Episode 187: Artificial Fandom Intelligence
In Episode 187, “Artificial Fandom Intelligence,” Elizabeth and Flourish respond to a listener's letter about AI tools like ChatGPT and Lensa, which are currently sparking anxiety in fandom and across the broader web. What are the realities of this technology, now and in the coming months and years? Are we looking at a future where any fan can plug in a few terms and receive a halfway-decent computer-generated piece of fanfic or fanart?
Show Notes
[00:00:00] As always, our intro music is “Awel” by stefsax, used under a CC BY 3.0 license.
[00:00:52] The two big ones we’ll be discussing: ChatGPT and Lensa.
[00:01:56] “Disability and Fandom”: Part 1 and Part 2
[00:07:22] Our interstitial music throughout is “Sad Marimba planet” by Lee Rosevere, also used under a CC BY 3.0 license.
[00:10:32] ChatGPT has been temporarily banned at Stack Overflow: “Overall, because the average rate of getting correct answers from ChatGPT is too low, the posting of answers created by ChatGPT is substantially harmful to the site and to users who are asking or looking for correct answers.”
[00:10:50] If you AREN’T familiar with Nick’s work: nickm.com
[00:13:08] Some explainers on these tools:
And some criticism:
[00:19:03] Flourish, Lensa-fied:
[00:27:11] There’s been a fair amount of reporting and commentary about the problematic ways Lensa renders faces, including:
“The viral AI avatar app Lensa undressed me—without my consent”
“A Psychologist Explains Why Your ‘Hot AI Selfies’ Might Make You Feel Worse”
[00:28:56] The AI-generated art that made the news was Jason Allen’s “Theatre d’Opera Spatial,” first-prize recipient in the “Digital Arts / Digitally-Manipulated Photography” of the Colorado State Fair art competition. It looks…exactly like you’d think something entitled “Theatre d’Opera Spatial” would look:
Respect to him for committing to the bit, though—as he told The Pueblo Chieftain: “I wanted to make a statement using artificial intelligence artwork. I feel like I accomplished that, and I’m not going to apologize for it.”
[00:34:42] DAMN IT the woman does wear the red sweater.
[00:36:02] One such piece about using AI to write a novel (with a pretty neat design and artwork!)
[00:36:08] NaNoGenMo actually stands for National Novel Generation Month.
[00:51:25] Casey’s breakdown of the GPT-3-fanfiction connection, with great additional commentary from marta-bee.
[00:53:00] On the University of Toronto researchers who analyzed Christie’s work.
[00:54:06] Bro??
[00:58:20] As you can from this episode’s cover—the creatively titled “DALL·E 2022-12-07 13.57.31 - Professor X (played by James McAvoy) and Magneto (played by Michael Fassbender) kissing on a bed near a conflagration, photorealistic,” fanartists…should perhaps not feel threatened by the current capabilities of these tools lol. A few other majestic images Flourish generated for this prompt:
Who is that third person?? Gentlemen, what happened here??????
Transcript
[Intro music]
Flourish Klink: Hi, Elizabeth!
Elizabeth Minkel: Hi, Flourish!
FK: And welcome to Fansplaining, the podcast by, for, and about fandom!
ELM: This is episode #187, “Artificial Fandom Intelligence.”
FK: OK, so, many of us may have observed that right now there has been a lot of discussion about AI-generated art, meaning both, like, visual art, and also text with—I guess ChatGPT is one of the big things that’s driving people talking about this in text, and the Lensa app, which uses Stable Diffusion in visual art. And we’ve noticed that there’s a lot of people talking about especially the text stuff in fan spaces, but the art stuff too. So it seemed like a good time to start discussing this issue.
ELM: Yeah, the out-of-context, [laughs] inaccurate, fearmongering strapline you may have seen is, like, “Elon Musk is scraping the AO3.” [FK laughs] Which, you know, I—look, I’m not gonna say he has better things to do, but—
FK: Also, like—never mind. I’m not even gonna start getting into this, because we actually have an entire episode to get into it with. [laughs]
ELM: Right. So yes, we—it was already percolating, and then we received a letter about it, so we will read that letter after the break. But first, we got so many incredible responses to the “Disability and Fandom” episodes. Thank you again to all our participants, and we got a voicemail that we wanted to play. So should we just get right to it?
FK: Let’s do it!
Diana: Hi, Flourish and Elizabeth! My name is Diana, and I am calling because I just finished listening to y’all’s fandom and disability series, and first of all, I wanted to extend a huge thank you to you and your guests for covering this topic and putting it out there in people’s minds. I unfortunately find that disability and all the ableism and accessibility issues that come with it are almost always in their awareness stage of their campaigns, and they can’t really progress. Progress is so slow. So thank you so much for the conversation, and thank you so, so much to your guests for sharing their experiences and their knowledge and the work that they’re doing, you know, inside disability and on behalf of it.
One thing that I did want to add to the conversation, you know, for people to keep in their minds—and your guests had definitely touched on it, but I kind of wanted to make it explicit and kind of underline it—is that making things truly and thoroughly and meaningfully accessible is a tremendous amount of work. And I just want people to really appreciate what goes into making something truly accessible.
And I say this because I’m coming to you from the accessibility services world. So, specifically, I am a professional audio describer. And the inadequate but quick-and-dirty explanation of that is, audio description is a little bit like your alt text, but for your TV shows and your movies. We’re getting into video games, you know, a comic book can be audio-described. Your museum tours can be audio-described. Your live shows can be audio-described.
And like your guests had mentioned, there is—or lived, not even just mentioned, but lived—there is a real dismissal of not just accessibility itself, but the amount of work that goes into making things more than just bare-minimum accessible. There is a lot of work that goes into making something actually functional and good. It’s not volunteer work, and it’s not an afterthought. It’s not just one unpaid person in an organization. It takes a lot of time, and it takes a lot of planning and dedication and skill.
And, you know, particularly because we are talking about fandom, you know, which is entertainment and art and people having these really personal and emotional experiences in relation to these things, and it’s such a travesty that we would deny anyone from having such human and meaningful experiences. Main point being, I’d love for everyone to really value the ways in which accessibility changes people’s lives, and this is really important work.
You know, if anybody’s listening to this, really, and you ever find yourself in a position of influence, please really, really value the work that goes into accessibility, and all of the different ways that you have to work to make something accessible. So I’ll wrap it up. That’s me on my soapbox. Once more, thank you all so much for the episode, and I’m looking forward to the next one! Bye.
FK: Thank you so much, Diana, that was such—I mean, that’s a really great point, and I think I really value you raising that up as something that needs to be paid and prioritized. And I think it’s—I guess I’ve heard from people a lot of, “Well, it’s just too hard.” And I feel like it’s really, really important for people to not let that be the end of the conversation, because there’s so many things that people do value that take time and effort that aren’t accessibility, and it’s really a shame.
ELM: Yeah. Absolutely agree. And Diana, I really appreciate this. Also, you know, exciting to hear from a person involved in a different kind of accessibility. You know, the descriptions that you can put on, of the movie, describing the action, or whatever.
FK: Yeah! You know, the other day, I accidentally turned that on for the first time ever, and I thought it was really fascinating. I mean, it’s something that I knew existed but I didn’t understand exactly what it was, and I think—I mean, I want to familiarize myself more with what that is, and I hope other people will think about it too.
ELM: I’ve actually encountered it in, like, public spaces. You know how they always have closed captioning on in, like, TVs at the airport or whatever, you know what I mean?
FK: Right.
ELM: I’ve actually encountered that in some situations like that, too, which I find very interesting. And I mean, of course, that’s all, you know, the curb-cut effect thing, too, there, to say there’s so many reasons why you might want something described or you might want it captioned for everyone, you know? So, it’s interesting.
FK: Yeah. Absolutely. I mean, I was actually thinking about how, like, sometimes I have, like, the TV on—I mean, as a person who’s not disabled in that way, but I have the TV on, and I’m doing work around the house and I’m not looking at it sometimes. And for a lot of things, I would want to really pay attention to, like, you know, whatever’s on there, but I don’t know. I was like, “Oh, actually, this could be cool. This could be something that I could be using so that I’m looking at something else while I’m—” You know?
ELM: Yeah yeah yeah. It’s interesting.
FK: It hadn’t occurred to me before.
ELM: All right. Well, Diana, thank you again. Really appreciate it. Anyone else, keep ’em coming. Voicemails, we love voicemails.
FK: Yeah!
ELM: And letters. Disabled fans want to share responses to the episodes or just, you know, bring up additional things, we’d love to hear from you.
FK: Cool. Well, should we take a short break and then get on to the main part of the episode?
ELM: Yeah, let’s do it.
[Interstitial music]
FK: All right, we’re back! Which means that, as always, it’s time to talk about how we fund this podcast, including how we fund stuff like our transcriptionists, which, you know, as Diana has pointed out, is a lot of work and takes, you know, money and support to make happen.
ELM: [laughs] Yes. Long setup to say, Patreon.com/Fansplaining. [FK laughs] Everyone I think at this point probably knows how this works. Maybe not. But there are a lot of levels you can pledge at. As little as $1 a month, as much as infinite dollars per month, or infinite euros, or Australian dollars. I really, actually, like—I don’t know why, but I find it so delightful, now that Patreon has allowed people to pledge in their own currencies, and I see people changing it, and I’m like, “Oh! [FK laughs] Oh, I didn’t know you were from New Zealand!” You know? [laughs] I don’t know why I’m so charmed by this.
Yes. Anyway, at different levels, you get access to different rewards. $5 a month, we’ll send you a very cute enamel pin of a fan in the mail, for example. And at $3 a month, as we’ve plugging in the last few episodes, but I think it’s worth mentioning again, you get all of our special episodes. There are 27 of them. The most recent of which is about our beloved Interview with the Vampire. And, you know, we’re gonna be doing more content around vampires.
FK: Yeah!
ELM: On the special episodes. We’re gonna be talking about my new favorite writer, [laughing] Anne Rice.
FK: I truly—it has been—you have been on a journey, [ELM laughing] and I love that I get to accompany you on this journey, Elizabeth. She’s giddy. She’s giddy with this fandom.
ELM: I’m just flying you around the world, like the Queen of the Damned.
FK: All right. And if you don’t have any money or don’t feel like giving us your money, you can also support us by sharing the news that the podcast exists, hurray! Especially also sharing that we have transcripts, which we know a lot of people really like. And you can write in, fansplaining at gmail.com. You can put an ask in our ask box at fansplaining.tumblr.com. There’s also a form on our website, fansplaining.com. Or you can send us a voicemail, 1-401-526-FANS. We really, really love hearing from you, and that’s how we make so much of this podcast.
ELM: All right. Business concluded?
FK: Business concluded! With that, I think that you were gonna read the letter that kicked off the idea of doing this episode.
ELM: Yes. This is from our friend Steve. [laughs] Steve, I just called you our friend. I hope that’s OK. [FK laughs] So Steve writes:
“Hi, Fansplaining!
“I’m in a bit of a spiral right now, reading a lot about these seismic advances in AI content generation and the impact on creative fields.
“It seems fairly conceivable that we aren’t too far away from a fanfic generator where you’d plug in your ship and trope modifiers and have a quote ‘unique’ fic in a matter of seconds. I feel like for a lot of people, that it seems like a dream, but I can’t help being horrified. Especially considering that the datasets that AI uses will probably be illegally scraped from sources like the AO3 or the entirety of human writing.
“I am conscious that there are positive use cases around the accessibility that these tools might offer, and of course nothing is stopping people from writing fic even with other AI tools, so it might be less disruptive than for professional writers.
“But does it change the practice of fandom, because will there still be the same impulse to write if these tools become so sophisticated that it’ll work on even the most niche prompts? Might fandom’s relationship to IP offer more protection than a romance of SFF writer?”
That is from Steve.
FK: Thank you so much, Steve. This is…I mean, it opens up so many questions. [laughs] So many things to talk about!
ELM: Hot topic!
FK: Hot topic!
ELM: For a little behind-the-scenes, I think we mentioned in the last few episodes that we had—we were all scheduled out til the end of the year, and the thing that was going to go into this slot has now been pushed back to January. Just, like, a little teaser. We’re not gonna say what it is. [ELM & FK laugh] And then we have our end-of-year episode, and so it was very fortuitous, because I feel like the kind of panic around this sprung out of nowhere within the last week.
FK: Yeah.
ELM: And not just within fandom, but academics, I think, in particular, and, you know, seeing people kind of just slamming the emergency brakes on, over at Stack Overflow, because people were, like, using these tools to generate actual bad responses to people’s code questions [FK laughs] and stuff like that. And they were like, “Stop. Stop doing this. This isn’t funny.” You know?
FK: Right. And it kind of blows my mind, because, I mean, as you know, Elizabeth. As people who listen to this might know, my partner works on, among other things, some kinds of text generation and so forth. And so a lot of this is stuff that—I mean, it’s been around for many years. But all of the sudden in the past week it’s like, [laughs] everyone figured out that it exists and that it’s easy to do it yourself, and it’s everywhere. It’s just like, [snaps] that.
ELM: Well, I think that the DALL·E art generator—
FK: Yeah.
ELM: That’s been, like, a heavy part of the culture for the last six months or so.
FK: That’s true. Right, right. When DALL·E came out, I was like, “Oh, wow, look at all these people who are using this tool that has existed for a while.” And now it’s like—
ELM: Yeah.
FK: You know, with these new—particularly, you know, with ChatGPT and with Lensa happening at the same time and becoming really popular and becoming really easy to use, right? Much easier than DALL·E, which required you to, like, join a Discord—
ELM: Right.
FK: You know what I mean?
ELM: Right.
FK: It still required a little bit more—it wasn’t just downloading an app from the App Store.
ELM: OK, let’s give a little bit of background for anyone who has literally zero idea of what we’re talking about. So ChatGPT, that is built on the GPT-3—
FK: It is built on GPT-3.
ELM: —language processing model.
FK: Right.
ELM: And that is owned by OpenAI, which was co-founded by Elon Musk, but Elon Musk has not been involved in that company for, I think, close to five years now.
FK: Right.
ELM: So just to kind of clear up where the “Elon Musk is stealing your fanfiction” came from. [FK laughs] Just such an absurd series of words together, there. But so OpenAI is a company that is really—began as a nonprofit with maybe loftier ideals than they ever—that they stated but never actually believed, and now is a for-profit company.
FK: Right.
ELM: So, you wanna explain the very—
FK: Yeah. [laughs]
ELM: —basics of how this kind of thing works? Not even just this specific one?
FK: Yeah yeah yeah. OK, so those of you who know details about how both GPT-3 and Stable Diffusion work, please forgive me, because this is gonna be, like, the most simplistic, like—
ELM: Quick and dirty. Don’t worry about it.
FK: —quick and dirty that you can possibly imagine. Please don’t yell at me for this. [both laugh] I promise—I’ll include some more links in the show notes if you wanna get deeper.
But, OK, basically what happens is, you have a computer and what’s gonna happen with GPT-3 is you put in a prompt, which is text, and then it goes into the black box, and then it comes out with an output. So the input prompt might be like, “Recite the First Law of Robotics.” I’m totally stealing this from somebody’s explainer, because I think it’s a good one, right? It goes into GPT-3, and then there’s an output that comes out from that. And the ideal output in this case, you know the answer. It’s Isaac Asimov’s First Law of Robotics, right?
So what happens is, how does it get there? Well, what it does is, basically, it’s presented with an example, and then you ask the computer to predict the next word. And the model’s prediction at first is always going to be wrong. So you fed into it a bunch of text, right? So imagine you feed it in everything in Project Gutenberg, and it quote-unquote “reads” everything in Project Gutenberg, which means that the computer, like, has all of those examples of text in it, right?
ELM: Mm-hmm.
FK: And so you feed it in all of that, and then you give it, “Second Law of Robotics.” And it comes up with a guess for what the next piece of text will be. And it’s gonna be wrong. So then it calculates—the system then calculates how off the error in the prediction is, and it updates the model, so that it makes a better prediction the next time. And it does this billions and billions of times, right? So that’s the training part of it.
ELM: So who is telling it that it’s wrong? The researchers.
FK: Yeah.
ELM: And then once it’s out in the world, the users.
FK: Yes. Exactly. Exactly.
ELM: Right. So it’s constantly learning from feedback from humans.
FK: Exactly. But the other thing is, the input, you gotta remember, it’s not that it’s—I guess in the unlikely situation that, like, you asked it—you gave it Project Gutenberg, and then you gave it a bit of Jane Austen, it might make a good prediction, maybe, for what would come out of that sentence, because the sentence is already in its corpora. But most of the time, it’s not gonna be—but there’s also probably lots of other sentences that start with the same words as the sentence you’re asking it to complete, right?
ELM: Mm-hmm.
FK: You see what I’m saying?
ELM: Mm-hmm.
FK: So it’s not—it’s basically never going to be generating the same text that you put into it. It’s not gonna produce, it’s not like—
ELM: But you don’t want it to. It’s not a search engine, you know?
FK: Exactly.
ELM: It’s not just spitting it back to you. The point is that it’s generating something new.
FK: Exactly. And I think that that’s an important thing to remember, because I’ve seen people say things—I think maybe more on the art side, and this—I mean, again, this is super quick and dirty, but this is kind of the same way that the art stuff works. Obviously, it’s different, because it’s like, you write in text, and then it produces a visual image. Ao there’s more steps, and also there’s a different way that art comes from—[laughs] you know? There’s a different kind of prediction model.
ELM: OK, but so—but the thing with the visual art one, though, is, like, you know—I think it was always presented as more of a challenge. And knowing some people who work in machine learning, too, they’re usually working with words and numbers, right? You know?
FK: Right.
ELM: To solve specific kinds of problems for specific fields. But—
FK: Right.
ELM: —the art one, you know, they always talked about the challenge was, like, the identification of images, right? You know?
FK: Right.
ELM: That was the thing that was going to take time. And, obviously, the way all this stuff works, it is compounding. The more you feed into it, the more you teach it, the more it can teach itself—
FK: Right. Right.
ELM: —and it starts circling and circling, right? And so it has to be able to know a picture of, you know, dogs playing poker, and those are dogs, and that’s a poker game.
FK: Right.
ELM: You know what I mean?
FK: Right, in order for you to write in text—
ELM: It has to be able to identify the element—
FK: Yeah.
ELM: Yeah. For me to be like—
FK: Right.
ELM: “Now do dogs playing poker on the moon and—” et cetera, et cetera.
FK: Right.
ELM: A cool prompt I would give it.
FK: Right. One of the things that’s hard to realize with art is that the way that it’s generating the art is not by looking at art the way you or I do. It’s by reducing the art to, like, effectively, the array of pixels that appears in a digital image of the art.
ELM: You don’t know how I look at art.
FK: OK, you’re right. I guess I don’t. [ELM laughs] Your pixel vision aside, though, it’s basically saying, “What is the probability—”
ELM: Mm-hmm.
FK: “—that this pixel will be this color?” Right? You know what I mean? So I think that one of the things that happens—and furthermore, like, the reason that it shows up, right, when you put something into any of these kind—whether it’s a Stable Diffusion thing, whether it’s DALL·E, whether it’s Lensa, whatever—sometimes it produces something that appears to have a signature. And the signature is unreadable, right?
ELM: Mm-hmm.
FK: Well, it doesn’t have a signature there because it has stolen a piece of artwork whole-hog, and then, like, erased the signature. There is no piece of artwork that it’s individually stolen. Now, I’m not saying that it’s ethical for them to scrape art, let’s note. But what’s happening is, it’s looked at a lot of pieces of art, and it’s learned that there’s a high probability that an artist will sign a piece of art in the bottom-right corner, because that’s where people sign pieces of art—
ELM: Sure.
FK: —and it’s looked at all of these examples. So when it gets down to looking at those pixels, it figures out what the probability is, and it’s like, “Oh, yeah, there should be a squiggle here.”
ELM: Hmm.
FK: And I think that can be quite confusing to people, because it can feel like, “Wow, somebody drew this picture and you just tweaked it a little bit, and then you erased their signature.” Again, I’m not trying to say that this is, like, a perfectly ethical enterprise, but that is not what’s happening—
ELM: Sure.
FK: —when they produce this piece of art.
ELM: Right, right. So Lensa, the thing that you’ve been referencing, that is a—what’s been going viral is, on Instagram, it’s a series of sort of fantasy avatars that you can make if you put in pictures of your face.
FK: It provides you with different styles, and one of the styles is fantasy, and one of them is sort of, like, futuristic, and one of them is like—
ELM: That’s—yeah.
FK: —cool, and one of them is romantic. You know? It provides you with, like, eight of them.
ELM: Because I, you know, see fandom people’s accounts, [FK laughs] I’ve seen people doing the fantasy ones, doing cyborg ones, you know, like—
FK: Those ones also look cooler, honestly. Like, having done it—
ELM: Yeah, the genre-y ones. Right.
FK: You know, I will say, having done it, it produced cooler ones.
ELM: You’re so fucking basic, Flourish. I love it.
FK: I’m interested in this topic! [ELM laughs] I did it for a variety of reasons, but one of them is that I’m genuinely interested in this specific topic.
ELM: [laughs] All right, well, you don’t need to go into specific, you know, like—there’s a lot of—I mean, that also touches on, like, these viral face-altering apps, right?
FK: Oh yeah. For sure.
ELM: And there’s a lot, a lot of very fraught politics around that, all of them, and I see that replicated in this one as well, which is interesting. But let’s bring this back to fandom. Hopefully that was a non-confusing explanation.
FK: [laughs] I did my best! It is a fundamentally confusing thing.
ELM: Yeah.
FK: Because it’s hard to make AI.
ELM: I mean, I think at the very, very—the very, very basic building blocks, hopefully that’s clear. There is a corpus, a body of words or numbers or visual images, right? And you’re teaching a model—
FK: Yeah.
ELM: You’re teaching a computer program, [laughs] I guess we can say—
FK: Yeah.
ELM: —to understand what that is, and then to start making that, right?
FK: And it’s all based on, like, probabilities of what’s gonna come next.
ELM: Right.
FK: So it’s not that you’re teaching it—it’s not like you would teach a person and be like, “Feel the vibes.” It’s like, “No. Here is a pixel. What is the next pixel going to look like, if this was the first pixel?”
ELM: That’s an incredible description of teaching humans. [FK laughs] Yeah, so, when—
FK: But I mean, kind of! Kind of, though.
ELM: When your husband gets in front of the classroom, and he’s like, “All right, we’re just gonna feel the vibes today.”
FK: A little bit!
ELM: And then the students are like, “Yeah, I’m vibing, professor.”
FK: A little bit. [ELM laughs] Sometimes. But you know what I’m saying.
ELM: Put that prompt in. Generate that fic. OK.
FK: We think on a more macro level, and it’s thinking on a hyper-micro level.
ELM: Sure. Right. I mean, that’s what computers are. I think sometimes it’s hard for us to—I think we can drive ourselves a little bit bananas, [laughs] trying to think of the really small building blocks—
FK: Yeah.
ELM: —that are making up every single little thing that your computer is doing, right?
FK: Totally. But back to fandom, though.
ELM: Back to fandom. OK, so obviously the logical step here is, I think what Steve is describing, saying, take the visual art one. You know, he’s talking about text, but, like, start with visual art. And I’ll say, “I want Professor Charles Xavier [FK laughs] and Erik Lehnsherr kissing next to a—” I don’t know. Where would they be kissing? [laughs] “—next to a bonfire, like, a conflagration.” Right? Like, you know?
FK: That sounds right for them.
ELM: [laughs] Yeah, I was trying—I was gonna say, like, in front of a Christmas tree, or something, and then I was like, “Nope. [both laugh] It should be in front of something much more—”
FK: Death and destruction!
ELM: “—much more dramatic.” You know, so, what’s it gonna do? It’s gonna—it depends on what the body of the visual images is. Maybe in this case it’s all the images that are on the web, right? You know?
FK: Yes. Very likely.
ELM: And it could be pictures of the various actors—
FK: Mm-hmm.
ELM: —pictures from the comics—
FK: Yup.
ELM: —and then all of the fanart, right? You know?
FK: Yup.
ELM: And they’ll generate something.
FK: Yup.
ELM: And God bless every single fan artist, but I have seen things tagged “Cherik fanart,” and I could not tell you who those two gentlemen are. [both laugh] And so at that point, I’m just kind of thinking, like—it’s interesting to me, thinking about that, and thinking about what fanart is now. Is the goal accuracy? Is the goal—what does accuracy mean? Right?
FK: Yeah.
ELM: You know, like, because a lot of fanart now is just, like, vibes and style, you know?
FK: Yeah, and I actually wouldn’t be surprised if the thing that it produced—because it’s also looking at, like, romantic images on Christmas cards, or, you know, [ELM laughs] like, on—I mean, OK, we’re not on the Christmas tree thing.
ELM: Bonfire! [laughs] It’s gotta—
FK: On bonfire cards. You know what I mean, though. But it’s also looking at, like, romantic images of people kissing that have nothing to do with any of this, right?
ELM: Yeah.
FK: That’s part of the information it’s receiving, too. I wouldn’t be surprised if—I mean, I haven’t done this, and it would maybe be interesting to, like, put this into DALL·E, or something, and see what happens. But I wouldn’t be surprised if it actually looked more like Magneto and Professor X than a lot of fanart.
ELM: Well, what is “more like” them? Like, more like the actors who play them—
FK: More like the actors.
ELM: —or the drawings in the comics, yeah yeah yeah.
FK: Or more like the way that the comics represent them.
ELM: Yeah.
FK: Which, I mean, doesn’t make it better, to be clear, necessarily.
ELM: Yes.
FK: And I think it’s also a weird thing to be like, “Well, what’s—what do we value in this art? What’s better in this case?” Because I value relationships with fan artists, and I value seeing their style and the vibes and the history and all that stuff, you know? That’s one of the things I value.
ELM: But, I think that—I mean, all right, we’ll bring it into the text thing now. What Steve’s talking about, you know. Imagine there’s a fanfic generator, right? And you say, like—all right, we’re just gonna keep it on my guys. But, you know, say, like, plug in, “Charles and Erik—” I don’t know, “—only one bed,” or whatever.
FK: Right.
ELM: Now, the cynical [laughs] thing that I’m going to say right now, is I think a lot of fanfiction already reads like this, right? You know? Where I feel like I could just find and replace their names, and it could be “Steve and Bucky” and “only one bed,” and the story will play out exactly the same way, and be, like, tightly written, good dialogue, but I could change it, and that—what I am saying, though, is, but every single one of those stories is written by an individual, whether they feel interchangeable or not. And there’s something very different to me here about the idea of, like, “Oh, well, now you don’t even need a person to bother to write it. You just stick in their names like a Mad Lib.”
FK: OK, if we are talking about a fictional thing that gets this all right, and feels right, then—
ELM: Well, what’s “getting it all right”? Sometimes I’ll look at the things that people say are, like—
FK: Right, and you’re like, “What the fuck is this?” [laughs]
ELM: You know, they’re like, “It’s an incredible fic.” And I’ll read it and I’ll be like, “This is the most generic thing I’ve ever read.” You know?
FK: No, but I mean, like—so, my experience with GPT-3 and with ChatGPT, is that if you skim it, it looks OK, but when you read it sentence by sentence, it quickly becomes clear that this thing has no idea what it’s writing. It’s poorly paced, the sentences don’t actually make sense with each other—
ELM: Ah.
FK: It falls apart as you get into it. And this is a real serious problem, and it’s been like this for a long time.
ELM: Hmm.
FK: So I used to have conversations with people in the film industry about, like, when is this—when are we gonna be able to generate film scripts, right?
ELM: Yeah.
FK: Because obviously people are interested in this. And the answer is, well, we can generate a lot of things, but it’s really, really hard. It’s much harder, actually, than with—actually, the visual art that it’s generating is not that great either, right?
ELM: [laughs] No, it’s not.
FK: It’s not. Sometimes it can be—one of the reasons why people are so into Lensa, for instance, like, having done it, is that it has ideas about what’s culturally flattering for you. You put in whether you’re male, female, or nonbinary, or anyway, not gendered, and it does different things with those, and it changes your features to make you look—for instance, even with the nongendered one, it made me lose weight, and improved my jawline, and made my cheekbones higher, in every picture I put in.
ELM: [laughs] That’s not problematic in any way.
FK: Right?
ELM: Yeah.
FK: And it made my nose sharper, and, like—
ELM: I mean, I will say, I saw one of my friends on Facebook who is Black put in her results and it was shocking.
FK: Yeah!
ELM: It was like—it just—it was like—I mean, it seemed very racist to me.
FK: Yes!
ELM: You know, it was like—[both laugh]
FK: Right!
ELM: I mean, and I think even the thing you’re describing, and about the way different people from different backgrounds’ features can—you know what I mean? Yeah yeah yeah.
FK: And it has an anime version, and it makes your eyes look different, you know? It’s really—so, the thing is, what I’m trying to say about this, though, is, A) this shows how shallow these models actually are, but then B) it’s also like, why do we like these things? It’s because, to some degree, it flatters our cultural expectations of who we are and what we look like, or what we want to look like. Genuinely, like, there’s a lot of—
ELM: OK, I mean, like, in the example you just gave. But in the one I just gave, that’s not “flattering”—
FK: Right, right. No, I know.
ELM: —the way she looks.
FK: I know that.
ELM: Yeah.
FK: I know. I mean, what I’m saying is, it shows—like, when it doesn’t work, it shows how shallow it is. But the only reason that people really love these Lensa things is, if they are indeed flattered by it—
ELM: Well, sure.
FK: If they, you know—
ELM: Yassifying.
FK: And—
ELM: Yes. [laughs]
FK: And then when you—right. And then when you, like, look at it a little bit longer, you’re like, “That’s actually not a good painting.” Maybe it, like, is photorealistic on some level, and it’s doing some interesting things, but it’s not the way a real painter would compose it.
ELM: Yeah, I gotta say, that one that won the prize that everyone was all up in arms about—do you remember this one?
FK: Hmm.
ELM: It was the weird-looking, like, space court. You know, it was like a royal court in space. It just—I find so much of the stuff that gets generated and these, like, some of these tech folks seem to think is beautiful, is so cringey and basic. [FK laughs] And it’s like, and I don’t wanna demean a teenager, but it seems like something that a teenager might—you know, like, a 13-year-old boy would be like, “Wow!” You know, that kind of thing.
FK: Yeah, well—
ELM: And it’s like, “You’re an adult man.” You know? This is—I don’t know, I guess because there’s boobs here you’re excited? [FK laughs] Like, I don’t—you know what I mean?
FK: Well, and some of it can be cool and fun. I have a friend who uses DALL·E to generate infinitely—like, she’s just really obsessive. She uses it to generate images of, like, treehouses filled with rainbow light and plants. And she does this all the time. Her Instagram is just, like, you know, thousands of images of this, which she evidently really loves and finds beautiful and comforting and is interested in playing with it. And that’s cool. That’s something that she’s enjoying. I’m not hating on it. But you can see how if, like, text generation can produce a paragraph that’s like that, right?
ELM: Mm-hmm.
FK: That’s not the same thing as producing a successful novel-length story that holds together, or—
ELM: All right.
FK: —a movie screen—you know what I mean?
ELM: Yeah, a screenplay or a novel-length story, but we’re talking about fanfiction here, and fanart, you know? And I just want Professor X and Magneto kissing on their one bed in front of the bonfire. Right? And I’m saying I sometimes don’t think that fandom is that deep.
FK: Sure, sure.
ELM: I think there’s some—and this is a huge group of people, right? I think there’s some people sometimes who—maybe that’s not them all the time, [FK laughs] but I think sometimes, would—I mean, maybe not as it is now.
FK: Mm-hmm.
ELM: But we’re talking about something that is learning at a pretty fast clip, right? And maybe we’re looking at this in a year, six months, or a month.
FK: Mm-hmm.
ELM: If everyone keeps fucking around with this, [laughs] it’s only gonna speed it up, you know? And so, like, I think what Steve’s describing is realistic. But I also don’t…you know, I think when I think about the fics that I think are best, I can’t imagine—not just because—obviously, there’s some fics that I like that aren’t so, so, so character-specific, right?
FK: Yeah.
ELM: You know? And they’re internally consistent, or whatever, but they maybe feel a little OOC, or whatever.
FK: Right.
ELM: But they’re still—I don’t know. There’s just something there that has to do it for me. Some kind of connection, some kind of understanding of the characters, right?
FK: Right.
ELM: And just, like, taking names and sticking them in scenarios, and I don’t know—I can’t imagine how this could actually do any real character work. For an original movie script, maybe that’s—because then you’re just sort of—
FK: You’re creating the text that someone’s then going to read and—you know?
ELM: Yeah. Yeah yeah yeah.
FK: —produce. Yeah, I mean, I don’t know. Maybe it’s possible that this goes there. But I think that then that also leads us to questions of, like, “OK, so imagine that all of this is true. Imagine that we are truly going to have an AI that—” Well, OK. Now let’s be clear, I don’t think it’s true for a variety of reasons, one of them being that we see how the corpus that it’s trained on is, like, fucked up in ways that will mean that it never produces things that are, like, actually innovative or interesting, right?
ELM: Mm-hmm.
FK: I mean, not to say that it couldn’t, I guess, but you know, everything that you put into Lensa always gives you the same, like, whiteness-oriented beauty standard.
ELM: Sure.
FK: Everything that you put into a corpus that has been trained on romance novels will give you a particular kind of romance novel, and it’s never gonna give you a thing—because it’s only looking at what the best prediction is, right?
ELM: Mm-hmm, mm-hmm.
FK: So it’s never gonna give you a thing that’s actually surprising or curious without a lot of human intervention. Like, you can get something back and be like, “No, I didn’t want it that basic. Actually, I want the hero to die in the first third of the book, and then what happens after that.” And then it has to rethink it, and then maybe it produces something unusual then. You know, maybe you can keep tweaking it that way.
ELM: Sure.
FK: But it’s gonna take so much human intervention at that point, that now we’re really not talking about an AI creating something. We’re talking about human writing that is computer-assisted.
ELM: Right.
FK: Do you see what I’m saying?
ELM: I do see what you’re saying, and I kind of don’t see the point of it.
FK: Yeah, well, I don’t either. [laughs]
ELM: You know, because so much of, you know, the broader culture industry is already somewhat templated, right?
FK: Mm-hmm.
ELM: That work has already been done.
FK: Yeah.
ELM: You know? Whether it’s a literal template or just the kind of sets of structures that people just sort of plug and play, and they’re like, “OK, well, these are the beats, this is what people want, and these are the kind of characters they want, and we’re gonna put it all together.” Right? You know? If you look at—
FK: Mm-hmm.
ELM: —people who write commercial fiction, like, you know, do for-hire work, right?
FK: Mm-hmm.
ELM: And they’re like, “Here’s the beats you’re gonna hit.”
FK: Mm-hmm.
ELM: “This is the kind of character. This is the guy. This is the girl.”
FK: Right.
ELM: “It’s a romance. Go. Write it.” And you know, the success of that is in the execution.
FK: Mm-hmm.
ELM: And obviously, there’s a sentence-level element, too, right? You know?
FK: Mm-hmm.
ELM: But I feel like we’re already there to some degree, and what you’re describing is coming at it from the opposite angle, but in a more tortuous way. [laughs]
FK: Yeah.
ELM: And so it’s like, why would you bother with that, when you already have a kind of standardization of a lot of these genres, or, like, different types of art creation, right?
FK: Yeah.
ELM: You know, and this happened because it’s commercial, you know? It’s—people wanna be able to churn out X number of books in this genre, or—
FK: Yes.
ELM: —X number of “man in red sweater, woman in green sweater, small-town returning” Hallmark Christmas movies, you know what I mean?
FK: Right.
ELM: I hope I got it right. I hope it’s not the woman wears the red sweater and the man wears the green sweater. [FK laughs] Right? But, like, there’s so many things in the culture industries that are already so cookie-cutter.
FK: Yeah.
ELM: You know what I’m saying?
FK: Yeah, and I mean, I think the other thing that it’s important to note about this, though—I agree with everything that you’re saying. I would also say that we’re—like, right now, when I’m talking about this, I’m talking specifically about GPT-3, which I find very uninteresting. [ELM laughs] You know? I think that there are other—
ELM: Enough to do a whole episode about it. [laughs]
FK: No no no, what I mean is, I find it interesting as a concept. I find the text it outputs usually to be pretty uninteresting, right?
ELM: Yeah.
FK: I think that there are ways that you can use other forms of machine learning or computer generation to make interesting, admittedly avant garde and not, like—interesting artworks that are not—well, OK, two things. One, either to sort of assist or come up with surprising or interesting phraseology, in that way. Or two, to create artworks that are not about humans, right?
ELM: Sure.
FK: You know, so there’s been a few different articles about a novelist who used a computer program to, like, help get them out of ruts, or to, like, propose ideas to them or something, and I think that’s, like—that’s a longstanding thing. There’s things like National Novel Generating Month, which is NaNoGenMo, and, like, every year, a bunch of people create these weird things—it’s not even—some of them, it’s like, “Is that a novel? If it is, it’s a very avant garde novel.” You know?
ELM: Sure.
FK: And they’re interesting, and they’re weird. So I just want to be clear that there’s, like, also that kind of stuff going on, which has nothing to do with this commercial end of things.
ELM: It’s amazing to listen to you say all these things, and I think, like, “Did Flourish find this stuff interesting before they met Nick? Or is it because they—” Is this, am I seeing, like, you know, a decade-plus of, like, Stockholm Syndrome, like…
FK: [laughs] I found this stuff interesting before. I did.
ELM: That was a big turn-on for you? You were like, “Generate something.”
FK: I was interested! [ELM laughs] No, I mean, but that all is very different to fanfiction, which also, you know, to your point, actually one of the pleasures of fanfiction is the repetitiveness for a lot of people, right?
ELM: Yeah. I mean, I think what you’re describing is very different, and obviously, like, I think there’s people in fandom who’ve done stuff like that too, you know?
FK: Oh yeah. No doubt.
ELM: And I mean, I’m sure you’ve encountered it in these spaces. I’ve even come to some of the stuff that both you and Nick have participated in—
FK: Mm-hmm.
ELM: —where some of this stuff definitely would fall within a fannish realm—
FK: Sure.
ELM: —whether people want to describe it that way or not, is—
FK: Right.
ELM: Maybe they just wanna say they’re engaging with pop culture characters, or whatever.
FK: Right right right.
ELM: But, like, you know, people certainly have made little projects like this, or, you know, generators or whatever. Like, people who are technically oriented in fandom.
FK: Mm-hmm.
ELM: I think the anxiety I hear—I mean, there’s a couple different pieces of anxiety, but the one that I hear in Steve’s letter that I’ve seen reflected around a lot is that kind of idea of, like, “Oh, well then it’s just gonna be this plug-and-play thing.” Kind of what I’m describing [laughs] that already happens in commercial fiction.
FK: Mm-hmm.
ELM: But just removing the actual fan creator from the equation, and saying—
FK: Right.
ELM: And you know, like, OK. Say they can get it to a state where it’s actually, like, a fine thing to read, you know?
FK: Yeah.
ELM: Could be kind of great for a lot of fans, [FK laughs] because, you know, they’d be like, “Oh, write it yourself? I’m not a writer. Oh, draw it yourself? I’m not an artist.”
FK: Yeah yeah yeah.
ELM: Go ahead. You know? Like, no one has put Professor X and Magneto in front of that bonfire? [FK laughs] They have. I’d love to see it, right? You know? And maybe being able to see your own ideas expressed, even if it’s not very good quality, is the valuable thing for you. And then you’re not reliant on other people to create things for you, particularly for free, you know?
FK: I mean, I think—yeah. I mean, I think that this is—I think in the non-fandom way, that’s what’s appealing to my friend who generates the infinite treehouses, right?
ELM: She doesn’t have to, like, [laughing] commission someone to make constant treehouse art for her.
FK: Right. And I mean, and I think that speaking about the commission, that is one of the big things that this is an anxiety piece, right?
ELM: Right.
FK: People feeling like, “Commission me!” You know? I mean, like, obviously—
ELM: It’s a different question for fanfiction writers versus fan artists here, right?
FK: Versus fan artists, right.
ELM: What I’m saying—yeah. Yeah.
FK: Yeah. Because, I mean, the people who I have seen who are most upset about this—and I think reasonably so, to some degree, you know—are people who make money off of, like, especially user icon commissions.
ELM: Yeah.
FK: Being like, “Hey! Is my job going away, effectively?” And also the product, I think that they’re correct also in saying, like, the product they produce is better than the thing that you’re generating on AI, but all of the sudden, everybody’s able to basically eat this AI-generated junkfood, effectively.
ELM: Yeah. Right. I mean, there’s not a hopeful answer there. I think that they are right to be worried.
FK: I think that’s true, although I think that the vast majority of people who are using AI to make these images, especially with Lensa and things, are people who are not going to pay for a commission to begin with.
ELM: Yeah. OK, that’s fair.
FK: I mean, not to say that they’re not correct, also, that it’s gonna take stuff out, but also, you know…
ELM: Yeah. Right, right. But I mean, I feel like that kind of underscores—I mean, well, you’re talking about profile pictures now, but it does underscore kind of the precariousness of, like, you know, fan artists as a profession, right? You know?
FK: Yeah.
ELM: Obviously, an already precarious situation with kind of skirting along the edges of the law, and—
FK: Mm-hmm.
ELM: It isn’t, I think, talked about as often as it should, how fan artists are still, you know, butting up against corporations and having their stuff taken down or whatever, right?
FK: Right.
ELM: I feel like there’s some sort of commonly accepted thing that all fan artists can just charge money now, and it’s fine.
FK: Yeah.
ELM: And it’s like, the law never changed, and in fact the law is not different for visual art [FK laughs] to fanfiction. But these weird, rigid, like, two sides of a binary—
FK: Yeah.
ELM: —about fanfiction versus fanart have somehow sprung up in the last, like, decade, right?
FK: Yeah, absolutely. Absolutely.
ELM: But, like, it’s also possible that fanart found a moment where people would pay to see the scenarios they wanted to see depicted.
FK: Yeah.
ELM: Ten years ago, they wouldn’t do that.
FK: Mm-hmm.
ELM: And maybe that moment is gonna pass, if people are just paying to see something depicted, and not paying for a style, right? Or, you know, artistic—like, you know, artistic choice, I guess. I do think there’s some portion of people who are commissioning fanart just to see the thing in their head—
FK: Right.
ELM: —down on the page, right? And that—and they’re doing it maybe because they like the person’s style, or they have good rates, or, you know, whatever, right?
FK: Right.
ELM: You know, and so, like, if there’s a way they can get a halfway decent version of that for free?
FK: Right.
ELM: I bet they’re gonna do it for free, and that just—that’s the hard truth, I feel like. I mean, I don’t know. I don’t know.
FK: No, I mean, I think that’s true. And it’s hard because it reminds—I mean, I think that I said this before, it reminds me of when photography came into existence, and suddenly portrait painters were like, “Photography is awful! It takes the soul out of images!” Like, all of these things, right?
ELM: Yeah.
FK: And nothing that they said was totally wrong. There are still some portrait painters, but there’s many fewer, right? [ELM laughs] Or it’s not as central.
ELM: Yeah. Yes, that is not—
FK: Right? Like, if you had a—yeah.
ELM: If you were a painter, and you told me you were a portrait painter, I’d be like, “Oh, wow.”
FK: Yeah.
ELM: I’d be like, “Oh, unusual.” You know? [laughs]
FK: Right. I mean, like, you know, if you haven’t commissioned an icon of yourself from somebody, like, you know, online, then probably the only way that you’re encountering that, assuming that you’re not, like, the dean of a college, [ELM laughs] you know, or the president of a bank—
ELM: [laughing] Yeah, like, a rich person.
FK: Yeah…is to, you know, go to Disneyland and have one of the caricature artists—
ELM: Wow, you don’t need to go to Disneyland, you could walk outside of your literal house.
FK: Well, that’s true. My literal house—well, Central Park and Disneyland—I mean, Disneyland was where I encountered it when I was a child, right? [ELM laughs] Because it’s not in, like, suburbs always.
ELM: I’m just reminding you that you live in the center of, like, clown city, basically.
FK: There are—yeah. That’s—well, there’s not actually any caricature artists in Union Square that I know of. Anyway, whatever.
ELM: Are you sure?
FK: I haven’t seen one.
ELM: All right. I’m gonna look around.
FK: There’s lots of other things! They sell ashtrays shaped like pussies, [both laugh] but they don’t sell caricatures.
ELM: Don’t try to make Union Square sound cooler than it is! [both laugh] It is not cool.
FK: Anyway. They sell pre-rolls. There’s a lot of people selling pre-rolls. And a lot of people selling chess lessons.
ELM: Again! Again, you’re trying to make it seem cool. [laughs] It’s not right.
FK: No, this is the opposite of cool!
ELM: That’s not right. [laughs]
FK: Anyway, chess lessons, pre-rolls, which you ash into a pussy-shaped ash—anyway. But right, like, so this is a real thing. And this is also a real thing for people who, I don’t know, like, made clothing, right? Historically clothing was better than it is today. It fit people better, it looked better, all that. But then technology moved forward, and now we all wear slightly ill-fitting clothes that are much cheaper, and there’s a lot of them.
ELM: Right. Right.
FK: And this is what we’re seeing, except in art.
ELM: [groans] All right. It’s interesting to me, like, thinking back over this conversation, how much I’m thinking from the perspective of a fan, too.
FK: [laughing] And how much I’m not?
ELM: No, yeah, I mean, obviously I agree with everything you’re saying about industries, or whatever.
FK: Yeah.
ELM: But I’m like, OK, well, you know, isn’t the point that you’re the fan, and you want this scenario—
FK: Right.
ELM: —so how are you gonna get it, right? You know, if you can’t make it yourself, get it some other way, you know?
FK: Right, totally.
ELM: And that’s hard. But I also—it’s interesting, because I feel like, you know, there’s a lot of discussion in fandom saying, with fan artists, saying, “If you don’t reblog our stuff, if you don’t commission us, then we’re not gonna make this anymore.” And people kind of vote with their—I was—they don’t vote with their feet, they vote with their clicking fingers, right? You know? [FK laughs] And it’s like, OK, then this is not—these people, like—I don’t know. It’s a sort of murky space to me, because it’s like, there’s a lot of conflicting motivations, right?
FK: Mm-hmm.
ELM: I think there’s a lot of fan artists who are in this space because they are talented artists and they want to draw, and they know that there are people—we talked about this in the past with furry commissioning, right? You know?
FK: Sure.
ELM: You know that there’s people who really want those drawings, right?
FK: Yeah.
ELM: You know, and they’re gonna pay top dollar for a great depiction of their fursona. And I absolutely get that, and I don’t blame anyone who’s taken advantage of that space, but I think there’s a lot of people within the space who are not thinking about anything beyond, “I love this show—”
FK: Mm-hmm.
ELM: “—I love these characters, I love their faces, I wanna see more of them.” Right?
FK: Right.
ELM: And I think that those two things—taking the AI piece out—I think those two things aren’t always in harmony, in ways that I don’t see reflected so similarly in the fanfiction space, though I think we’ve talked about this recently, too. There are fanfiction people who are trying to build a brand, get a lot of hits, right? You know?
FK: Sure.
ELM: And that feels like it’s coming from a different place than, “I love this show, I love their faces—”
FK: Right.
ELM: “—I wanna see them in different scenarios.” That kind of thing, right?
FK: Yeah. Yeah, absolutely. I think—I mean, there’s one other piece in here, which is, I think, this question of sort of, um…understanding of what quality is and means, and taste within this.
ELM: I think I already brought that up a lot by insulting fanworks on a whole—
FK: You did, you did.
ELM: —[laughing] and saying a lot of people don’t have taste.
FK: You implicitly brought it up. But I think that there’s an explicit way to bring it up, too.
ELM: Go ahead.
FK: Which is to say that there’s—you know, one of the problems here is that artists have—because they have to in order to develop their skills at drawing—developed a sort of more fine-tuned taste for what’s good and what’s bad than the average person who’s looking at art. I know that I don’t have the same—
ELM: Sure.
FK: Like, I have a fair bit—like, my dad’s an artist. I have a fair bit of art appreciation in my life, you know? But I definitely don’t have the same perspective on art that somebody who has dedicated their life to making art would have.
ELM: And studying it as a craft, absolutely. Yeah, like—
FK: Right?
ELM: I mean, you know, yeah. And that’s not to say you can’t enjoy it. I think that people who don’t know music theory can still enjoy a symphony. [laughs]
FK: Sure, and, you know, and when I think about, like, whatever. I’ve—[laughs] this Christmas I had to actually actively tell people, “Don’t buy me commercially made knitwear, because I don’t wear it anymore, [ELM laughs] because I have too many opinions about it.” Which is not to say I think everybody has to—”
ELM: Shit, I got you a bunch of H&M sweaters. You’re not gonna wear them?
FK: No. [both laugh] But that’s not to say that, like, you know—but I think that the—right? So I have, like, because I knit all the time, I have a very highly tuned idea of the craft of that and the work that goes into it and all this stuff. And now, the average person just wants to wear a fucking sweater, right?
ELM: Mm-hmm.
FK: And I think that it’s hard, because there’s so much moral value that we place on art, and I don’t think it’s necessarily wrong. I think people are better for understanding art more and understanding quality of craft and all of these things. I think those things are good. That kind of education is good. But not everybody is gonna have it at all times. And so there’s gonna be a lot of people for whom, as you were saying, they just wanna see the two dudes kiss, and they don’t really care that it’s a bad composition or that, like, this looks funky or whatever. They’re gonna be like, “That’s fine.”
ELM: I mean, it might look great, also. It might look photorealistic, and, I mean, whatever. It depends on how we’re defining it, right?
FK: Right.
ELM: I think there’s—some of the stuff that you see—I mean, most of the stuff that you see is like, “What are we doing?” I mean, people are also putting the most deranged combinations of words into DALL·E, right? You know?
FK: [laughing] Yeah.
ELM: And you’re like, “I didn’t need to see that part of your brain, but thank you.” [FK laughs] But what you might wind up getting is something that is more polished than some of the fanart you see, right?
FK: I mean, yeah. I mean, that’s also true with things like knitting, right? It’s always got perfect tension.
ELM: Yeah. Yeah yeah yeah.
FK: There are certain parts of knitting, and there’s certain things that you, you know—there’s certain aspects of knitting that are definitely better when they come from a knitting machine.
ELM: Sure. I don’t know, I just think a lot of people aren’t thinking about it that deeply. Like, I absolutely agree with you. Yeah, I just—you know, you just wanna see it, and it looks good. You know?
FK: Yeah.
ELM: You know, put the sweater on. Not from H&M. Just don’t do it. It’s not worth it. [FK laughs] Unless it’s for, like, one night only.
FK: From Uniqlo. [FK laughs]
ELM: Yeah. They make nice sweaters. You know, they’re not—yeah. They’re explicitly machine-made.
FK: Yeah, I mean, they’re fine. Whatever. We can talk about, like, wool quality and machine making and stuff, but they’re totally fine.
ELM: Yeah. They look flattering.
FK: And they look nice.
ELM: I’m wearing one right now.
FK: Yeah.
ELM: I’m wearing a sweater from Uniqlo. Doesn’t it look lovely?
FK: Yeah, I mean, it looks like a functional sweater. [ELM laughs] It looks nice, you know? [laughs]
ELM: Right.
FK: Again, I don’t, you know—but I think that culturally then this also causes some problems, right? Because if I were walking around being like, “Elizabeth, you know what, you’re taking jobs away from the knitters who knit really good quality things that are really important. Why are you not hiring people and paying them a fair wage to knit this sweater for you?”
ELM: Well—
FK: You’d be like, “What the fuck are you talking about?” Right?
ELM: I mean, and then there is the—or are you saying the next moral step here is to say, “Oh, so many people can’t afford to commission fanfiction and now we’re democratizing it by allowing them to create their own?”
FK: I—yeah, I mean—
ELM: Or—not fanfiction, sorry, fanart. You know what I mean.
FK: Yeah, I mean, I don’t—I’m not, again, like, I really do value both the work of handknitters, of whom I am one, and the work of fan artists. But I do have to raise these questions, because I think that whether or not we are into it, that’s the direction it’s going.
ELM: It’s tricky. I would love to hear from some fan artists on this.
FK: Me too.
ELM: Please, please write us in. Because, you know, we’re making some assumptions about fanart whenever we talk about it, and we’ve actually gotten a few comments recently asking if we would do a full episode on fanart, and this isn’t it, but, like, [FK laughs] it would be great to have a fan artist on to talk about the state of things right now, so I think that should be a 2023 goal.
FK: I think that’s a good 2023 goal.
ELM: I think with the fanfiction side, though, I do wanna say, before we wrap up, there’s another piece of anxiety here. I mean, I guess it is mirrored in what you were talking about, with people saying they’re stealing art—
FK: Mm-hmm.
ELM: —to do Lensa or whatever. The fanfiction “Elon Musk is scraping the AO3” fearmongering post, it started on Reddit, and then it went around on Tumblr, and it’s, like, a hundred thousand notes or whatever within, like, a week, you know? And I’ve seen some good pushback. There was a good, very clear-eyed breakdown of what’s actually going on from Casey Fiesler that we can include in the show notes—
FK: Yeah, definitely.
ELM: —just kind of explaining the basics. But I feel like we’re in that space again, where people will get very—[laughs] whatever. I haven’t seen as much of this on this as I have in other situations, where people get very, um…hypocritical is too strong a word. But a little bit, like, “How dare!” [FK laughs] Like, “My fanfiction is my sacred—!” You know? And it’s like, “OK, you put that on the internet.” First of all, you’re transforming a work that already exists.
FK: Mm-hmm.
ELM: Second of all, you put it openly on the web.
FK: Right.
ELM: Wide open on the web. Right? You know?
FK: Yeah.
ELM: So it’s all these things. And I feel like we’re back in that space where the broader fanfiction community can’t handle kind of the idea that they are involved in a continually remixing conversation.
FK: Right.
ELM: And, like, yeah. It’s shitty to think about computer science researchers just yanking this text just because it’s there, right?
FK: Right.
ELM: But that is also what they’re doing with Project Gutenberg or, you know, any body of public-domain literature. And, like, you know, I have some feelings about some of this stuff as a person who went to school for the digital humanities, [FK laughs] where there was a lot of text-scraping, and I think—
FK: Oh, so much.
ELM: [laughs] I think it’s rare, I can say—and I don’t know how you feel about this, being adjacent to this world, as well—it’s rare I ever think that the researchers who are pulling these bodies of texts are asking worthwhile questions. [FK laughs] Like, the most famous digital humanities project of all time is the one where a researcher analyzed Agatha Christie’s works, and then determined because she had less word variation in her later works that she had dementia, right?
FK: Yeah.
ELM: And it’s just like, “Bro? Bro.” [laughs]
FK: Yeah. There’s a lot. I mean, I once was—yeah.
ELM: For example.
FK: Yeah. No no no, I once was speaking to somebody about AI, like, dealing with movie stuff, and he was so proud to say that his model had figured out that women don’t like Angelina Jolie and do like Jennifer Aniston, because Angelina Jolie has, like, a black widow, you know, kind of homewrecker persona, [laughs] and Jennifer Aniston’s the girl next door. And I was like, “Wow…you trained a model on all of IMDB comments and everything else, and this is what you came up with. It correctly observed the world, I guess.” [laughs]
ELM: That is so sad. What a sad anecdote you just told me.
FK: Yeah! You know?
ELM: Also, I—no offense to Jennifer Aniston, but that’s completely backwards. [FK laughs] One of those people is great, and the other one, could take or leave.
FK: Yeah, I mean, I also feel that way, but it is true that—I mean, that was the tabloid narrative, and it just totally—so, you know, there’s all of these things, where it’s just like, yeah. No, but I definitely agree with you on the point of, you know, I think Casey has a lot to say about this ethically, and I think that there’s a lot to be said about it ethically, but the reality of living in this space is that when you put your stuff out unlocked online, it is going to be scraped.
ELM: Right. For a variety of reasons.
FK: It is going to be reused. It’s probably going to be remixed, for—yeah, for—exactly.
ELM: Yeah. Some man in the digital humanities may be determining that you have dementia right now [FK laughs] by looking at your—[laughs] your repetitive choices.
FK: Or, like, I may be looking at—I mean, not now, but like, the me of three years ago may be looking at it to try and, like, prove to your favorite franchise that they shouldn’t be assholes to you.
ELM: Yeah. Yeah.
FK: Or, I might be doing it to, like, try and make more money off of you.
ELM: [laughs] To have them make a decision that they despise. Yeah, absolutely.
FK: Right. You know, like, any of these things are possible, and they are happening right now, and that is the price that you pay for having unlocked things on the internet at the moment, like it or not.
ELM: Yeah.
FK: It just is what it is happening.
ELM: Yeah, and I deeply value the work that Casey has done and some other researchers, by trying to get through to the people scraping—
FK: Mm-hmm.
ELM: —whether they are on the industry side or in academia, to say, “Hey hey hey, [FK laughs] could you think for a second? Like, maybe just think about that?”
FK: Totally.
ELM: “Just give it one thought.” You know? [laughs] She’s a little firmer than that in her writing, [FK laughs] but you know, that got to be like, “Uh…” But yeah, I mean, you know, it’s easy to say—I don’t know. It’s not—we have skin in the game here. I’m not locking any of my fics. Go for it. You know?
FK: Yeah.
ELM: I don’t know. If a story showed up and it seemed like it was plagiarized from me, you know?
FK: Mm-hmm.
ELM: I mean, whatever. What we’re talking about right now, the easiest way to do that would be to actually plagiarize someone, you know?
FK: [laughs] Completely.
ELM: Just copy-paste it and change bits, you know? Change 20%, right? Change the names—
FK: Yeah.
ELM: —change little things here and there, right?
FK: Mm-hmm.
ELM: I don’t think that people are gonna be—I don’t know. I just, I think that we’re seeing a lot of anxieties on display, and I get it. I understand why everyone has been panicking over the last week in all these fields. And frankly, some of the things that people are panicking about outside of fandom are a lot more valid. I think that the professors saying, “Oh, you can realistically generate a paragraph response to this question that sounds fine.”
FK: [laughs] Oh yeah. Absolutely. It’s not gonna sound any more incoherent than some of the things I’ve received in classes before, for sure.
ELM: No. Right, and I mean, I think a lot of this also is a question of, like, well, then I think maybe people need to rethink some of how they’re doing this stuff, you know? You know, it’s interesting. I’m not a professor, so I don’t want to speak out of turn here, but you know, also the lengths that people are going to, to give these exams online—
FK: Right.
ELM: —and it’s just like, “Would a better demonstration of the knowledge this student has garnered not be to try to replicate this very strict, you know—”
FK: Right.
ELM: “—surveillance-laden, multiple-choice test-taking? Maybe there’s another way to see whether they’ve learned the material.”
FK: Yeah yeah yeah.
ELM: “Like, maybe it could be open-book, because they’re gonna be generating—like, expressing ideas, as opposed to just, you know—”
FK: Right.
ELM: “—selecting correct answers,” et cetera. Whatever. I am not a professor. [laughs] Just take that with a grain of salt. I’m just saying, like, I wonder if these are opportunities to kind of rethink the way the world works. But I think that the instinct—I understand why the instinct, in all of these scenarios, is to panic and say, “Oh my God, this is happening too fast. It’s already happening.”
FK: Yeah.
ELM: “This is gonna destroy what we know.” And it’s like, yeah. It probably will change it, but we don’t have to approach it with this kind of reactive, like, immediate hard shell going up and saying, “Ah no! How do we stop it!” Because it’s not stopping.
FK: Yeah. I agree with all of that.
ELM: OK, cool. So if anyone wants to make—I actually think, for the show notes, we should make some of the things that were described. I would like to see [FK laughs] Professor X and Magneto on one bed in front of a fire.
FK: All right. Professor X and Magneto on one bed in front of a fire. I will see if I can get on that.
ELM: Oh my—[laughs] I can’t wait. Do your—do the best. Do the best.
FK: Yeah, pulling up Discord right now. [ELM laughs] Getting on DALL·E.
ELM: Thanks, Flourish. Thank you for making me some fanart.
FK: Wow.
ELM: [laughs] OK, all right, should I leave you to it?
FK: Yeah, probably leave me to it. I gotta work on this.
ELM: Tinker.
FK: It may take a little bit of human guidance.
ELM: Do some tinkering.
FK: Some tinkering.
ELM: Sure. [FK laughs] Yeah. All right, great. OK.
FK: All right. I’ll talk to you later, Elizabeth.
ELM: Bye, Flourish.
FK: Bye.
[Outro music]