 Welcome everybody. Thank you for coming out today. Thank you to Elena for for having this. I think yeah Let's give a round of applause for Elena So yeah, not only is this a lot of work, but there's a lot of risk involved with doing something like this too So I really admire Elena and her team for putting this together and I think with without these we don't really have the community and You know digital art doesn't grow the way it should so very appreciative for her for her setting this up So I'm Jason Bailey. I run art gnome.com it's a blog that talks about the intersection of art and technology and I also Started a blog last year called the dank rarest podcast that I get a lot of questions from older people about because of the name but that would focus mostly on blockchain and art and I'm excited to be here today to talk about AI art mostly though blockchain may Come up a little bit and generative art So one of the things that that I've noticed as sort of I'm a long-term art nerd geek This is like I got my my master's in art and tech in like 2010 and grew up in a family of engineers So this isn't a new thing for me, but it's exciting to see that other people are starting to care about this stuff So, you know one of the things I want to touch on today is that there's before all of this AI art there's been this long rich history of generative art that a lot of people don't necessarily know about and somewhat for selfish reasons I'm seeing this window of everyone being fascinated in AI art as an opportunity to kind of force feed everyone the history of Generative art because I think it's really important and we've got this opportunity to do it So I'd like to quickly introduce my panelists, but they'll kind of introduce themselves a little bit more as we kind of walk our way down the line, but we've got My co-curator for an upcoming show automaton munch George Bach We've got John Watkinson who is best known probably for crypto punks But we're gonna talk a little bit about auto glyphs and his partner in crime mats and the second row there David Young who is an AI artist who will be talking about some of his work on the panel today Sheeran who I don't know how to say your last name yet Anlin Anlin all right Sheeran Anlin who I met last Friday and is doing awesome stuff and changing my mind about AI art and Then Kevin Abosh where panel veterans together. So excited to have you back on the panel so I'm gonna start by sort of having a Targeted question for each person you can all weigh in on any of the questions and then we'll kind of open it up from there To a broader conversation So George we're lucky to have him here He's sort of an expert on early generative art specifically in photography So maybe help folks understand a little bit more about where this all started and a little bit more about the history to Give some background before we kind of go further down into the panel Thanks Jason well generative art that is a That was a movement in the mainly in the 60s and Generative aesthetics is an expression which was defined by our German philosopher mathematician named Max Benze Who was teaching in Stuttgart? He was also? later teaching in Ulm when Max bill has invited him to become a professor there But he was very influential because he defined art as something Well, it was a time when the first computers came up and So artists were thinking about how can you create? visual aesthetics with algorithms with computers machines and So he created this expression generative art and he talked about Aesthetic states so he was not talking about artworks now. This is important because When you talk about generative art It's basically important to know that there is a program behind it, you know a program that Creates the art or a machine that creates the art so you can basically recreate I'm sorry Better is it better? so so We're talking about aesthetic states and these are basically Visual effects or visual artworks that can be recreated anytime Just by using an algorithm or by using a machine that recreates them Now This was very interesting at the time and in 1965 two of the pupils of Max Benza There were two German artists skilled nays and Friede Nake they did an exhibition with a Plotter drawings which were made on very early computers now Those artworks well nowadays if we look at these artworks we think wow It's just like a very simple print, but at that time these plotters Which were printing these artworks. They took hours and hours the artist didn't just have Access to these machines. They these machines were owned by big companies like Siemens Bell laboratories etc or universities and So the artist sometimes just got access For example throughout the night to work and to print and it took sometimes six hours to make a print Which nowadays takes not even a second So and they had to create their own programs and everything so In 1965 that was the first exhibition in the history of computer art where those two artists displayed their works in the University of Stuttgart and later on There was a whole movement of early computer art or generative art and In 1968 There was a some sort of a peak because a lot of artists Wanted to display their work so there was a show in London for example Cybernetic serendipity which you might have heard of that's a like a classic computer art exhibition Curated by I don't know how to write say her name, but it's Yezia Reichard I think if I pronounce it right and She she curated an exhibition with computer art and Well, it was also that there were also artists who did geometric art Kinetic art etc. So that was a seminal show at that time and At the same time there was also a movement in Zagreb the so-called new tendency movement Which is Yeah, I mean Was a place where a lot of artists met from Eastern and Western Europe Europe and Who were thinking about? Computer art about new aesthetics. It was very influential. So these were maybe the two most important art Exhibitions at that time But also at the same time there were the first magazines created like the Leonardo magazine Which is a very important source for early computer art There was a magazine called bit bit international and Also in London they They founded the computer art society, which was also collecting art and Thanks to this computer art society the V&A Could buy their whole collection They have preserved a very important source of these early artworks my special field is actually generative photography and That happened also at the same time in 1968 there was an exhibition called generative photography it was in in Bielefeld in Germany and That was actually the first exhibition and manifesto where they used this expression generative and It was got three Diego a very influential and important photographer who Showed his works and also the works of Pierre Cordier Kylian Breyer and Hein Gravenhorst and They were basically the founders of this art movement, which is now In the last few years they started to rediscover these artists. I mean they were quite unknown for a long time, but People started to realize that there was this was the beginning of digital art digital photography and Well, I could A solid intro will probably come back to more of the history, but I think from from there There's a continuous history. It doesn't just start and then stop in the in the you know late 50s early 60s One of the pivotal points that you know, I think got a lot of maybe the people that are here interested In the late 90s early 2000s flash comes along action script, right? So artists For the first time don't have to be engineers to be able to start programming their own work and then processing Which is a programming language built by Ben Fry and Casey Reese out of MIT comes out and then it just explodes Right, so people are using processing for data visualization and all kinds of art and you can cook You know hook it up to sensors and things like that So I think you know one of the things I'd like to do is what we'll ask John here So John how many people here know what a crypto punk is or what crypto punks are yeah? Solid crowd. He's like all your relatives guys are So crypto punks is like a sneaky generative art I call it sneaky generative art because you don't actually have to know what generative art is or even know that It's generative art to appreciate it, but I think it's totally generative art I'm gonna ask you if you agree in a second here and then they've more recently done this auto glyphs Project, which is actually a tip of the hat to a lot of the folks that George was just talking about so John Maybe you can tell us a little bit about crypto punks your thoughts on generative art some of the influences for auto glyphs Yeah, absolutely. Yeah, so crypto punks is generative art. It's You know these works are generated by a computer program I guess the unflattering way to describe it is a mr. Potato Head, you know, or it's just like all right Take a random face, you know, put a random hairdo on it, you know that sort of thing So it's it's a very simple type of generative art, but that's how we got to be able to make 10,000 of them and and then yeah the real sort of Important thing I guess we did there that the reason we did it that what mattered to us was that we We we registered them their ownership on the blockchain and made it so that the way you own these things was Control them by a smart a smart contract on on a theorem and at the time when we launched it We didn't know if anyone would acknowledge that as real ownership, you know What does that mean? You know that anyone can download these things look at them do what they want with them So does it matter if you if you own this little record on the blockchain and then the answer was happily yes that people did care about it and and They sort of took off and they're still traded usually a few crypto punks change hands every day two years later almost and So that's that's them and then to speak briefly about auto glyphs, which I think have been rolling through on the slides here that was our idea to kind of it's like can we go even kind of deeper on this on a similar concept and Try to make it even more pure And so what we did there was make the entire generation of the art happen within the blockchain in this the smart contract and People I've heard Ethereum and other blockchain Networks referred to as like the world's super computer, you know people and but really it's it's not that it's the world's punch card computer You know me that you can do so little it's so expensive and so time-consuming to even do a very small amount of computation That in a way it really is a throwback to what Georg was talking about where you know, it's quite difficult to To to run these things you have very limited resources and so because of that yeah, we thought oh, you know, what can we do? With this very limited computing resource We can actually take inspiration from what they did in a very similar constrained environment in the in the 60s and people like Vera Molnar and Michael know and and some of some of the artists that Georg mentioned so so out of this is definitely an homage to that and And we love the purity of that everything comes from within That smart contract environment so the entire art is created sort of in public view by the by the You know by the participators in the network so part of it's a hat tip a part of it's just sheer limitation of the Confines that you were working within right yeah, that's right And yeah, I mean the entire generator is only about 30 lines of code and it's I think we were limited It's only seven variables that we you know it's very limited and and it still takes it still takes almost an entire Chunk of they get uses more than half of what you're allotted to to fill in higher block and a theorem so And at the time when we wrote it it was even closer to the limit the limit pushed up a little since then so So yeah, it's it was hard for us It took us many iterations and we even gave up for a while to try to make what we wanted within that constrained resource So yeah, they were nice enough to let me write the sort of the exclusive Article about auto glyphs and as much as I love generative art like most people I talk to glaze over pretty quickly So I was kind of like these poor guys. Nobody's gonna buy this stuff, right? So they put it out and within two or three hours all 330 some odd were sold is that right? Yeah 384. Yeah, there was 512 total But we kept a quarter of them for ourselves so we can make prints and do other things with them And yeah, we were surprised at their reception Because is this what you say? We said this is pretty esoteric. These aren't nice characters anymore. These are Yeah esoteric kind of unique little patterns and But somehow it yeah, it struck a nerve and It's cool. Yeah, I think you know, that's what we're seeing with and what's when we'll talk about the GAN stuff And you're not works up to that that there is a bit of a coming-of-party for this generative art and people are It's it's resonating more. I think that it has ever before I love it This could be like our only shot to be cool because you know, we've been loving this stuff for so long So I'm sure that when it's like fingers crossed, please so David I I always smash AI art under generative art But a lot of AI artists that I've talked to a don't like to be called AI artists and be Point out that they're sort of different I think it'd be helpful and I know this isn't easy if you could help people understand I tell people when people say AI art in the last two years what they really mean is kind of GAN art, right? So these generative adversarial networks, which I can't really explain that well Maybe you can help us understand a little bit what what this AI art is that people are talking about and you know I know not easy so when I did this in Bahrain Robbie Barrett and Mario Klinchman flipped the coin to see who had to explain it because it's It's just not easy, but I appreciate the any help I can get there I appreciate the question and if you want I can connect a laptop and bring code up on the screen for everybody to look at I mean, I think people don't really like the term AI artist because it's a little too Specific I think instead people would like to think of themselves first as artists and AI is the tool that we're using Just as I'm not a well, I guess you could say I'm a painter as opposed to an artist that uses paint So okay, so maybe I and unsay that But I think that you know AI aren't using AI fits into a long history But that what's interesting now is how AI itself has changed over the last five years from what it was previously so when I started my career I Actually started in computer science and doing AI in the late 80s and The AI in that era before there was this kind of nuclear winter that kind of ended all the AI work that was happening at the time Was around sort of this notion of sort of top-down Defining of rules so expert systems were a big part of what AI was in those days And so people were embedding rules into the code to kind of in some way embody expertise And I think of generative Generative design or generative art as sort of similar that generative artists were Embedding rules into programs and so they were you know, generating the the images based on this kind of rational approach to image making And then what happened most recently with AI in the last five or six years as the emergence of neural networks And it's a really different notion of what AI is as well as what art made using AI is So if you think of previous versions of AI or previous versions of generative art as being a kind of top-down approach Where the the programmer or the artist is sort of embedding rules into the system The new notion of AI is around training and so you may set up some kind of Framework for the system to learn and you set up what information you're feeding into the system But what it what it kind of learns what it develops is much more Organic sort of as the nature of a neural network and much less Understandable by the by the AI team or by the artist and so the work that emerges has a very different kind of quality And I sort of go back and forth between this is both AI in the Industrial sense and also AI in the generative art sense and so the works that come out in the in the art space are kind of I think are much more They they're part of a tradition of generative art But they're they're also a bit of a break because it's a new type of expression that's coming out of it No, I think that's really helpful in terms of the background. So it's not AI didn't come out of nowhere, right? So it's people have been talking about AI since the the 50s I think and it's just its definition has evolved and now arts evolving along with it, right? So I you know I might pass this question to the rest of the panel too but one of the things I want to get back to you on is a lot of AI artists find these massive data sets online and train their work, you know Using huge amounts of data and they're constantly racing for what I call Firstism which is like the plant the flag and be like I use big ends first or I use this first And I think you've got a slightly different approach But maybe we'll come back to that and kind of catch all three AI artists. Sure. Yeah. Does that sound good? Okay, great So Sheeran We talked on Friday for anyone who reads art gnome. I have sort of like this personal vendetta Against anthropomorphization, which is probably not even a word of AI I feel like so I I didn't say it in my intro But I worked for a machine learning and AI company based out of MIT for the last four years So even though I'm a nerdy art marketing person. I've been around legit You know machine learning folks from MIT and Harvard for the last four years and watching them try to do anything with AI It's like it's embarrassingly simple or like the embarrassingly simple stuff It's really hard to do with AI But then I come home and I like read the news and it's like all our jobs are gonna disappear next year and like You know all artists are getting replaced by like autonomous a You know robots or whatever and none of that's true And it drives me crazy right because there are actually these AI artists that are doing like remarkable things But we never get to the nuance of what they're doing because everyone wants to have a debate about whether or not robots are taking over the Planet with art or so it's it's frustrating So then when I was talking to Sheeran and she was like oh well I look into sort of the intersection of emotions and AI art and I'm like no, please don't do anthropomorphization or whatever but Immediately was excited when you started explaining your work because I think you open my mind You're saying well, you know for some reasons that you'll probably share You've had to think about emotions in sort of a structural almost programmatic way And now that you're thinking about AI as AI evolves Do we want to carry everything that we know or think we know about our minds into these models that we're building and Can it cause us to reevaluate things like, you know, what mental illness, right? So what does that look like an AI? So I'd love to have you share a bit more about that. I think your works fascinating Yeah, it's funny that the only woman here is speaking about emotion I'll talk about emotions too Okay, good Yeah, so what I'm researching is about the possibility of mental illness in AI and I'm looking at the What happened between input and output and the distortion between there and I think it's what make us human is like Me being happy to me being sad something happens there and I'm looking at that in the machine And I'm exploring different kind of machine learning models and looking an error bug in bleach and asking whether that can evolve to some kind of mental state and maybe that's already kind of like a Hint for something that can evolve to if it would be like independent thinking entity what kind of independent thinking entity A dysfunctionality of thinking independent thinking entity will grow out of that. So that's my question and I'm looking at that From the art point of view, although it sounds really hardcore Technical question. I'm speculating about the possibility of that and I'm looking at gun for example. What is my first? Network that they explored and I speculate what kind of mental illness gun will evolve to help So I started really big and I immediately diagnosed it as a schizophrenic because he had He has a two network that he he competed with each other like one is the country debating what is real and the second one is Generating new image and fake image, but then I I had a chance to exhibit a prototype of this project. So I I created a prototype that's exploit in a small assumption of like gun he has gun generate fake memories of fake family and the data set that we train was about on Perfect family data set and the user came and played the perfect family But the gun view them through the perfect family data set and actually the distortion was the result We're very distortion and the story through the text that this is an example But we we developed a text to image that as the user speak Guns trying to understand what they're saying And he view them through the data set of the perfect family data set And it was a kind of a room that evolved through the generative sound and everything was created through Generative gun model. So the room was embodied Trying to teach about what is gun actually and how we can embody that. So yeah, too. It's not really a spiritual a Statement is really very grounded statement about how we can bought brought in the conversation and bring more people to understand what is machine learning and the next one is like Reinforcement learning models. So I'm exploring different kind of machine learning That's great. Thank you. Yeah, and I think You got me to change the way I think about AI and art more than a lot of people and recently So it was a good conversation. Appreciate that. Thank you. Yeah so Kevin I think a lot of people know you as a photographer sometimes as the guy who sold the million dollar photograph of a potato Or the forever rose. I think of you as a conceptual artist And I think you use a lot of different tools to get your points across your more conceptual points You've done a lot of work with blockchain and you're starting to do some work with AI that Folks can see downstairs. So the the works around the nails What is it that got you interested in looking at AI as a tool? And would you agree with my assessment that these tools are just tools and really you use the ones that best express your point Sure, so so I mean I've essentially been an artist my whole life But I also have this dirty secret as a technologist and in 1989 1990 I was exploring some of these expert systems that you spoke about We call it. We didn't call it AI. We called it fuzzy logic and you know, so fuzzy logic plus some Crude computer vision. I was using to analyze pap smears because there's a There was an inordinate amount of false negatives coming back for cervical cancer from women. It's a real practical issue and But this was you know, we and the limitations at the time were as they are today all about compute processing power So, you know, I've always said that for me the most Exciting place are is the intersection of art and technology when it when it works for me my work is about identity and value and existence and I I'm always using my subjects as proxies to distill emotional value With the blockchain I was able to do that too. I use these alphanumerics as these procs as a proxy usually to sometimes even a secondary and tertiary proxies to distill value AI in the You know form du jour with GANS with these adversarial networks is particularly interesting to me I'm not interested in training on these public data sets you talked about or data in the wild But creating my own data sets based on my own work If I can I you know, I'm always interested with if I photograph you or a potato For instance, where do I end and where does my subject begin and vice versa? It's really interesting when you consider when you're let's say we're collaborating with an AI not to anthropomorphism Maybe that's not a word but anyway the the Where do I end and where does the AI begin and when it's training on my own work? So in the downstairs, you see some little photographs of little one and a quarter inch nails the kind you hammer into a wall They're bent because What I did was I took 200 of these hammered them into wood Pulled them out made sure they had a little bend in them and then those photographs I trained on those the resultant generative output this generative photography, which also Starts you have to ask like what's is it a photograph? It's not computational in the kind of traditional computational sense, but generative based on these GANS I look at it and it's me. It's more me than me. I'd argue It's and that's really exciting is I feel like it's me plus some sort of insight and that's what it's about using this AI to glean insights That I might otherwise not be able to So yeah for the moment. I'm on the I'm on the GAN bandwagon as well on the GAN train to do on the GAN train So here's a question for everybody. I guess this would be a discussion question So with with both traditional generative art and with GAN art What is the art is the art and I get a lot of different answers to this Is the art the output or is the art the system or the code or whoever wants to take that to try to kick that off? It would be great What are people if people want to buy it, you know generative art or AI art should they be thinking like oh, man I really want the code or I want the output or how do you guys think of that as the artist, you know I'll say for the more traditional generative art where it is just a simple, you know a Discrete algorithm that generates the work There's some trade-off there right obviously if If you just have the code and you don't run it or anything or you don't use it Then that's certainly not what the work was intended You know you were intended to see the output But then again if you have something where you can just generate an infinite number of them at some point You know the million and tenth one is not really the art anymore, you know, you know, you know It's like you can't it's not infinite that way. So I think it's it's a weird combination of Definitely, you know that the code is the art and what it generates those together kind of form The result and and you and you they always have a limitation You know you once you've made a certain number of them then you just kind of know what to expect And you don't you're not going to be surprised anymore So so I think in certainly the traditional stuff I you know, I think that that the algorithm is you know is the work That's what the that's what the artist made but then it's you know, and then it's expression The way you experience it is with what it makes I can barely relate to that I totally appreciate it and respect it but from from where I'm coming from that's that's really interesting because I don't relate to it at all the the code is this tool. It's like buying a paintbrush or or some medium and Especially like in some of this AI work. I'm not looking for I'm not trying to win a Nobel Prize in science or technology You know with creating the the most kick-ass algorithm I'm actually looking at taking algorithms tweaking them breaking them and and and using them as a method to make art I'm also not interested in making two million iterations of something I might only be interested in making ten Mathematically perhaps what I'm doing breaks down if I maybe if I tried to make that 11th one I'd realize oh, it's it's a it's a bullshit algorithm But yeah, that's interesting that yeah, and especially if it's open source right if you're doing things in an open environment Where you're giving out your code then other artists will be will take that code and and do things with it You know so really you know that also I think elevates the code as as being more cord of the art if you're working in that environment You know so the autoglitz that code is open source and people can take that Change it adjust it or just run it again. They can do it They want to pick up on the previous talk about open source code I think we're all using stuff that at some point was either open source or from somebody else and where we're messing with it and there seems to be some discussion around is that ethical is that You know should we be monetizing other people's works? So it's like near impossible to figure out Where to stop it, right? I think a lot of people in the audience may or may not know about the Robbie Barrett piece So the obvious piece that was sold by Christie's end of last summer for $450,000 it was heavily heavily heavily heavily heavily based on Robbie Barrett's work Which was open source, right? So I don't think there's an opportunity for like a lawsuit there But like the court of public opinion was basically like well, this is bullshit, right? But that's not unique it happens pretty much if you're a nerd on Twitter with no life who like looks at this stuff all day Like me it happens pretty much weekly You know and I think there's maybe precedent for it in the software where we've had open source software for a while And you know do you give how far up or down do you give credit? Should you give credit to the person who makes the graphics card or the person who like makes them on it? Well, maybe that seems silly, but what about the guy who wrote the algorithm, right? So I don't David you have thoughts on that I mean I would sort of echo what has already been said that a lot of the code is public domain in part because in the AI world You know in the business world of AI what's valuable is the data not the code And so it's easy for companies like Google or Facebook to make the code available because they're the ones that own the data And so, you know if anything what we're doing is we're kind of taking sort of variations of that code or a code That's from the same spirit and then feeding it, you know content. That's our curation And then likewise, it's spitting out, you know depending on how we work Hundreds of thousands of images and you know our job is sort of one of curation Sort of like a photographer with a contact sheet that doesn't say every single photo that they take is is the one that they want To put out into the world, but to go through and kind of curate what we put out Yeah, that makes sense. Sharon. Do you have anything you want to add on that on that front now? No, I totally agree But I do have a question. Do you work with as open source as well? Do you bring back? Yeah, yeah Yeah, so who the question is who on the panel contributes to open source in addition to just using it I Don't At this point probably because I don't think I have much to offer I am working on some things that I think are pretty clever And if they work out and they they prove themselves to be as clever technically as I think they are which has nothing to do with My art actually then I I might feel some obligation I Don't think it has to be clever, you know, like we are in the middle of the point like we can just contribute together to the To the effort of making it like social and community and like work together to make it I get that I appreciate it and respect it. I'm just not sure I'm the type of artist who wants to collaborate with the masses yet Can I can I add to that? I mean, I think you know, I'm in a similar position as Kevin It's like I don't think I'm doing that much to the code yet Although I've also just started to learn how to use github so that I can you know check in and check out code But I think part of at least for for me what I'm Embracing in my work is that you don't need to be an expert to do this stuff with AI that you know It's it's such an important part of what the future of our world is going to be and for us to say that technical Proficiency is a requirement to engage in what that future is is sort of too big of a barrier Considering how important it is and so what I'm embracing is the notion that you can just take something That's open source off the web and do something with it and you don't have to be an expert at it I mean you have to kind of know how to set up a Amazon instance and all that stuff But you don't have to get super deep into the code in order to start developing, you know Your own intuition for what AI can be So I'm just wondering what the traditional analog is to this particular You know subset of artists who are somewhat technically proficient and there there sometimes is this hierarchy Amongst tech artists, you know, well, he doesn't write his own code or he's this or she's that and you know and Which is kind of nasty. I actually really like to stay out of that Yeah Like open source can be like I write a blog post about what I did and I share what I learned, you know Open source is a state of mind So it doesn't have to be like you have to commit new code to get If like, I don't know if we like it's allowed or what we think about it So I was wondering if we bring back Yeah, John, do you want to weigh on open source at all? Well, just in our work, we don't have a choice You know, we're not running the code to run on the blockchain So it has to be open source and so it was one question, you know You know with with the autoglitz because they're all generally there Someone could just take that code and copy it and run it and make as many as they want and they would be You know, they that could happen and I don't think anyone's done that and I think obviously they did do that It wouldn't be the official ones or anything, but it's interesting that that's that's possible that we limited It to 512 but that's an artificial limit You can just take the code and do you want with it hasn't happened yet But when it does and they sell it at Christie's for a big boatload of cash, you're gonna be upset We'll see yeah, maybe I maybe I'd plan to see I think Sellability actually is part of the problem. So open I learned how to use processing because there's this great community around Open source and teaching people how to code I'm a dumb artist, but I learned how to code because there's this great community, right? But now that we're seeing the prices go up through the roof now people are starting to say like maybe we don't want this So open source anymore. So maybe my question for you George would be Sort of this Sellability part, you know, you've got experience as a gallerist and you and I are working on the show coming up with generative art and AI and Especially the younger artists. They don't want to print anything out, but we are a commercial gallery show, right? So it gets complicated, right? Yes, but I think We have to look back a few years in 2015 I believe there was actually an auction at artsy where they were selling pure coats Which is quite interesting and they were sold for like $20,000 or something and and all of the codes were sold all of them and Actually, we had this discussion because in our exhibition we will also show some algorithms, right? Yeah, and For example, Gottfried Jaeger has made an edition of one of his early so-called algorithms About his pinhole structures so we're gonna show that one and I remember I sold a few of those previously and Well, John, we had the discussion. What are we gonna do with? your code and I Think one can look at it as a brush like Painters brush or it can be a artwork. I don't know Yeah, yeah, I think for us for Matt and I because we come from a pure software background But we love to express ourselves, you know and create things that we really think of ourselves We think of that is the case that we make code that makes the visual thing so the code is our brush and But but certainly there are other approaches for example, you know, there's more curatorial approach where it's like I'm gonna use off-the-shelf GAN software, but then I'm really gonna do an amazing job Curating what goes into it and then the output will be fantastic And that's that's a different and that's kind of what's cool about and interesting about the neural network stuff It's it it gets you away, you know, if you if you don't want to do that If you don't want to be like Matt and I a bunch of bunch of dorks You know, you can focus now again on like on on on visual and and you know You can get back to that space while still doing something. That's very technical and cool and this is a dorky build on that which is that I Think what's interesting about a lot of the way that sort of contemporary machine learning works is not through beauty It's through this kind of weird engineering hackiness And so, you know as people build the kind of layers for the neural nets a lot of it It's like they don't quite understand why certain numbers of layers work or why certain configurations work It's much more of like an engineering mentality as opposed to a more traditional programming kind of a beauty mentality So there's a there's a kind of interesting parallel there So David you touched on a little bit earlier about AI you don't have to be an engineer to start But we've got one two three four five White guys roughly the same age on a panel about AI art and one female So is there something essential or you know that has to do with The technology the expense or we did we just not get the proper mix on the panel or you know, what are you seeing? What it what are you seeing David because I know you're sensitive and have you know thoughts around this You know, what do people need to do to start with AI? Do I need a degree from MIT and to be like a white guy who's like over 40 years old or like? Probably statistically that might be the case You know But it's it it can't be the case we have to change what the future of AI is gonna be You know part of the reason that I kind of re Got re-excited about AI was because of my own sort of terror of what AI could be in the future That you know, there's a term that Farhard Manchu who's the tech editor for the New York Times? He refers to as the frightful five So it's the five biggest companies Google Apple Facebook Amazon Whatever Microsoft that basically have all of our data and so these are the companies which are essentially saying and steering where AI is going to go into the future and You know my my own personal sense was like I can't change what Google is going to be What they're going to change what their new new businesses are going to be you know If I want AI to be more democratic to be sort of less colonial I need to find some other way to shift what the future of the technology is going to be and so you know I think that you know It's great to sort of do stem programs where we educate more people and bring more diversity into these companies But I think there's also a huge opportunity through outsiders and through creative expression and aesthetic experiences to get sort of a more cultural sort of idea of what AI could be in the future and so it's through I think through art and through the kinds of things That we're all doing here on the panel that that sort of gives us a new imagination for what the technology can be and hopefully You know enables more people to participate in what that future is That's what drives me crazy about some of the artists who I would consider almost charlatans who are smart enough to understand How AI works, but they buy into what the public wants what the reporters want to hear about like the robot Taking over the world Conspiracy stuff because really art you know an artist in my opinion artists one of the reasons They're the most important throughout history is because they're less afraid and they embrace these new scary weird things for all of us right and when you embrace a false narrative and especially around something that people are afraid of you're really kind of coming up short I think of Filling that role that artists can fill this AI art could just be an attack vector for the AI. This is where it all starts Kevin and I were talking about this earlier that you know so much around AI is about Bigness and speed and kind of who's doing something bigger and better and those are really the the narratives that we're getting from Kind of the big tech companies and we can see if you know through art we can say AI can be slower AI can be smaller it can be so it doesn't have to be about consumption or efficiency perfect even absolutely Yeah, these massive massive systems are well sure and you were saying too that People immediately are like oh look at all of these really bizarre scary weird images and you're like well They're not necessarily bizarre scary and weird we can like look at them like in a different way, right? Like the media is always Declaring them as horrifying and stuff like that Yeah, yeah, yeah like you said before like the robot is gonna take over the world and take our jobs. Yeah Exactly so Not gonna take over our jobs. I don't think at least any time soon It doesn't sound like it usually the the people I talked to who know the most about AI are they're like It's just like augmenting like what we're doing and not freaked out. So oh because it's still slow and stupid It's amazingly how slow it takes to train one of these systems I mean you think of AI is this all powerful super thing and it's it's We just bummed out everyone here who was hoping that way we're gonna talk about like robots taking over the universe You're like it we're like it's slow and it's stupid But it does do this thing, you know and talk interviewing a bunch of these AI artists There is this point where artists using AI. What's that artists using AI? Oh, sorry in interviewing these artists using AI There is this black box point that they never get to talk about because it re Sort of reassures this false narrative or whatever But there are things that are going on that you guys don't even understand right is that is that fair? Can we talk about that a little bit in the game? The black box the idea that there are things going on and again that that aren't knowable Yeah, I don't know if we can call it a black box. Like I don't know if I agree with that it's like It's it's like kind of like an ecosystem of code, right? Like we we code in layers of layers and we kind of don't really understand what already What's changing what or why firing what but it's not really a black box I mean you find yourself laughing sometimes at the result like sometimes I'll see something like wow like I would never like You know we talk about these things thinking, you know or acquiring knowledge It's a completely different process obviously from the way we we acquire knowledge. I think and sometimes you see the result And you're just like oh, okay, that's interesting. I had a simple thing my such as quickly my nine-year-old son of the science fair project He says Papa. I want to do something in AI so we decided okay. We'll train on a bunch of photos of apples We took a bunch of photos of apples and I don't remember what we were going to do It had something to do with image classification at first But then I had this idea that whether a green and red apples now is it going to be able to know which oh the idea was We're going to convert all the color photos to black and white first bring them back into the color realm You know there are apis to do that, but will it know which are green and which are red? I thought to myself okay logically When it trains on enough data it probably takes some sort of tonality and if they're on this side of the tonal tonal scale They know it'll be green and the other ones will be red, but then of course all the photos had a shadow side So how is it going to interpret shadows and anybody who trains with with these gans knows that shadows complicate things and That's when it was really strange. It didn't know it doesn't know that's a shadow It just knows that there's this tonal shift I guess and so we ended up with these kind of marbleized green and red apples And it just brought a smile to my face because just when you think you understand how these things think you realize you don't Fair I've just did a quick time check. I don't want to make sure that we have time for questions So I think we have about 10 or 15 minutes. Is that's fair? Yeah, so yeah, let's let's kick it to the crowd. Who's got questions? Well, you want the mic? Yeah Hey guys All right question. Yeah, so I have a question here So Looking at you guys, it seems like you all have a very strong background in engineering and computers at the same time your artist, but And you mentioned this earlier. I would argue that you are not representative of artists at large in terms of your background so my question is about access to accessibility to To AI using AI as a tool It seems to me that the cost to use it at the moment is pretty high like the barrier of entry is pretty high Do you see this changing at some point and just in general what are your thoughts about this, right? Like do you see having tools like having a Photoshop for AI for example, and if you had that would AI generated with this tool be actually Considered the same type of art that you guys generated the moment with your models Yeah, I'm coding but I'm not machine learning quarter. So for that's for example, and I work with a encoders and other people when I want to create projects and now the project that I described is a It's very big. So I work with a team But I do have like other other platform that I see evolving and we work on tools and I see other people work on tools So I I see this field growing as like a creative field for like Creative tools like there are a few amazing things that already developing. Yeah Yeah, that's specifically to that runway ML, right? So runway ML is essentially AI Photoshop, right? So it's designed ground up to take all the latest and greatest research academic research on AI Algorithms and make it available to other people based out of New York, Cristobal, Venezuela. Yeah. Yeah, so yeah It's coming and I think the second part of your question When everybody on the planet can use these algorithms, is it just going to get boring like a Photoshop filter, right? I think what's what we're starting to see is that It's helpful for the really talented AI artists because their stuff is still different Even when we all get the algorithms, right? Their work is just like we could all have a paintbrush and paint, but we're not all gonna be you know, Michelangelo or whatever, right? So it's beneficial to the artists who are actually good at this stuff that it's getting into other people's hands because you can Differentiate what's you know why they're different and what they're doing. I don't know if you guys would agree Absolutely, I think you know the the tools are gonna get better but to go back one step You know, I come from a technical background, but I haven't coded until just recently for 20 years And I was interested in AI and so what I did is I found you know websites and blogs Sort of to the point that you were talking about earlier that said here's the steps that you need to go through To get this code running and they would have links to the github place It would have you know instructions on how to set up an account on Amazon to do the processing Which can actually be quite inexpensive when you're first starting out like you know a couple dollars a day And simply by following those instructions and not doing any code at all But having a sort of a sense of like you know I can just follow these these instructions and I'm not going to be scared of the technology I was able to feed it images First from sort of the common repositories of whatever ducks or things They are there are that are out there and then begin to put my own images in and just substitute it that at that out And it was it was quite easy Once you sort of figured out what the steps are to do and so I would say you know Although we all have some sort of technical proficiency It's not it's not a requirement to do this stuff. Also if I can add I Started like three years ago with like just reading a lot of research Because academia has tons of data sets and it's beautiful how much is accessible So I like one of the things that are showing here is a tool that I started to develop out of just like research So I read a lot. So that's also Thank you another question. We got a question here. Yeah I was wondering when you when you feed visual images Into AI and have have a generate artwork. Are there any intellectual property concerns with the original images even though they aren't Representative in the final artwork Yeah, so I've talked to a lot maybe a lot more lawyers than you guys maybe you talked to lawyers, too I don't know I talked to the the woman who runs a cyber law clinic at Harvard and she's got I'll pitch art No, I'm here. We did a long interview and she goes in depth on a lot of this She basically says that if it's a lot of the same artist, you know, if you're using a hundred works by the same artist She thinks that there could be problems because all the training material came came from the same person But it's a little bit which is interesting because you would think law would be kind of specific I've talked to probably five or six different sets of lawyers and every single one says something different Which is why I was kind of talking about maybe black box is the wrong term But there's a thing that's happening in the middle that people don't fully understand yet So it's hard for them to govern it So, yeah, I don't know if that really gives you an answer other than asking you to go read my blog Seems like the early days of hip-hop when people would sample with no problem and then all of a sudden there was a Point where it became a legal issue. Yeah, I think you're right I will say and I don't know if you guys agree with this But the the artists that I talked to laugh at it, right? They basically say all these things that the lawyers are writing show that they have no idea what's going on I think this one woman Jessica field from Harvard might be one of the exceptions But there are a lot of lawyers racing in to want to weigh in on this stuff because again anytime you say AI though The readership goes up most of them It doesn't seem like they really know what they're doing and we know from experience that that the entities that have the most Leverage in the most money and the most to lose will probably help shape legislation So won't just one more question is the um when you create a work of art Is do you usually? Like let on what the original data was that led to that or is that kept secret or like citing your sources kind of Do you guys say you're well? I'm yeah I mean I from for me personally I'm using images that I take myself and they're all images from my farm in upstate New York partly to also contrast the nature of nature versus big tech and I have shown a Sample of those images so people can kind of get a sense of how similar and how different they are But it's a it's a good Question because I think sometimes some of the AI the art made with AI Has a kind of gimmick equality to it like you know they train it on all of the you know churches in England or something and so I think it depends a little bit on how unique the art is I mean I think your point around how you did your nail series is interesting I train on my own my own data set, but I mean I don't know I mean there there've been derivative works There's a lot of you know artworks that are Clearly derivative satirical was satire is a whole thing unto itself, but I mean yeah, I don't know I mean it's just a debate again. I think it really comes down to How big are the lawyers and how big are the the moneyed entities behind who are making their case? Yeah, I think our panel is unique in that everyone as far as I know kind of rolls their own here Whereas you know like a Mario clenchman or a Robbie Barrett those folks are pulling for massive data sets on online They might weigh in on something like that any other questions Well, we got one in the front row This is a fascinating panel. Thank you so much everybody We've talked a lot about kind of on the generative side of things and conceptually kind of how you're you know coming up with The artwork and the concept I'm interested in kind of the the other side of that as well in the marketplace and kind of what the reception has been from traditional collectors or institutions curators And kind of what you're putting out there and how that's being received commercially and and in kind of the fine art marketplace Yeah, I give this one to George maybe to start Well, I I can talk about the generative photography maybe because that's the market I was working in when I used to have a gallery and Certainly at the beginning it was a something very new and For most curators and collectors They had to read a few books to understand what it's about but I was quite surprised that I was selling to collectors who had Picasso's on their wall and Kenneth Mowland and I mean like It was it was really strange that they got fire for it and I think I Think it's a question of education. So it's it's Yeah, I mean, it's it's not a type of art which is Easy recognizable like Damon Hurst and You don't need to read many books to understand it I mean Most of the artworks have a very interesting concept behind it and if you take time to learn about it you catch fire and We managed to get it quite a few really important collectors to buy these artworks But I think it needs more. I mean I Believe the we need more dealers not more curators. I think what we noticed last year was that because it was kind of like 50 years anniversary of cybernetics and deputy and all these computer art movements There were a lot of exhibitions I was surprised how many they were like I think I counted at least 10 exhibitions That were showing these early computer artworks from the 60s There was an exhibition in the MoMA in the Whitney in the Santa Pompidou the Grand Palais there was an exhibition on robots and V&A etc. So And I realized there is some interest for this and Actually, we did a Yeah, blockchain art exhibition last year in Zurich at Kate Voss Gallery and We showed like the Cryptopunks for example and some works by Kevin and It it was there was a lot of interest a lot of interest a lot of new collectors so I think Especially the new generation of collectors. They are Looking into this stuff and they understand it more than maybe a traditional collector, but I had I don't know I had Collectors visiting they were like over 80 years old and they were interested in the Cryptopunks Yeah Yeah, I think I would just add to that my pet theory is that in the 80s if you use computers You were a nerd, right? I grew up in a family of engineers. We're all nerds now We all use computers so we're at this threshold in history where computers isn't a weird thing that just a few of us Do it's like maybe one of the things that we all do that actually brings us together that we have in common Right, so I think we're at a tipping point where the market's gonna grow I'm pretty bullish on you know investing in digital art and generative art in particular I don't know if anyone else has anything that saying your nerd is such a hipster thing well Saying I'm a hipster is such a nerdy thing though, right? Do we want to throw in one more? Yeah, we got one more question Just one. Oh, sorry. Yeah, I'm curious about how this art form is developing globally in particularly in Asia Do you have can you speak to that? AI art in Asia anyone have that's that's a blank spot for me. It seems like it'd be a sweet spot for them That's why I'm asking Yeah, I know on Twitter a few folks I follow so Google Advances a lot of this technology and there are a couple of folks I follow closely on Twitter that I know are very heavily involved on a research side But I don't know that they would call them cells AI artists you guys have any off the top your head Missed exactly I've been I just got back from Hong Kong and there was a lot of interest in my work that deals with AI Yeah, so I was just there for art puzzle, but artists and other I mean, you know, obviously you're up There's like Mario Klinchman and folks like that, but In Tokyo there is David who I don't know if you follow him on Twitter But he's doing a lot of reinforcement learning art with Google brain. He's a hard hard Maru on Twitter Brain brain research and reinforcement learning less gun I think Less can less can and more research So I think the real answer to your question is that probably a lot of folks just we don't know as many of them But we know some so yeah, I think that's it, right? Yep. All right. Thank you everybody appreciate that. Thank you