80. Art and AI collaboration with Sarah Brin

Transcript created using DeepGram.com

Hello, and welcome to episode 80 of the Machine Ethics Podcast. This time, we're joined by Sarah Brin, and this conversation was recorded on the 12th September 2023. Sarah and I talk about the types of AI, AI, process of making artwork with AI, how is an artwork culturally valuable in of itself, curatorial practices for AI art, collaborative artwork with AI, AI in games, and much much more. If you've enjoyed this episode, you'd like to find out more, you can go to machinedashethicsdot net. You can contact us via email, hello at machinedashethics.net.

You can follow us on Twitter, machine_ethics. Instagram, machineethicspodcast youtube, youtube.comforward/at machinedash ethics. And if you can, you can support us on Patreon, patreon.comforward/machineethics. Thanks very much, and hope you enjoy. Hi, Sarah.

Welcome to the podcast. If you could please introduce yourself, tell us who you are and what you do. Hey. My name is Sarah Brynn. I am an expert in digital creativity.

I have a background in art history, especially surrounding newer technologies, and I also work full time at a video game AI company called Caithera AI. Awesome. And I was extremely excited to get you on the podcast because I had this notion that we've had a couple of episodes, which kind of touch on art and creativity in AI. And since then, we've had literally an explosion of, like, creative AI services and products and, large language models, and image stuff, which has really contributed to this conversation. And I think we I really wanted to revisit it.

And your you put yourself, above the parapet. And I thought that you were, like, literally the the perfect person to to have a chat. So thank you very much for joining me. I wanted, to ask you about creativity. But the first thing we always ask in the podcast is what is AI?

Oh, gosh. So the first the first part is is is easy. Right? So that's artificial. Right?

So so for the purposes of of this conversation, we can see say made by a machine. Intelligence is the harder, trickier one. Right? So we can say that intelligence might mean the ability to solve problems or complete tasks, and that's that's, like, the the shorthand I often use. But you might also know that there's a whole bunch of different types of intelligence.

Right? So there's, like, kinetic intelligence. There's interpersonal intelligence. And I and I think that the the current definition of AI, has a very limited definition of what intelligence is and can be. It's it's interesting because I see parallels a bit with you had this, recent talk at Gamescom this year, and you were talking about different types of fun as well.

Lots of different types of fun. And it's very hard to actually define some of these words in a succinct way. And you have to it's multifaceted and stuff like that. So, are you kinda referring to, like, the kind of the Boston Dynamics robots? You know, they are maybe, let's say, intelligent at kinetic stuff.

And then you have things like, stable diffusion or whatever, which those systems might be more intelligent in this image or, like, interpretation sort of way. Oh, there are loads of different types of AI as well as I Yeah. Know. Right? So there's, like, spam filters are a type of AI, and that's something that that does one thing really well.

And so and then, yes, there's also generative AI, which is the one that's getting a lot of, like, attention right now. And then there's there's there's a type of AI that I work with in in my day job, and that that's, very much oriented towards solving a problem or completing a task. And then there's also stuff like theory of mind AI, which is, you know, does does the AI itself, like, have a a, like, a an awareness of itself for or then there's self aware AI, and it's there's just so many different types. So Should should we clarify that there there isn't self aware AI? There is not.

Yes. Good. There is not. That is correct. But there well, according to a study that came out a little while ago, there there there might be an AI that has theory of mind equivalent to a 9 year old child, but, I think there's there's there's more to be discovered there.

Yeah. Yeah. I don't think I've read that one, but that sounds like there's a lot of nuance there because I have a 6 year old child. Yeah. Yeah.

And that's sort of terrifying that there might be something that is more let's say, more powerful. I'm doing inverted, like, hand quotes here. Yeah. More powerful than that is slightly unnerving. Right?

Yes. Yes. I think about my my dog as well. Like, maybe has a level of intelligence of a 3 year old child. And and I I don't yeah.

I'm it would be very, very scary. But, yeah, it was it was GPT 4 that that some some researchers decided that theory of mind might have spontaneously emerged in it. But, who knows? I think oh, this was just a couple weeks ago as well. I think, we could definitely go down the rub hole of talking more about that.

But I want to kinda segue us back to, the more creative endeavors because we have this idea that, like I say in the beginning, we have these new tools and there are new issues or impacts with those tools that people are either literally feeling or are, you know, on the horizon somewhere. Mhmm. And what I wanted to get into really today was this idea that we have human made things and that we derive value from those things. And I'd love to have your feeling about how that change that relationship changes when actually the the work or the the product, the creative endeavor, the the thing that is being produced either has, let's say, a little or limited human input. Right?

Mhmm. And should that change how we feel about the artwork, or how we, talk about it, or how we, obviously, there's there's legal ramifications that we're going through at the moment. But just just on that point, do you have any kind of gut feelings about how our situation is changing because of these facts? Well, it is and it isn't in some way. So, like, a lot of my work in an art historical context and working with, like, large cultural institutions like SFMOMA and New Zealand Contemporary Art in Los Angeles is around working with artists made video games.

Right? And I think or and I've also done a lot of work with digital fabrications, like 3 d printing and laser cutters. So I think it's very normal for a lot of people to say, hold on a second. I can't see the hand of the artist in that. I can't see the hand of a human, meaning they can't see a brush stroke or, like, the role of the human in the process.

They just see the object. And I think in a lot of ways, that's a curatorial problem. Right? So, like, a lot of art and design focuses specifically on the finished object. But a lot of what's really interesting in our process, art to me is the process the process of making the artwork.

And so when we start to think about how humans and AI work together, what's really interesting to me is that process, that back and forth, that exchange, and not so much the finished object. Let's say that you are let's hype let's make a hypothetical situation for you here, Sarah. So So you're in a gallery, let's say. I mean, there's a very contextual space already. So we're walking into a gallery.

We're expecting certain thing to happen to us. There's probably white walls. There's some writing on the walls somewhere. And you see maybe a a large painting, let's say, because that's the very easy medium for everyone to kind of grok, have in their mind. And it says just, you know, has a title and has a name, and and maybe that's what you get given.

Let's say that that painting or that image, was created for a system. And let's, for argument's sake, say the the individual artist had little to do with that. Maybe just they put in a prompt, And the first thing came up, they printed it really, really big. And it's it was awesome. Okay?

It was so awesome. I I think I'm trying to get at it's like how is it okay that people still get a lot from that painting? Right? Like, the aesthetic of it maybe. How do they feel when they view it?

You know? That sort of thing. The emotion are they emotionally charged? And, actually, should we care about the fact that it's produced by AI and and less about that and and care more about how the visceral reaction of being human being in a space viewing something is, you're nodding. I'm just gonna First of all, I'm a populist.

Right? So I am kind of, like, art historian and art worker who understands that a lot of people feel alienated in museums and I am the person who's here to tell you it's okay to like what you like. It's okay to not like what you don't like. And the same goes with wine. The same goes with cheese.

It's the same thing. And so, yes, it's okay to like things. And there are also really different personal reasons why we like things. And if that is that that thing is created by a machine, fine. Great.

However, there are also mechanics in terms of how we decide if something is culturally important or not. And, so, who are those arbiters of importance. Right? So that's gonna be museums, that's gonna be art critics, that's gonna be, press. Right?

And so usually when, one of these organizations is deciding if something is art historically meritorious, there are criteria, that they apply to the work. So is it art historically valid or even art historically interesting if it's just, okay, a computer made this? There's not much of a story to tell there and in terms of deciding, okay, if this is this valuable, is this culturally valuable, is it monetarily valuable? That's a huge part of museums as well. Yeah.

Is it is is a really important question for to to ask. So, yeah, there's a little bit of a distinction between high art and everyday art. Yeah. Yeah. Yeah.

So there's a there's a validity to its kind of presence there in the gallery itself, like, and and and how that's, what the mechanisms of that are. Mhmm. Is by extension, do you let's take it to a little bit more, like, I keep saying visceral, but, like, in music, there there is this response as well. And I'm going to say that for me, listening to a bit of a music produced by some sort of AI system probably has the same reaction. Right?

They're they're not. And the the only kind of thoughts I have that on that initially, knowing maybe that, a piece was made by AI is, maybe that it's mediocre or fine. You know what I mean? Maybe it doesn't have the highs and lows that, maybe a a human made piece, or the inaccuracies may be. And I and I think when we talk about, like, appreciating particular artworks or pieces of music, it talking about the context in which the works are produced adds a lot to our understandings of and our feelings about the work.

So if, you know, we were talking about, like, a, like, a poem, you know, being part of the story is like, oh, yes. Langston Hughes, you know, was part of the Harlem Renaissance, and this is what life was like. And it the artwork becomes a piece of a puzzle, though, our overarching story. With AI created works, I think we need to tell a different type of story. Doesn't mean that story can't be rich and interesting, but, we need to figure out how to tell it and what it is.

And, again, that comes back to, like, this curatorial issue. Right? Mhmm. I remember because I I used to be interested in interested in art. Should I say no.

That's bad. I am still interested in art. Yes. Good. But, I had this this problem, I guess, back when we were talking about new media artworks and how to present those in in the gallery, or present those to in different spaces, and and how, for example, Internet art or, screen based media could be presented.

Right? And I guess that that that's another extension of that. Like, how how should we tell this story of this artwork or this creative practice? What does that look like? All these different things that, this new technology is enabling.

Absolutely. And so something something I I really enjoyed doing when I worked at the Pure 9 Digital Fabrication facility. So it was a a building in San Francisco that's run by Autodesk and is full of all this, like, really sophisticated digital fabrication hardware. And so what I would do when I was curating the artists in resin shows, I would show the artworks inside of the machines that produce them. So that way I could have a water jet cut sculpture displayed on the Omax water jet cutter, which is like this massive massive machine that has basically, like, a nozzle on a gantry that moves back and forth that sprays highly pressurized water and sand that can cut through 6 inches of steel.

So and it's also, like, in this like a kind of big big bed that looks like a a nightmare swimming pool. And so being able to, like like, show those things in conversation with each other really helps tell that story of how the thing got made. And I'm sure, you know, if we were to think about it a little while, there's there's so much we could do territorially to to tell story and provide that as well. However, that's not really conducive with that kind of white box of museum that a lot of us are are more or less acquainted with. Yeah.

Yeah. Yeah. Exactly. It depends on how thoroughly you would like to invite that into the gallery space. I always find that there's quite a few shows at the Barbican I went to over the years that do that where they may be curating the process as well as the, the artworks themselves or at least having extensive information about it.

Yeah. It's funny because you you kind of often there's this other tension, right, where you might have some sort of reaction to a piece, a sculpture, let's say, for example. And then you read the title, and you read that it's from AI. And maybe you have now a different response, and you feel differently about it. And it changes your perspective on on on the scene, on on how it is as, if you didn't have that label.

But, obviously, that's always been a tension in gallery space anyway. But, I feel like more so when it's like the hand of the, artist is is less prominent. Absolutely. And and, you know, sometimes sometimes seeing that AI is on a label kinda add to an experience, and sometimes it can challenge an expectation. I mean and also very few people read these labels, unfortunately.

Because sometimes sometimes you'll look at a label, and it'll be like, talk about, like, the components, and you'll be like, oh my god. The art is so good. Yeah. Yeah. Yeah.

Yeah. And it's like, woah. So, you know, I think it's it's interesting to to have that context, and people will bring whatever they they want to the interpretation. Great. Okay.

We we've solved that. That's it's we're just gonna get we're gonna get all the curators in and then just kind of, like, load onto them. They'll be like, go nuts, guys. You're ready. Yeah.

That's it. Mark. Do do you think do you think it's important for those, types of spaces to really understand a bit about the technology to be able to display it in a way that is gonna be communicative? You know? Absolutely.

And museums are usually the last to integrate emerging technology. So, usually, you'll see it, something a technology pop up in industry, then maybe some early adopters in the creative sector. Military is also unfortunately likely to be an early adopter of emerging texts as as well, but, and and then education and then cultural institutions. So, and that's why, you know, sometimes you're in a museum and you're like, oh, this interactive is broken. Because, yeah, they they often don't have the resources or the infrastructure to to keep up with timely things.

A lot of museums are meant to to collect things that are static, like sculptures and paintings, but you don't have, like, the, the operating system for painting go out of existence. Right? You don't have, like, an an engine stopping supported for a sculpture. So, yes, it it takes a a lot of cultural change in education to to get a whole organization on board with a new technology. I've I've I've done that firsthand with with video games, and it's a lot of work.

Yeah. Yeah. Yeah. Is that with, like, archiving video game stuff and presenting that, you know, out of date? Or Out of title stuff, visitor services.

So, like, the people who are working on the gallery floor, like, they need to know what to say. They need to be able to restart something if it You know, they're they're curatorial, of course, and then there's security. There's there's loads to to factor in. Let us let's sidestep now to the kind of impending doom, right, which is, like, the generative AI is gonna take our job situation. I like doing like, I like coding.

I like, having that kind of intellectual puzzle, which is, creating some, something that that works. Right? And it feels like lots of people who were concept artists or illustrators or photographers are being essentially slapped in the face with this kind of ease of, production, which is kind of, getting in the way of maybe work or, or in fact, you know, taking some of their their work and, melding into the system to produce new artifacts. So I know you mentioned in that video that that I mentioned earlier at Gamescom that you you're pro unionization, for for the maybe the games industry. Does that extend to those sorts of people as, creative art workers, let's say?

Absolutely. And my you know, we can see what a powerful force it is for the writers guild and the screen actors guild to unionize. They've shut down a whole industry, and I am heartened by the unionization efforts I'm seeing in the US, at some Amazon warehouses and Starbucks, etcetera. Yeah. And it is my hope that they'll they'll take a page from that book.

I think if you're working as a freelancer, it's a lot harder to think about. However, there are freelancer unions. But, I think it's it's important for for people to make it Technological advancement doesn't have to be at human expense. Mhmm. But, it's, unfortunate that not everyone thinks that works.

Yeah. And it's not always distributed evenly as well. Like, someone who is, you know, a concept artist might be impacted, initially today, and then someone further down the line, developers, whatever, doing kind of basic gameplay programming, maybe more so tomorrow. And he's making sense out of that stuff from an economic point of view, which is, what we're grappling with at the moment, I guess. Yeah.

We're I mean, what we're seeing is happening in a in a few different channels. So there there's a there's a drive an economic drive by studio heads in major companies. And I I think they're not necessarily thinking about the craft. They're not necessarily thinking about the process. But an interesting side effect of this is that if we keep on training datasets on things that exist already, there's only certain types of innovation we can get.

So it's it's gonna be more of the same, and we're gonna see things in in different combinations. Mhmm. Yeah. But, we're we're taking a certain part of human innovation out of the picture. And this doesn't necessarily mean that it's show over for concept artists.

I I do think that it can be an opportunity to complicate and evolve new types of art practices. Just like when, you know, photography was invented, a lot of artists were like, oh, no. Painting is over forever. And then what they did instead is they kind of move more towards abstraction. They move towards expressionism, and it kind of pushed the medium forward.

What is different about a lot of types of AI is that the the impact of AI is is not just limited to a specific field. It is it is across seemingly everything. And, again, that doesn't need to be a bad thing. We need to have certain protections in place to make sure technology can serve people. People.

Awesome. Yes. I mean, my hope is that we are creating a better situation, you know, at the end of this instead of destroying the, wealth of stuff that we had. It'd be a disaster if we look back and went, oh, the naughties. They were like just the pinnacle of human, the human epoch.

And and now it's been terrible ever since that AI stuff was, came in. We're obviously doing the wrong things if that's that was the case. Do you do you see many people kind of really embracing the technology at the moment and getting a lot out of it, in terms of those creative practices? Yeah. And I think there's there's different types of of enthusiasm around it.

Right? And so there are I always like to point to something called the Gartner hype cycle when I talk about about new technologies, and what that does is it shows kinda gap in the hype around a a new technology and what it can actually do. And then there's always, like, the a point called the trough of disillusionment in which it Mhmm. Do something. So I think, you know, there's there's some hype, and I I think that, people might be disappointed about where AI is at in terms of its ability to generate certain things, certain things it does really well.

And then there are people who have been using AI for a long time to complicate complicate and complexify the practices like Seungkoon Chung, who's been working with, datasets and industrial robot arms for many many many years now. What she does is she trains robots based on her gestures, she's drawing or painting, and then they draw and paint along with her, and they make collaborative artworks together. Or there's another piece that I love so much. That's a a game artwork called how to not get hit by a self driving car in which the AI itself is the mechanic, and you win the game by tricking the human detection in an AI into thinking that you are not a person. And, of course, funny in that situation is if you were to effectively trick the AI, you would get hit by a car.

But and it's also really interesting because in the process of playing, you're training the AI in order to work better. So, yeah, I I think there there are loads of of artists and and creative people who have been using, AI in really, really interesting ways. But for me, I think the most interesting projects come with, like, a back and forth between the player. It's not just, okay. Press a button on the AI, and it's done.

Mhmm. It's interesting to me when it's a conversation, when it's a back and forth. I mean, I think that, you know, if you think about AI and artwork, there is a lot of people think about art as an end product as a visual object, and I think that what I'd like to see when we talk about the evolution of different art processes is under evolving our understanding of what art can be. Mhmm. If you're a person like me and you've gone to art school for a 1000000 years, like, you understand that performance can be an art form.

You can understand that, like, you know, reading some instructions, like draw a straight line and follow it. That's also an artwork. But that is like a, I would say, like, a fringe weirdo opinion. Yeah. But, what's interesting to me about when you have creatives and artists working with AI, we have a discussion about a technology that isn't led by industry and it's not led by military.

Mhmm. And those are tend to be the biggest factors about how products take shape and manifest in our world. So I do hope that artists continue to experiment with AI and have more opportunities to do it So we can continue to invite broader audiences to consider different perspectives of this technology that is going to affect the ways we live, Whether we like it or not. Hopefully a lot of those ways will be positive But we need everyone on board So it's not just like the Jeff Bezoses of the world who are deciding what this technology is. Yes.

Yeah. Yeah. Yeah. I think part of what an artist does, I guess, is reflect what's going on in the real world in a way that is, saying something. Right?

You know, sometimes it might be, you know, check this bad thing out, or check this beautiful thing out, or I have this feeling and I'm gonna express it in these different ways. And art, generally, and and visual arts, especially, have this dialogue that they they can use with, you know, showing something about our current situation. You know? And I guess from the AI and ethics point of view, it'd be really nice to get more of that. You know, the example you gave earlier of that game and and the automated car kind of detection game, That's all that's that's there.

Right? That's like, you know, these things are coming in. Let's, like you say, like, teach you about it and also try and train it in this fun participatory way. Like, that's really, exciting and and cool. So it'd be nice to see more things which were enabling people to kind of understand what's going on almost in in a in a new way.

Sounds great. That that's one of the things that sounds exciting for from what you, mentioned earlier already. Yeah. And when you have artists, you know, pushing the boundaries of newer technologies, that also serves an r and d function for the company itself. So you can in the process of doing that, you can find new, yep, use cases that can potentially be monetized.

So it's not just like this, you know, this this thing for the good of the world to have artists making art, but artists are gonna bring new approaches to tech that, the limited amount of people that people are considered as users for a particular product aren't going to bring to the table. Mm-mm. It kind of reminds me of speculative design as well. Just having these art artworks or these, these items or real world objects tell a story about, you know, in a way that Black Mirror does. Like, what where are we going?

This could be an artifact from that place. And isn't it interesting or terrifying or, amazing? And we can have good dreams of AI. It doesn't have to all be black mirror. It's it's an easy easy touch point.

Right? But you're you're right. You're totally right. Yeah. And and, you know, it's it's it's much easier as even for me to to go down that road.

But, like, what if what if we could imagine? What if we could imagine the the world we want? That'd be awesome. We should try. Yeah.

I'm I'm just hoping that the world that I want is also the world that other people want. You know what I mean? You know? We'll have a dialogue about it. We'll see what we get to.

You know? Awesome. So before we ask the last question that we normally ask in the podcast, did you wanna briefly talk about how the the AI, some of these AI techniques, but also maybe AI historically have worked in the games industry as that's where you're you're currently spending a lot of your time and and have, curatorial practice there as well. Absolutely. So in video games, we have something called NPCs, which are non playable characters.

And, actually, if you've been on TikTok lately, you you might see people pretending to be MPCs and repeating, like, weird phrases like ice cream so good, for an over. But I you know, if you played a video game before, chances are you've encountered an NPC. Those are the characters that you often give you a task and say or they set context to be like, oh, last cowboys across that ridge we haven't seen again, or, I heard there's gold in their hills. And, I'm so good at writing video. And we also have enemy NPCs.

Right? So those those are those are the bad guys that are gonna try to to catch you or capture you or shoot you in your video game. So that's actually AI. That's not that's not generative AI. That's that's kinda more like machine learning.

But, yeah, that's that's what the company that, I work for does, things which helps developers with things like marking up digital terrain to see where is navigable and where is not navigable, Doing that for 2 d and 3 d dimensions. Also designing the behavior for different types of MPC. So is this MPC gonna remember that he's passed you before, or is its approach to you going to change based on the fact that you threw something at him? So yeah. So so NPCs have been around in video games for almost as long as as video games themselves.

So when people get really excited about AI, but it's this particular technology has been around for a long time. Mhmm. It's very well, but it's it's it's not the same as chat gpt. Yeah. It's it's kind of like a we talk about all these things in the same umbrella, but they we've got all these different techniques.

And almost the different industries talk about it in different ways. Like, games have been talking about AI as a term for kind of, automated behavior for since the beginning of of games, really, computer games. But that's definitely not like machine learning or or some but machine learning has all these other techniques in it as well. Yeah. Yeah.

So it's like it's very muddy this area. If you I'm gonna come back to some of the episodes that I recommend on some of these subjects at the end. So so, Sarah, thank you very much for joining me on the podcast. The last question we always ask is what excites you and what scares you about our AI mediated future? I am I'm sure everyone says this.

I'm excited for AI to possibly help us solve the climate emergency. You know, I saw a tweet about this not too long ago, and it was like, why does all AI wanna be a screenwriter, or can some of that help us pick garbage out of the ocean? And so I would love some of that good AI that can pick garbage out of the ocean. And, yeah, the the stuff that scares me is, you know, we've seen a lot of tech companies that prioritize economic growth above anything else. And so why would the approach to AI be different, which is why I've been outspoken about labor rights and protections for individuals whose work might be impacted by AI itself.

Thank you very much for your time. How do people find you, follow you, all that sort of thing? Oh, this is so uncool, but I have LinkedIn. My my name is Sarah Bryn, s a r e h b r I n. You could find me there.

I also have a website that's pretty good, I think. It's ww.com. Thanks. Yeah. I I please please give me a shout.

I, love to to meet interesting people who, care about anything, really. What an invitation. Cool. Well, thank you very much. And we'll see you next time.

Thanks, Ben. Hi, and welcome to the end of the podcast. Thanks very much for bearing with me. I had extremely creaky chair in this episode. So I'm really sorry about that.

That came through a few times. So I'll try and get on top of that for next one. We also mentioned in the episode, there are other episodes you should check out. So if you're interested in more AI and art, you can check out episode 41. We talk a lot about speculative design in episode 55 and lots of game stuff in episode 65 and a few after that as well.

Thanks again for listening. If you can support the podcast, please go to patreonpatreon.comforward/ machineethics. And you can sign up there for some extra bits and bobs. And you can support the podcast continuing into the future on AI and ethics, tech ethics, all that's all great stuff. And also, if you'd like to get hold of me for talks, workshops, or anything like that, go to hello at machinedashethics.net or just say hello, and that'd be great too.

Thanks, and I'll see you next time.