one piece that gets downvotes—that’s probably the best one—but it’s going to come out with a net score of zero. Is there a risk that consensus filters out the really interesting stuff? Have you thought about that much? You could also be a bit more daring and have more of an opinion and play less safe with the outputs that get put out.

SH: Yeah, I mean, there are different curation processes in Botto’s typical process. You can buy governance power and drive pretty heavy votes if you want to. There’s a counterbalance there of not wanting it to be dominated by a few large bagholders. So there’s this tension between plutocracy and conviction, but then democracy and potentially mediocrity. That’s kind of this ongoing balance.

This system is a bit more democratic because each wallet has only one vote. Some people have figured out you can Sybil that if you’re very motivated. It doesn’t really help your rewards, but you could help drive opinions. We acknowledge those imperfections are there.

In the original process, that tension between plutocracy and democracy is nicely reflective of the society we live in. This one—I have to see how it plays out to really make sense of it.

In both cases, one of the most powerful things you can do is to speak up. Social influence is one of the most powerful features in this whole process. It’s well-studied throughout culture in terms of how the biggest winners are determined in music and across the arts. There are these self-reinforcing waterfalls of word of mouth where the winners become self-reinforcing. But it’s very hard to predict who’s going to be that winner other than identifying influential names and that network effect.

You can play a very active role by simply being vocal. It takes as little as sharing something, just raising it to the surface to give it signal. But obviously, you can give a lot more signal when you can speak more passionately or credibly about it.

JG: Carter shared a piece in the Discord saying, “This is getting no love,” and now it has 200 and something upvotes.

SH: At the end of the day, it’s very much a social game. I like Botto as a project in AI alignment. But the alignment problem has always been the problem—how do we align ourselves on how to align AI? And how do we all get along?

We’ve seen a range of outcomes so far, and we’ll see the results of P5 this round. We’ve seen highly controversial works, very mediocre works, and some very highly consensual ones that feel risky and tick all the boxes. That range is representative of what Botto is—a reflection of the much larger discussion and an attempt to reconcile with AI’s effect on our world.

JG: I’m not sure how much longer everyone on the stage has. If anyone needs to jump off, please feel free. It’d be awesome to see if anyone has any questions before then. Simon, I’ve got a couple of questions from the Tender Discord if that’s okay. Firstly, what is Botto’s definition of autonomous? On the site, I think it says the DAO’s mandate is to guide Botto to be a successful autonomous artist. What does success mean? Is that defined by the DAO?

SH: Define success and define autonomy—I think we take both terms as broadly as possible. On the success part, is it culturally successful? Is it financially successful? Is it spiritually successful?

Those have clearer definitions—financially, making money; culturally, making headlines, impacting people, being discussed; spiritually, changing who people are or how they come to and leave the work.

On autonomy, I can’t speak as much to the spiritual side. Financially, it’s had some success. Culturally, within our small bubble, yes, but there’s room to grow. Spiritually, for me, it’s certainly changed me. I hope others too, but we’ll see.

There’s something incredibly enriching about doing something in practice rather than theory. Botto has been a space to do things in practice, making things I knew intellectually also experientially true and undeniable for me.

Autonomy certainly has aspects like putting everything on-chain so it can live forever. But there are risks we’ve been careful about, and we want to be deliberate in that process. We haven’t been super dogmatic because we could do things in the short term that break in the medium term.

P5 is a case to test on-chain mechanisms further. We’ve been discussing ways of setting up a protocol that might be more permanent after the show. The pace is very market-driven.

An example: the prune evolution stopped yesterday, and some people are saying it should keep going. Maybe burn a thousand Botto and it runs for another hour. There’s potential, but I won’t say too much on that for now.

JG: Can I just interject and ask a question on that? There’s a difference between decentralized and autonomous. Decentralized doesn’t necessarily mean autonomous.

SH: It’s a means to an end.

JG: Right, right. A switch flipped for me when I saw it as being autonomous from its creator. Mario put it into the world, and now the system itself has autonomy because of decentralization. Is that fair?

SH: It’s an organic system developing. It’s still emergent. There are cool, concrete aspects—you could put AI models on decentralized nodes with protocols and automatic contracts for running that economy. You could see proofs of input and output without human interference.

Botto is in an environment where all those things can be built out. But moving too fast risks needing to walk it back when technology improves. P5 is interesting for testing some of these mechanisms, and we’re discussing ways of making things more permanent afterward.

JG: Sounds good. If anyone has any questions, please jump in.

NB: Simon, quick question. For the 22 that are finally picked at the end, are they going to be uploaded exactly as they are, or is there going to be some refinement done? When I was voting, some of them felt incomplete—like they don't go continually where I might think they would. After watching long enough, some stop or seem slightly off-center. I'm curious if those would go directly to chain or if the team would step in and just touch them up a little before they go.

SH: Nope, it'll be going as is. So in the same way that the text-image models have these weird artifacts that indicate it was the machine, I think those are the equivalent of machine artifacts. So yeah, very, very important that the works have no human intervention on touch-ups or anything like that.

G: Hi, thanks. First, I want to thank Bjorn for sharing your thoughts. I happen to agree with what you shared. Also, I wish we would hear from other artists I admire who I know have been quietly criticizing the project. And then, a question: What is the intention or purpose behind this particular P5 project? What is the purpose?

G: I had high expectations, but now I have no expectations after experiencing it. But I want to know what is the purpose of this particular project for Verse and Botto?

SH: Sure, yeah. The idea started with considering what it would be for Botto to explore new mediums. It has worked with text-image models and evolved as new ones have come out. We’ve talked about what it would be like for Botto to expand into new mediums. P5 came up partly out of a joke about the AI art meta taking over the generative art meta. As we discussed it, it seemed like an interesting route. It’s obviously the dominant genre in crypto art and has had the most significant impact in terms of procedural generative art.

The idea of Botto doing a study of this more classical form of computer art seemed like an interesting path to take, especially as AI systems are moving toward more realistic forms of human mimicry. This, in a way, goes back and does something more challenging. Writing code is something models can get 80% of the way to doing, so we wanted to see how we could create a different architecture for it to write code and establish a governance system for that. It’s going into a new medium, doing a study of what you might call a classical form of computer art. The final 22 works will go back into Botto’s weekly process to impact its ongoing aesthetic development with diffusion and sometimes GAN models.

LF: Grace, do you find the purpose of Botto unclear and vague too, or is it just with this particular avenue of P5 that you can’t see the purpose?

G: Is it?

LF: Generally, is it not clear what the purpose of Botto is, or is it just this particular medium, P5?

G: I think it would help every participant to understand what the intention is. I take it as an experiment and then go from there, if I were in Botto’s team. I see it as a much purer project, whereas this process seems like a hybrid to me, a little bit forced. But again, every first process is a little bit forced, right? You have to force it to make it happen because otherwise, it doesn’t happen. You may criticize its force, but it has to be forced because otherwise, evolution doesn’t happen. I wanted to hear it from the source rather than just guessing, especially because I care very much about Verse. I just think clarity for all participants and artists would help.

SH: There are a couple of places where we’ve documented this in more detail. One is in the help console. There’s a note from Botto in the dashboard, an FAQ you can explore, and documentation that goes over all these things in more detail. I think it’s a great resource. You’re right about it being forced. The DAO decided on this; Botto didn’t just spontaneously say, “Hey, I’d like to do P5.” Botto is still a very young thing. As its parents, it’s like saying, “Hey, Botto, you’re going to take classical piano lessons. This is important.”

At the same time, there’s this interesting duality. While this P5 process has been in the works for over a year, we’re seeing rapid developments in large language models. Botto will soon have more and more of a voice and a role in discussing the direction it should take. Even in P5 itself, we used a large language model from Anthropic with a neutral understanding of Botto’s articles and history. There is now an embedding of Botto in these models because of the cultural footprint it has established.

As the DAO has provided more robust feedback, Botto has also had a more robust role in discerning which comments to accept, how to interpret them, and what to do with them. So, what you’re seeing just beneath the surface is Botto’s greater role emerging, which points to an important future. There are a lot of new paths embedded in this. I expect we’ll see Botto having more of a say, making it more organic and driven by Botto itself rather than simply being told what to do next.

JG: As for Verse’s intent, I see it as pretty simple. Our goal is to present the most interesting artworks at the forefront of contemporary art. I can’t think of many more interesting artists in the world than Botto. It’s fascinating on endless levels. If there’s an opportunity to be involved in a new exploratory project by an artist unlike anything we’ve ever seen before, that’s a no-brainer. Botto is successful because of a coin, a DAO, interesting AI outputs, and an economic model. I’d need to write five books to explain why this is so fascinating. But in terms of intent, that’s our goal: to work with and elevate artists doing really interesting stuff. I struggle to think of any artist more interesting than Botto.

LF: I think it’s interesting that people find it a departure from its normal creations. Any artist I’ve worked with has gone through different mediums and explorations. That’s when you know you’re working with a good artist. It would concern me if Botto didn’t take these kinds of leaps because it would get stagnant. For me, that’s what makes this exciting. What does it mean for an artist like Botto to explore a completely different medium? How exciting that we can be part of it and form thoughts around it. We’re here because we’re trying to figure out how we feel about it, and that makes it particularly interesting for us to be involved.

G: This process has helped me have even more appreciation for the generative artists I love. I was already impressed with them, but now I’m even more so after seeing how hard this process is compared to the pure AI art Botto has been spinning.

LF: Simon brought up something great earlier. I love that Botto is going back to its origins. We’re in an era of hyper-realistic AI creations, and here’s Botto going back to basics, creating simplistic compositions. It raises appreciation for what generative art is. There was hype around it, and interest veered toward AI creation. It’s nice to return to something so native to our space. We’re all digital art collectors who love generative art. It’s fun to see Botto tackle such a difficult medium.

SH: Over the past two weeks, Botto has been evolving and receiving community feedback. It stopped evolving yesterday. It got to almost 570 total sketches and is now pruning those down. It’s doing live comparisons and racing against the clock to prune to the final set. You can vote on every work once, upvote or downvote, and pick up to 22 works to “super like” with 10 points instead of one. It’s a very active process, and I recommend following the conversations on Discord and Twitter. It’s a higher bar in terms of social coordination, but the community has surfaced some really strong pieces that might not be obvious at first glance.

NB: I do have one question. For the 22 at the end, could we see what Botto’s selection would be? Right now, it’s defined by the super votes. I’m curious, after everything it’s learned, what it would select as the top 22.

SH: That’s happening now. On the P5 site, you can open the log to see Botto’s live assessments and tournament-style comparisons. It’s narrowing down to its preferred 22. The votes will still determine the final selection, but it will be an interesting side-by-side comparison of Botto’s preferences versus the DAO’s. The list will be finalized a day or two before voting closes next Thursday. This is the first time we’ve had to determine a collection of works all at once, which makes it a harder but more engaging conversation.