Why AI Needs to Stay Weird with Kate Compton | Approximately Correct Podcast

Speaker A: We had professional creatives try to use the adventure creator and end up saying like, uh, ah, I guess I'm just no good at this. And then we would hack literal 6 year olds making Pixar armies with. So it's like, okay, how did we go from totally winning it to totally losing it? With the same team, with the same tools, with the same everything.

Speaker B: Hey, welcome back to Approximately Correct. I'm Alana Fish.

Speaker C: And I'm Scott Lowell.

Speaker B: And this time on the pod, we have Kate Compton.

Speaker C: Yes.

Speaker B: Which we were so lucky to get her. And it was such a cool conversation.

Speaker C: Yeah. She was here for a conference on digital entertainment and artificial intelligence and we knew we wanted to talk to her because she's just fascinating.

Speaker B: She's so engaging and also so excited about creativity.

Speaker C: Yeah. So, uh, she also, she lives in Denmark, so we did a virtual one for this one. Uh, very different time zones. We're coming with morning energy. She's got evening energy.

Speaker B: She matched us though.

Speaker C: Yeah. I think she exceeded our energy.

Speaker B: She made it there. Yeah.

Speaker C: But I mean, Kate is a person who didn't get her start in AI. She was working with video games and a lot of like creativity type stuff. But as things went on, she's become more in the space and she has some really interesting thoughts on creative culture, the importance of remixing, and how that is even more important now in the age of artificial intelligence and creativity.

Speaker B: Yeah. She said that she's been working on the same thing, but things have been moving around her in different ways. So she's been thinking about creativity, how to get people to be creative, what are the things that unlock creativity. And as she's been doing that, things have changed and now AI is such a big part of our culture and of course that has intersected with what she's interested in.

Speaker C: Yeah. So it gives her a really interesting, unique perspective.

Speaker B: Yeah. Here's our conversation with Kate Compton. Kate, thank you so much for joining us today on the podcast.

Speaker C: Yeah, it's really great to have you here virtually.

Speaker A: Thank you. I'm really excited to be here.

Speaker C: You were visiting Edmonton last year and I, uh, was reading up on you and I'm like, oh, we need to talk to Kate. She sounds really interesting. So you work in a, in kind of a different. Most people may not think about it in kind of the tech space. Can you talk a little bit about what your work entails and, and why you find it so interesting?

Speaker A: Sure. I've really taken a pretty, uh, to me it seems like I'm not moving at all, but the fields around me are kind of changing. So I've been in I guess kind of video games for a reasonably long time. So uh, I wrote my undergraduate um, dissertation on um, performative gender in video games. So how do you represent gender? Kind of from a procedural level, so actually in the simulation. Um, and then from there I went to grad school at Georgia Tech. And um, that was really the first kind of procedural content generation focused program which was um, my favorite definition of procedural content generation is make uh, stuff that makes stuff. So little programs that come alive and make strange things. This is like, well this is a solid 20 years before all of the uh, artificial intelligence genai stuff that we know now. So this is kind of an older fashioned version of it. Uh, but I really fell in love with you. Like, oh man. Like I can get computers to make weird, weird and strange things and then kind of show them to people. And from that I got an internship and a full time job at Maxis working on Spore, which was the first big kind of procedural content making stuff that makes stuff game.

Speaker C: Um, so when you say like computers make weird stuff, like what kind of weird? Show me the weird. What are we talking about?

Speaker A: So this is a set of kind of patterns that I've used AI to make. I made this in about 2012, layering a bunch of repeated patterns and then repeating those patterns, ends up making a pattern that um, people have described as having eye feel. I feel, yeah, that was my m favorite description of it. When I first made these I made uh, 54 of them and I had them hanging up around in a circle in my apartment and it just kind of gave your eyes asmr being surrounded by all these delicate little slightly alien patterns.

Speaker C: So we mentioned a little bit that you were uh, here in Edmonton in November, you were talking at Artificial Intelligence and Interactive Digital Entertainment Conference. Bit of a mouthful. And yeah, your talk I loved. I got to see it afterwards where you were talking about the way that people create and how remixing things is such an important part of creation. So what inspired you to give that talk?

Speaker A: A lot of things. So one of them is kind of my past research is on casual creativity which um, is a term that I coined for. So a lot of academics will study creativity and they'll be like, ah, ah, how do you make the next Mozart? How do you make the next Steve Jobs? Kind of the big money, big history creativity. I'm really interested in uh, kind of small everyday creativity. So people just creating for fun, doing weird doodles, doing zentangles, doing Sketching, making terrible music, um, listening to wind chimes, um, all sorts of things where it's not that you're making the biggest, best thing, but you're enjoying making it. I'd spent a long time studying that. I was really interested in how often that kind of casual creativity ends up happening socially. You think about quitting quilting bees, but, um, here in Denmark, people are very club focused. And, uh, so there's clubs for painting miniatures, and there's clubs for knitting, and there's clubs for making tiny dollhouse furniture. So this kind of. All right, maybe if you're painting a warhammer mini, it's not that you're making the single greatest warhammer mini, but there's something about getting together with a group of buds and kind of talking about what paint colors you're using.

Speaker C: Yeah.

Speaker A: So I'm really interested in, like, okay, that's wonderful. Like a really great form of, uh, kind of being together and being together in a way that you don't feel like you have to perform, which, um, I think stops a lot of people from wanting to be social is the feeling that you should be good enough. You need to come to this party to win. Um, you know, if you're going to make a TikTok channel, you have to like, like, be good enough to show off to others. Um, or at least extreme enough. Um, so really, how do you get people to overcome their fear of being social and of being creative at the same time? And so I'm really interested in kind of because it's been done before in some really great ways. Like both quilting bees and the early Internet show us that, yeah, people love to make weird little things together. Um, how can we start making that happen intentionally? And how can we kind of looking as broadly as possible across, again, technology, like the Internet, uh, social things like quilting bees, weird art installations that people have made, uh, games that unlock casual creativity. Uh, look across all of these and think, okay, are there common patterns that then if I said, oh, I want, uh, my local art space to have something that brings the community, to get the community together to be creative, Um, I could intentionally think, ah, these, ah, are the patterns that I'm going to use. I'm going to have a reasonable expectation that I can make, uh, casual community creativity happen.

Speaker B: So what are some of the patterns that you noticed?

Speaker A: Again, it's kind of this getting over the fear of not being good enough. Yeah, there's a thing that you can add as a community, um, but your, your individual contribution is quite small. Um, Together as a community. You're making a big happy pile. Uh, so there's a great one by Yayoi Kusama and I'm forgetting the name of it. I want to say, um, Abuishen piece. Um, but what it is, it's a big, a big blank gallery room. And then you have those little like round colorful stickers that you might put on at a tag sale.

Speaker B: Right.

Speaker A: And anyone could take one of those stickers or many of those stickers and start putting it all over the room. So you get a room that is covered in colorful stickers, um, and there's more of them at about like 3ft high because that's where kids can reach.

Speaker B: Yeah.

Speaker A: So you start seeing like this kind of like accumulated trail and you can always, you have a little bit of a choice, like, okay, what color do I want to put on? Um, but then, uh, not enough of a choice that you feel, you never feel like you were doing this room wrong.

Speaker B: Right.

Speaker A: Um, there's times where you, if somebody invited you to draw something in the SF MoMA, you would feel like, ah, uh, I'm on stage. But you don't feel like you have to perform here. So I think it's that I'm doing something together and I'm performing as a group.

Speaker B: Right. And my contribution is something. But it's.

Speaker C: Mhm.

Speaker B: Yeah. Maybe it's less scary because it's part of a bigger hole.

Speaker A: You always know that you're enough.

Speaker B: Yeah.

Speaker A: I think of this like also like when you go to a big sports arena and you're wearing your team's color, you know, there's definitely guys who will strip down in midwinter with the full body paint and everything. Yeah, that's fine. You can like, you can go. Yes. And go higher up. Um, but you can. And also just like, hey, by coming here to watch my sports team and wearing blue, I'm doing enough. I'm a part of things. I'm enough. I'm participating and I'm helping out my team.

Speaker B: Right. And I'm part of the whole. I'm like a participant.

Speaker A: Yeah, yeah.

Speaker B: So one of the things you talked about in that talk was the idea of these m of mixtapes and remixing things as a form of creativity. Do you want to just tell us a little bit about that?

Speaker A: Yeah. So what mixtapes are for is, um, kind of like with the uh, sticker exhibition, um, a mixtape, the amount of choices that you have to make and kind of the heaviness of the choices, um, is a little lighter. So If I start with a blank piece of paper, um, it's like I could be drawing anything. There's too many choices here. But, um, if I'm like, okay, I'm going to make a mixtape for a friend, I kind of have a list of songs, and then it's like, okay, I'm just picking one from this and I'm putting it over there and I'm picking. Maybe I'll pick something that's like, a little funkier. Uh, or maybe I'll pick something that's like, a little weirder, but just kind of like picking from a list and then figuring out where it goes in that list. Um, I think that's. That's something that people feel very comfortable with. And that's why I often talk about mixtapes and collage in kind of the same breath. Because if you're doing collage likewise, it's not like, oh, what am I going to make? But, aha, ah, I found a cool picture. Um, I'm going to cut out that picture, I'm going to put it somewhere on my sheet, and then I'm going to look for other pictures that I think go with it. So each time you're making a much smaller choice, and you're choosing, uh, from a bunch of stuff that already feels complete in itself. So I'm not like, is my mark good enough? Am I drawing good enough on this? M kind of smushing together things that are already good enough to make something. That curation is a little bit more than the sum of its parts.

Speaker B: So one of the things that students say when I ask them about whether or not AI is intelligent or whether or not AI could be creative is they say AI is just mixing things. AI is just choosing from what it has and putting it together. And that's not creative. But I think what you just said was taking things and putting them together is creative. So how do you. How do you figure those two out together?

Speaker A: Right. Um, I generally don't think about computers as being creative or not being creative, because this was a long thing in a lot of the computational creativity community for the past 20 years. Is, can computers be creative? Is this algorithm being creative? When I made a bunch of particle systems that kind of drew the scarf, were they being creative? Um, and so my response was, um, computers lack two things that I care about. Feelings and money. And, uh, so I don't actually care about whether or not they feel creative. Um, creativity isn't a thing. Creativity is a great thing socially. Uh, it's a great thing for Us to feel. And especially when I'm talking about casual creativity. This is about me feeling proud of what I've made. Me, um, making something while I'm thinking about a friend. This kind of creativity while I'm thinking about you. Uh, it doesn't matter that I'm just putting things together. If I have that feeling of thinking about you and feeling proud of what I've made. Now the computer's not thinking about you. The computer's not proud of what it's made. Um, it's great that it's putting things together, and it can be a part of my casual creativity process. We've all had the awkward friend who sends us a computer generated poem and, uh, is very proud of it, or a computer generated image. And they're very proud of it. And often they'll have put something about us in there. It's like, hey, it's like, yeah, they're using the AI to make a mixtape for you. Um, but it's not that the computer made a mixtape for you, and it's not that. Is Lana Del Rey creative because she made the thing that is part of my mixtape.

Speaker B: Right.

Speaker A: Um, not really part of the question there.

Speaker C: Okay, so that intention is a big

Speaker B: part of the intention.

Speaker A: Yeah. Like creativity to create is a verb. Um, I like to think of art as a verb. So especially casual creativity is about the process rather than about, like, oh, if I make you a mixtape, should that mixtape be in a museum? No, because that mixtape, kind of divorced from our relationship, is not worth anything.

Speaker C: So you have kind of talked in the past. You kind of draw in a contrast between the AI systems that we have now, a lot of which are a bit more of a closed system. And then, I mean, earlier AI systems, smaller personal things that people there was a lot more. It seemed kind of like collaborative, making things, just putting them in the world, sharing them. Uh, so I guess, what. What was it like back in the old days to give us a sense for those who might have missed that sounds like a pretty wonderful time.

Speaker A: Yeah. So let me take you to the magical world of 2014. You had a Twitter account and you felt good about it, right? Yeah.

Speaker C: Seems so long ago.

Speaker A: Yep. And so, all right, you had a Twitter account, um, and you, uh, wanted to make a, ah, weird little art bot for your friend. So you wanted, you know how in Wall E, um, the humans have left Wall E behind and he keeps cleaning up the trash long after all the humans have gone. So there's like a little helpful part of humanity that's still Trucking away, doing nice stuff for the world, uh, even though the world is not paying attention anymore. So I think I want to make a bot for a friend. Why do I want to make a bot for a friend? Um, one is that just the urge to make something that makes something is. I want to have something that comes up with new podcast ideas. Um, because after probably a few years of running a podcast, it's hard to think of new conversations. All right, I'm going to make a little bot that says, uh, I'm going to take InterestingPerson and ask them about topic. Um, so it's like, okay, I'm going to have my list of interesting people. Um, so all of your local Alberta celebrities and anybody who's coming in, and maybe a couple funny celebrities. So it's like, okay, I'm going to take Abraham Lincoln in and ask him about casual creativity. And I'm going to ask Lana Del Rey to come in, and I'm going to ask her about, uh, whether computers can be creative. And then it's like, okay, that's starting to make a bunch of weird little things that are sometimes funny. Or maybe I'll ask them about their favorite smoothies. Um, or I'll even add both. It's like, hashtag big topic and silly topic. So I'm going to ask Lana Del Rey about computational creativity and her favorite smoothies and what they have in common, right?

Speaker B: And it's like when we have those mixtures, then the person is bringing the intent. The person sees the mixture and creates the intent. Intent that.

Speaker A: Yeah, There's a fun academic term called computational reading, which is saying, I'm going to look at a field that I deeply love, and I'm going to figure out how to fit that into my little simulation. So if you deeply love podcasts, or you deeply love mixed, uh, hipster cocktails, or you deeply love Taylor Swift genre music, breaking that apart and saying, uh, ah, I think I see what this is really made out of is this, this, and this. And so then you make that for your friend. You put it online. Uh, you use a service called cheapbots. Done quick. Um, you write it in this language called tracery that I made, which is just like a very simple, um, kind of Mad Libs for computers, where it's like, okay, here's my list of people. Here's my list of silly topics. Here's my list of deep topics. If you sent this to a friend, maybe they're the only person that follows this bot. This bot only has two followers. The two of you but every morning you get to wake up and see what your bot has made for your friend. And maybe a couple of other people, like other podcasters start following it and they think it's funny. Um, this eventually led to, um, the runner of it, um, of cheap bots and Click V. Buckingham said that by the end of its life in 2000, what is it, 21, where Elon Musk turned off all the bots. There were 200,000 bots of all time that had been created with this service. So 200 weird little messages sent between people, 200,000 weird little mixtapes from person to person. Um, and that was really beautiful.

Speaker C: And I guess this seems like a question that we're asking all the time in 2026 about everything. But like, what changed? Yeah, where'd that go?

Speaker B: Yeah, right. Things seem different now.

Speaker A: Mhm. I mean, part of it is I get really into infrastructure, which is like, how so you know, you're a person who wants to make a little podcast bot for your friend. Um, you're probably not going to figure out how to run your own Twitter bot. And you're probably not going to figure out how to run your own Twitter. So when Twitter said, okay, no more bots, suddenly people started this wonderful big 200,000 person bot making community said, shoot, all of our bots died. Um, I miss my bot. Where should my bot go? I missed bot o'. Clock. The sound is so empty at 6pm with nothing having podcasting opinions. Um, except for humans.

Speaker B: Right, Boring.

Speaker A: So they started moving them to Mastodon because, um, the cheap Boston quick. So both Tracery and cheapbots done quick were small, simple and portable enough that kind of the most technical person in your friend group could figure out how to run their own. So somebody figured out how to run it on Mastodon. Somebody figured out how to run it on bluesky. So now there's a whole bunch of services kind of doing the same thing for other social networks. Um, and that was because, okay, Tracery, which was the language that ran these, was something that I wrote as somebody who had just learned JavaScript. So this was not a big complicated program. Um, I open sourced it, but, uh, kind of with the assumption that if anybody wanted to do anything with this, they're more than welcome to. And it got poured into 16 different languages and I used it to teach undergrads because if you've learned a little bit of JavaScript or a little bit of any language, you can, you can implement your own Tracery. It's like, okay, podcast inviting hashtag person to Talk about hashtag deep thing and hashtag not so deep thing. Many people can kind of break that apart and then like pick a random, uh, pick one of these random objects. So a lot of people could. A lot of people can pick this up and run it themselves. Um, so I think it's that. Pick it up and run it themselves is. And again, I often think about this as kind of. It's like music. Um, I've made something so small and so catchy that anybody feels like they can hum their own version.

Speaker C: So it hasn't completely disappeared. It's just maybe not as easy to find as it used to be.

Speaker A: Yeah.

Speaker B: Another thing I've noticed in the evolution that was 2014-2026 is that AI has also changed because it's mostly become a thing that requires a huge amount of money to train. The data is often closed, it's owned by these big companies. And so what kind of little creative person do these days? What can you make with AI? Has that also changed?

Speaker A: It's less that anything else has vanished. So all these tools like Traceray is still available and all the weird little, um, weird smaller versions of AI that we used to mess around with beforehand, those all still exist. And in fact you can use big, uh, gen AI to talk you through how to use any of these.

Speaker B: That's amazing.

Speaker A: It's in fact more accessible than ever if you're like, hey, um, hey, LLM of my choice. Can you make me something that runs Tracery scripts on Tumblr? And, uh, will do you a passable job of it so you don't even have to kind of go through the lightweight technical lifting that we used to. So it's kind of like we used to have these lovely little walking paths, uh, and we walked everywhere and it was slow going, but we loved the scenery and somebody built a giant superhighway through. And now we have lots and lots and lots of people going on the superhighway, um, getting to where they're going mostly, uh, not looking at the scenery. Um, but that doesn't actually stop us from kind of continuing the side paths. It's that we're mostly regretting that there's so many people who are coming through so fast that they're not getting to experience the little side paths, they're not getting to experience the little gardens, they're just kind of like rushing through. Um, but it hasn't actually stopped all the little side paths from existing. It's mostly a wistfulness of gosh, I miss when mostly we were just little side paths and we didn't kind of hear the, the overwhelming rushing of the traffic going past us.

Speaker C: Yeah. And there was more people to run into on the side paths and have conversations with. How do you, how do you lure them back?

Speaker A: This is that sort of casual creativity. It's look for the places where people are doing weird little creative, like weird little creative things for each other. Um, one of the big ones is Roblox. Kids are making just extremely weird little love letters via, ah, strange obstacle courses to each other and to strangers. Um, this is great. One of my friends had a lovely birthday party where, uh, their birthday party, they took a bunch of friends into Roblox, um, and they would only visit abandoned uh, birthday sites from other Roblox users. And so it's just a tour of like welcome to Timmy's seventh birthday. And just kind of this frozen in time moment of some kid building for their friends exactly what they thought the ultimate birthday zone was and just kind of jumping from abandoned birthday party to abandoned birthday party.

Speaker C: Oh, like a weird Roblox anthropology tour.

Speaker B: Yeah, People love to hate on Roblox, uh, in Minecraft. But I do see also the casual creativity that comes from that. Also that it's building there is pretty easy, much easier than creating something in the real world. Um, and so it does open up possibilities. And I've seen that in kids. Like you know, some of them are interested in playing the zombie game, but some of them are interested in creating a building and that's what they're excited about.

Speaker C: Yeah, I spend a lot of uh, time over Christmas, uh, playing Minecraft with my nephew and yeah, just like the creativity that he shows and like the weird, I mean he's a weird kid. I love he's a weird kid.

Speaker B: I don't know where that comes from. It's weird that somebody in your family would be weird.

Speaker C: It's not genetic.

Speaker A: All kids are weird little grandpas. I can probably say.

Speaker B: I know. It's too bad we convinced them to stop being weird.

Speaker C: Yeah, it's a shame we lose that.

Speaker B: Yeah.

Speaker C: Uh, so you mentioned weird little creations and uh, I did want to talk because one of the projects you talked on, you talked about it is you had worked on the uh, character creator for Spore. It's my understanding and I mean for people who aren't familiar, Spore was a video game about evolution, uh, and uh, creating new creatures and the creature. I mean I personally skipped a lot of classes in my first year of university playing with the creature creator in Spore. So I'd just love to hear about how you came to work on that project and what that was like creating this tool that other people could create things with.

Speaker A: Just really amazing. That was what I did a lot of at, uh, Georgia Tech when I was there. Um, and then we had this cool speaker come in, Will Wright, who had this new video game that he was pitching called Spore, which was, hey, what if we use procedural content generation, which again, there was almost nobody doing at the time. Um, what if we use it for a whole video game that's funded by Electronic Arts?

Speaker B: Ah.

Speaker A: So I was really excited. Um, I managed to get an internship on it where I was doing special effects for generating, um, uh, Interstellar Nebula. So how could I make a little recipe for generating H2 nebula and stellar Nebula and all these kinds of different phenomena? Um, so then I went back for a full time gig. Um, I actually never worked on the creature creator itself. Uh, I worked on the planet generation.

Speaker C: Okay.

Speaker A: Um, so I made a lot of planets. Uh, but I also saw a lot of the people using the creature creator. And that was, it was really interesting to see how much kind of pride and ease people felt making these creatures. We would often get wonderful situations where, um, a kid would come in wanting to make Pikachu. You come in wanting to make Pikachu. It does not make Pikachu. Well, but very soon you've got like, branching arms and you are making something very far away from Pikachu into the monsterverse. And you're really proud of it and share it with all your friends. And then you're like, okay, now I'm going to make Pikachu. And you're immediately back into the monster space. You never know where you're going to get to, but you're always happy you got there. Uh, and that was a really beautiful thing to watch kids and, uh, adults do. And we got some handwritten letters of people saying, hey, I'd written myself off creatively. Um, but this game, specifically the creature creator, made me feel creative again. So I thought that was really fascinating. Um, but a bunch of our, we had a whole lot of other editors in the game. So in our expansion you could edit the planets. Um, so using the same tools that I used to make the original planets, you could make games. Um, you could make spaceships, you could make all sorts of stuff, but people didn't feel the same intensity of, I'm proud of this. And in fact, our, our adventure creator, which should have been a smash hit, right? It's okay. Now I can use Now I can use all my weird little creatures and I can have them, I can make a video game around them. So very rudimentary Roblox kind of situation. Except that with the creatures, everything you made felt like a joyful, wonderful, perfect monster. In the game editor, everything you made was broken. So it's like, uh, oh, I'm falling through the landscape, I can't get to the exit. I like, I've spent an hour making this thing only to test it out and it's not even playable. So many people never even like play tested their, their levels.

Speaker B: So it's like the zone of proximal development. Yeah, it's like you can't be too hard, but it can't be too easy. Like there has to be some challenge to it, but also not too hard that you fall through the, the level.

Speaker A: Yeah, yeah. Like we were, we, we had a hard time even getting people to make levels that you could like walk through and get to the first if it was like walk through the level and pick up an object. That was often out of scope for most of the people who are making dozens of characters. Uh, so I found that just like fascinating. Like, oh my God, with the same team, how did we make something that was so different? It had such an effect on people that like we had professional creatives try to use the creature, uh, try to use the uh, adventure creator and end up saying like, uh, I guess I'm just no good at this. And then we would have these literal six year olds making Pixar armies. So it's like, okay, how did we go from totally winning it to totally losing it with the same team, with the same tools, with the same everything. And that was what I ended up trying to, uh, getting my dissertation on was, um, not only for those editors, but, uh, for every kind of casual creator trying to find the patterns of what makes some of them really great and what makes some of them really just dreadful.

Speaker B: And so beyond just like, this is too hard. Are there other patterns you saw that make it.

Speaker A: Um. Uh, the term that I coined in my dissertation was the Grok loop, which is, um, how long it takes you from having an idea to trying it out, to understanding how well that went. Um, this was before Grok had weirder conversations.

Speaker B: Not named after Elon Musk's.

Speaker A: Um, but the idea was that this is, um, if I'm, I'm a six year old using the creature creator. And so I start snapping on an arm. Okay, so I snap on an arm, I have an idea that it's Going to look good here. It doesn't. I move it, it looks better. I move it, it looks better. Move it. Looks better. I move it. Looks worse, looks better, looks worse. So I'm doing that probably 20 times per second. So 20 times per second I'm trying something out and getting feedback on it. And that lets you learn amazing things. So having a really fast grog loop where I'm trying something out and I'm immediately getting to understand whether that that was any good makes you master at things so fast. Uh, and makes you feel totally in control. Um, and it makes you feel very creatively freeze because I'm like, ah, uh, I'm going to put the arm on the head. No, that looked terrible. That was a zero cost risk. Um, you know, so people take wild creative risks without thinking twice about it. Now if I'm making a level and that level takes me 60 minutes to build, so I have to lay out all the things 60 minutes in, I start playing through it. And even the process of seeing whether or not it's any good is so grindingly slow. How long does it take to play test a game? Um, just aching amounts of time. So now you've gone from a 0.1 second Grok loop to a 2 hour Grok loop and nobody's even going to get around that. Once you feel you're uncertain. Uncertain, uncertainty. Oh, you screwed up. You screwed up. You screwed up. Oh, heck with this. I'm not going around for a second time.

Speaker B: Exactly. You just spin right in.

Speaker A: So nobody takes creative risks. Um, I think we're seeing on things like Roblox where they've managed to nail that much faster. Notice how many obstacle courses there are in Roblox. And uh, that's because an obstacle course is something that you kind of can't screw up. The default of an obstacle course is to be there's an entrance and an exit. Congratulations, you've won at game design. It's not a good game design, but you have a successful game design and then you could start like adding stuff, but you kind of go from valid to maybe better. Uh, and that turns out keeps people kind of in for a long time. I had a whole chapter on my dissertation of just how hard game editors are because of these slow Grokli problem.

Speaker C: Have there been other lessons that you've learned through your work in creating these kind of tools that tell you what types of things you should look for in a creative tool, like a creative AI tool? What elements will help people use it in creative ways?

Speaker A: Yeah, I really Saw that expand out in kind of the social space. Um, like everyone else during the pandemic, I was really interested in how social stuff works, uh, how loneliness works, how do people create together? Um, so we were all trying to solve loneliness in our own way. Uh, so I think one of the things that made spore work really well was you could immediately share your creature.

Speaker C: Yeah, that seemed like a big part

Speaker A: of it, I think. Yeah. The two major axes of how do you make a, uh, community creative scene is you have to have ease so that any doofus can make a terrible thing. Like the early web. Nobody closed their tags. Nobody is writing valid HTML in the early web. It's fine. Yeah, um, an HTML page, you can smash your face into the keyboard and you get. And it'll render as some key smash text. Like there is very little you can do to break a web page.

Speaker C: Dad dancing baby gif. And you're done, you're good to go.

Speaker A: So anybody could make a web page that they felt was like enough that they wanted to show it off. And then, okay, on the early web we had a bunch of easy hosting services and we have, um, so we had GeoCities back then. There's sites like NeoCities now which are trying to do the same thing of. Anybody can post a weird little website. Um, but okay, I can make a doofus little thing. I can share my doofus little thing with my fellow doofuses. It's this kind of creativity as a connection between people. Uh, why am I making an NSYNC website? It's because my friends love NSync and I want to share it with them. Or I think that there's other people who need to know about NSync. So I'm going to make an NSYNC website and share it with them. So this sociability as a huge driver for why people create, uh, they really want to connect with other people. And it also has this wonderful thing of, okay, I see your NSync website, but I think that you really chose the wrong side, like the wrong song for your MIDI auto playing in 1997. And so I'm going to make my own NSYNC website with a better MIDI auto playing song or better pictures or pictures of the best boy in the band. So this kind of. Ah, yes. And so everybody has both an easy move to create, an easy move to share, an easy move to remix and to reshare. And then you get this wonderful engine where just people are building communities, people are trading back and forth. If one person discovers something Then flashes through, everybody else discovers something. We had a wonderful, um, we didn't ship with asymmetry on Spore, Uh, but somebody kind of tortured their creature enough that they managed to glitch it out and get an asymmetrical skeleton. They then shared that everybody else then immediately brought that into their creator. Uh, because we did this incredibly creative thing. And I really want to find out who on the team did this. But your characters on Spore were saved as a png, so you had your little image full of pixels. Uh, how do we save all the data with that? So you take all the transparent pixels around the border of it, and you use the color data of the transparent pixels to store all the data about your creature. That means that I can then have my creature in Spore, but then I can drag it out to my friend's forum, I can send it to my friend in an email, basically any place that you could put a png, then became a social network to share things back and forth. And that made that kind of engine go.

Speaker C: And I hadn't really thought about that. But when I'm thinking, like, AI tools that I have used, shareability doesn't really seem to be a, uh, consideration at all. Yeah, like, I can't think of any. I mean, like, you can always copy and paste. It's always, like, safe. But it seems like if it's there, it's an afterthought.

Speaker A: Yeah.

Speaker C: Or like a workout.

Speaker A: I've got a good one that I like that's called, um, Glyphosate. And Glyph is, uh, for making pipelines. So, like, okay, I want to make something that captions, uh, images to make little memes. It's like, okay, take an image, have, um, the AI detect something about that image, make up a funny caption, paste it in so you can write your kind of little script, and it'll tell you how many tokens this takes. So, like, okay, I now have a really good image captioner, and you could upload that and then other people can, uh, then edit it. So now instead of sharing our Spore creatures, we're sharing our AI pipelines. And this has caused kind of a nice little micro community there, where you now have some really funny ones. There's a wonderful talk by Darius Kazami called Platform is Shitpost, where he makes the argument that the speed of posting determines, uh, whether or not you can shitpost on any platform. And to shitpost is to post something so quickly that it is still funny by the time you finish posting. It.

Speaker C: Yeah.

Speaker A: So Infinite Scream was probably only funny for the five minutes that Nora Reed spent making it. But because they posted it that quickly, then Infinite Scream could kind of discover its life. Um, so one of the funniest ones on Glyph is somebody made, you know, like the Bayou tapestry memes, where it's like old timey medieval illustration and then there's like a medieval caption underneath it that's saying something funny about modern life.

Speaker C: Yes. Okay.

Speaker B: Yeah, I think I know what you're talking about.

Speaker A: There's a famous one that's like the hold the field where I sew my. It is barren.

Speaker B: I've seen that one.

Speaker C: Yes. Now I know what you're talking about.

Speaker A: Yeah, so somebody made one of those and it's like, okay, that's great, because that is not something that, like, OpenAI is not going to come up with. The, like, we've put a team of five PhDs on the biotapestry meme Generator, but somebody was able to shit post this. But now also because it's a social network, somebody can then say, oh, that's great. But you know what I love is like, 19th century French impressionist paintings. I want to do the French impressionist painting Meme Generator. And then you can copy that and then you can send it to a friend and be like, okay, you have to run. Like, we both love French impressionist paintings. You have to run the French impressionist meme generator. So again, it's that, like, ease of making, ease of sending, ease of remixing, and that starts the engine turning over.

Speaker B: And I think it also comes back to intent getting in the way of intent. That, like, my intent can, like, last long enough to come back around the Grok loop. And then, like, sharing it is also, like, it's easy for me to share my intent with you, the intent behind the idea.

Speaker C: One thing I was kind of curious about is when you're. You're kind of using these tools and doing the remixing and things like that, does that change the feeling of ownership that somebody has over. Over what's created?

Speaker A: That's a good question. I, um, think it's that there's so many different niches and so many different kinds of creativity that you let people self define their creativity. So if I make you a mixtape, I still feel ownership of that mixtape. It's a different ownership than if I had recorded you five terrible songs of my own. Um, but I still feel like when I give you the mixtape, I never say, Shania Twain made this. I say, I made this. For you. And there's always little moments of kind of ownership where I put my name and your name on it. I draw a little picture on the mixtape. These kinds of little like, okay, I'm declaring ownership by kind of marking this up in some way and putting a sticker on it. I, uh, think we see this a lot. Um, things, uh, like Pinterest boards. Um, I have not made anything on my Pinterest boards, but I feel like I've made my Pinterest boards. So this kind of curation, um, and I think we're seeing this a lot now, is these different. Another thing that makes these ecosystems work is having a curator role where somebody can say, ah, these are my top, uh, spore creatures. These are my top, um, glyph image captioning bots that I thought were really funny. These are my top Twitter bots. And so then I can start. Can you mixtape that? Uh, can somebody make an art gallery of it?

Speaker C: Yeah. Like, anybody could have done this with this tool, but I am the one who did it this specific way.

Speaker A: You need vanishingly little interactivity, uh, for somebody to feel like they've made a creative choice. Sometimes it's just. Sometimes it's just being able to leave a caption on something like retweets as a form of creativity is important. If you can only retweet without a caption, people wouldn't feel like they owned it. But if you can put a caption on it, suddenly you feel like you've owned your retweet.

Speaker B: Uh, that's my best Tweets are hilarious. Quote retweets. Right. That's some of the loveliness that comes out of social media.

Speaker C: Well, I mean, I think that's because. I mean, I think even if you look past technology, I mean, a lot of. I would say sometimes the best art that has been done has been our reaction to somebody else's.

Speaker B: Um, a remix, if you will.

Speaker C: Yeah, yeah, exactly.

Speaker A: Yeah, I don't enjoy it, but the whole, like, TikTok genre of somebody, like, emotionally reacting to another TikTok.

Speaker B: Right.

Speaker C: Yeah.

Speaker A: They 100% feel like they've created that TikTok. That's just them making extreme faces at somebody else's TikTok.

Speaker C: Yeah, it obviously connects with something because, yeah, like, they. They do get watched. And.

Speaker A: And again, it may be one of those things where this art on its own does, uh, nothing. This art, when it's made by your friend. Um, and you know what they're making a face about, because this is just like, your life Suddenly again, you've been given a mixtape and you found a pile of mixtapes in a thrift store, and you're like, I don't get what this genre was about.

Speaker B: Right. But it hits different if it's from somebody and you have some sort of shared history and you're sort of making a comment about that history that nobody else would get. It's sort of like an exclusivity thing too.

Speaker A: Yeah, yeah, yeah. If I. If I took all of your discord reacts and put them in a gallery, probably, uh, most people would not find it very compelling. But I love.

Speaker B: I would love it. Yeah, I would. I would walk through that for sure.

Speaker C: Yeah. It's been a great conversation, and I think we're kind of wrapping up here. But one thing I wanted to ask you is, like, are there any areas of creativity that you think are kind of unexplored with AI now that you think it might. Might have like, a really positive impact in the future?

Speaker A: Yeah, that's a good question. Like, what could AI unlock? That's not. And again, I would come back to what's hard now that you would never share.

Speaker B: Right. Uh, it's closing that grok loop.

Speaker A: Yeah, yeah. What would I never make a mixtape for my friends of now that AI might make easy. Um, I think we're seeing a little bit of video. Although video is not as much fun as I think many people hoped it would be. Um, so we can send each other little generated video snippets, and yet we don't see ourselves doing that.

Speaker C: Yeah, sure.

Speaker A: It's like, okay, there's something that is not, I don't know, acoustic enough about this, uh, for lack of a better word, like, there's something that, like, putting a sticker on top of a meme is enough. And yet making a lavishly rendered 22nd Star wars video is not. Um, I'd love to see, like, also people who aren't cool teens. So, you know, I loved Pokemon Go. Um, Pokemon Go was super fun. It made me, like, walk around towns, find different art that I had never explored before. Because you open up Pokemon Go, and it'll show you all the different, um, locations that people have tagged.

Speaker C: Yeah.

Speaker A: But I always wanted, like, it's like, okay, you know, I don't actually have any connection to the franchise. Um, there are these little monsters. But mostly what I. What I want to do is, like, walk around and see civic art. Um, so could somebody make this for, like, you know, elderly folks who don't get out much, um, to get them to you know, they know their town, they know what interesting stuff there is. Granddad really wants to tell you a story about this tree. Could you have something such that Granddad could shitpost something interesting about this part of his town? And everyone in the town gets to see what that is, and then maybe somebody else riffs off of it and Granddad sees that, ah, uh, not only did he get engaged at this tree, but these are all the other people who also got engaged at that tree or all the other moments that, you know, that this had. So what are the things that everyday people, kind of having everyday life experiences, wouldn't be able to share about their world? Is there any way that AI can help us connect with that and then get the heck out of the way? So it's not an experience about AI, it's just a tool to help you to help Granddad shitpost about trees.

Speaker C: Wonderful. Yeah. I wish that I had, uh, had spent more time hearing my grandfather shit post about trees. Thank you very much for joining us. It was. It was fantastic talking to you and getting a real. A real different perspective on all kinds of stuff. Creativity, shareability.

Speaker B: Yeah. I think it's been really energizing too, and makes me want to go out and do some more creative things. So I hope that our audience also feels that. Yeah, I hope they feel the intent.

Speaker C: Yeah. And if they do, I hope they share. Let us know. Like, yeah, right. Does this conversation inspire you? And you're like, I'm gonna go make this bot.

Speaker B: Right.

Speaker C: We'd love to see them.

Speaker B: Yeah.

Speaker A: The best advice I ever got was to, uh, is for a video game jam. It was to make a game as a love letter to another person. So.

Speaker C: Oh, uh, don't.

Speaker A: Don't pick a weird thing that you're going to make. Pick a person and then figure out what you would use to send the strangest love letter to them.

Speaker C: M. Okay. Wow. That's a great way.

Speaker B: Okay, I'm going to take that away.

Speaker C: Well, thank you very much, Kate. Uh, it's been great talking to you.

Speaker A: Likewise. Thank you so much.

Speaker B: I'm so glad we can make this work. I was so inspired by talking to Kate, and I want to go out and be more creative. It's actually one of my New year's resolutions for 2026 is to try to get some creativity in my life to create with my kids. They're so full of creative, kooky energy and capturing that and making art together is one of the things I want to do this year. So this was really invigorating yeah.

Speaker C: She's one of those people that when you talk to her, you're just like, I want to get me a paintbrush. And I just want to go, yeah.

Speaker B: Or like, yeah, just the. The love of kooky, creative. Whatever you. Whatever you got is what I want. I love that energy.

Speaker A: Yeah.

Speaker C: If you out there were also inspired.

Speaker B: Yeah.

Speaker C: I mean, let us know. Like, uh, did you create something after this conversation? What was it? Share it with us. You could find us on. On social media. You can send us an email.

Speaker B: Put in the comments.

Speaker C: Put in the comments. We would absolutely love to see what you're creating.

Speaker B: And, you know, I bet Kate would, too. I bet she'll drop into the comments if. If there's some cool creative ideas.

Speaker C: Oh, yeah, we'll definitely let her know.

Speaker B: Yeah. Or it turns out she's on Blue sky, so hit her up over there. Yeah.

Speaker C: Yeah. And, uh, you know, if you want to see more about what we're doing here on the podcast, the best way to do that is to subscribe. You can find us on Spotify, you can find us on Apple podcasts, or we also have a video of every episode up on YouTube.

Speaker B: Yeah. And if you're on YouTube, hit. Hit the like. And also leave us some comments. As we said, we love to hear what you think of our episodes in the Absol. So with that, I'm Alana Fish.

Speaker C: I'm Scott Lilwall.

Speaker B: And this has been approximately correct.