Rebroadcast: AI Livestream - Kris Kashtanova - AI Educator, Influencer and Consultant

Kris Kashtanova: Welcome back everybody. This is the Artificial Intelligence Livestream. Today I am joined by a really awesome

Jon Radoff: guest, second time guest actually, although not originally in the live format. My name is John

Kris Kashtanova: Radoff from the CEO company called Demable. We make infrastructure for games to be online.

Jon Radoff: And I'm accompanied by Oscar, our producer who's been helping to keep this thing on track for episode after episode. Chris Cashinova was on my YouTube channel, I think about three years ago. And we talked at the time because Chris is famous for having filed the very first copyright of an AI assisted work. And there was all kinds of news about generative AI and whether you could copyright it or not, there were no guidelines at the time. And that was the context that I got to know Chris in. Now since then, the whole market has evolved a ton. The copyright office has some new guidelines now that as I understand it are actually based on some of these early interactions that the copyright office had with Chris will be able to talk about that. But well, let me take a moment here just to introduce Chris to you. Chris, welcome to the program. Why don't you give an introduction of yourself and where you've come from, where you're going. And then we'll dive right in starting from that work that we were just referring to and how things have changed over the last two or three years.

Kris Kashtanova: Hi, everyone. So happy to be here last night when I talked to John. I just remember how inspiring was that conversation, even though it wasn't live and there were no other people on stage. I just remember the way I felt. And now I see all of you. I see Andrew. Hi, Andrew. It's so cool to see you. Okay, live is so much better. Hi, Peter. I know all of you. You probably all know me. So I didn't really need the introduction. I just make stuff with AI. And I love, I just love doing creative stuff with AI and also learning from the community and doing stuff together and interacting together. And I like creativity of all sorts. Before I was doing pretty much everything I could get my hands on like claymations, photography, sculptures made out of garbage, all kinds of stuff.

Kris Kashtanova: And then when I discovered Generator AI, it was so new to me and it was also so not so good.

Kris Kashtanova: As it is now, you couldn't really make such amazing things as we can do now. But I felt something. I thought, oh, there's something in it. So I started creating and then shortly after a few of you actually told me to share. And at first I was really hesitant. But then I started sharing and the community to grow. And I met all of you and I met John. And yes, and here I am. I just want to share all the ways I created with AI in this three years since John and I talked last time. And hopefully maybe you will either try AI or try something interesting and also maybe mix up my ideas. And yeah, I would love to keep this conversation going. So is this live? You can

Jon Radoff: all jump in and wait, we can just talk. Yes, I love it. You're doing my job inviting the audience to participate. So the exact thing I was going to point out. So there's a reason we do this is a live program. We could do it as a YouTube video. That's exactly what Chris and I did. I think it was three years ago. And it would be pre-recorded. And then we'd go through editing and there'd be all this post production and we post it and people could watch it. We've switched a completely live. And the reason is you, the audience get to participate that way. It's an opportunity for you to ask all the questions that you're really curious about. Chris is just an expert on creative technologies in using generative AI in particular towards that. As Chris was talking about, she was an artist long before generative AI became an influence on her career. So we do want you to come in and ask questions because you're not going to have as many opportunities as this to get FaceTime really with Chris and ask about her journey as an artist as someone who uses generative AI. And also just what she's discovered about these technologies and how they've changed over the last few years. We've already got a couple hundred live viewers. So that's awesome. Five minutes into the program. I suspect this is going to be super popular with our audience. So again, you're here, you're live, ask questions. And by the way, if you want, you could even come on camera too. So if you want to join in the discussion, Oscar will shoot you a stream yard link. We can make that happen. That's

Kris Kashtanova: happened maybe a quarter of the time, maybe. I want to see everyone on camera. That was my

Kris Kashtanova: expectation that this will look like a Microsoft team or Zoom call and we'll just talk. But it's okay if you feel shy, please feel free to just put your questions into the chat. And meanwhile, I would love to share, like I prepared. I took some of the videos I made in this few years. And I just want to share with you some kind of journey that I was on since the beginning. And John

Kris Kashtanova: probably remembers first part, first one slide. I know that he will be excited to see that because

Kris Kashtanova: he's on that slide as well. So this was if I remember correctly August, 2022. I made my first comic book, Zara of the Dawn. And it's probably one of the most famous things that I've done in my life, which I didn't expect. I was just experimenting and playing and sharing my process very openly. I shared every single day what I was making on my Facebook page actually. So I had just friends and family commenting on that. I did have ex account, Baghdad Twitter. I was not using it at all. I maybe had like 100 followers. And I never thought about it because you also couldn't write long text, Baghdad. And you couldn't post long videos.

Kris Kashtanova: I've never got to meet actually the share there. And then I remember I posted my comic book on Reddit. And that's when it went viral. And that's probably how a lot of people discovered me.

Kris Kashtanova: Then finally I joined X. And I started sharing all my stuff there. There's a few pages that are in the comic book. It was translated into 33 languages. People just wanted to participate. The story was extremely simple. It's actually only one paragraph from a novel that I wrote a few years before. And I just tried to visualize like a small part of it. And because I was so limited, I couldn't really visualize the entire novel. It would take forever. I remember like one page would take me a whole day to get the right images to like, there was no generator

Kris Kashtanova: fill back then. Like I had to use like clone and other tools. It took forever to do it.

Kris Kashtanova: But I was so happy with it because I felt like, you know, you can tell stories. You can tell your stories. Your old stories, something you've done before. But now you could use some visual things. Before that, I actually also tried to do that. I use self portraits that I'll show you

Kris Kashtanova: later that I used to kind of visualize my novel. But this is it. I remember how this page with

Kris Kashtanova: postcard, it took so long. And still, I just had to kind of frame it because she just didn't want to hold it. You would see all the hands wrapped around. And if you are into it, I remember how hard it was to like generate hands, like someone holding something or someone doing something. So this was the first creative walk that I remember for, I think that comic book has 18 pages, which means it took me about 17 days because I was making one page a day. And I remember it just took me and my imagination into such state of flow that I just wanted other people to experience. I think before that, the only time I experienced similar feeling of the flow was when

Kris Kashtanova: I got my first computer. I was 11 and my grandma got me a computer. She saved my

Kris Kashtanova: for it for a very long time. And I remember she got me that computer and I learned how to use different software. And I started writing my first code. And that was the same feeling. The difference was that I wasn't 11 years old anymore. I was an adult and is just childhood curiosity.

Jon Radoff: Of course, let's talk about that flow aspect for a moment because a lot of people are talking about that a little more recently around vibe coding. If you're picking up the sirens conveniently

Kris Kashtanova: coming down the street as I'm talking. So a lot of people are talking about vibe coding is

Jon Radoff: putting them in that flow state as well. And sometimes people are very focused on the output of the generative AI process, whether it's sitting down at chat, ept, and having a conversation, and getting to a result or creating a piece of art or creating a piece of code as is the case of vibe coding. The output is one thing, but sometimes this experiential aspect that you're referring to, I think is maybe overlooked too. There's something to the experience of working interactively with generative AI systems that are capable of actually responding to you quickly enough in real time that you're going through this process of shaping things and seeing your idea

Kris Kashtanova: come to life. The concept of flow that you're referring to, as I understand it, comes from a

Jon Radoff: psychologist and I'm going to probably butcher his name, but I think that it's something roughly cheek-sent, Nihae. And he came up with this concept of flow, which is also used in game development, by the way. That's mostly why I'm familiar with it. As being the space between things being excessively hard, where it's so hard, you're kind of stopped in your tracks, but not kind of so simple that it becomes boring and you get uninterested. So what was the state between those two things where you can get super engaged? You know, it's great that you

Kris Kashtanova: mentioned the game aspect of it. So for me, the way I understand flow is that when you are in pain, in physical pain, maybe you had your tooth pulled, I actually had my dental surgery at the same time when I was doing Zara of the Dome. So I just wanted something to get away from that

Kris Kashtanova: nagging feeling of like sobbing pain. And I was looking for something and they were not that many

Kris Kashtanova: things that I could do at that time. By the way, hi, Ari, if it's so good to see you, hi, Aaron. Oh, yeah, all my friends I hear now. I feel so good. Yes, all of you on camera, guys. I hope you

Jon Radoff: join us. So I'm not going to ask, we'll get you the link. Just ask wherever you are linked in

Kris Kashtanova: X anywhere. We're monitoring all my house. You can join me here anyway. So yeah, and the thing is, it's like when you are in pain and many people shared with me actually that when they have like a minor surgery or they're recovering from illness, they sometimes play video games actually. I find that modern games are actually quite complicated to play. You need to like figure out the gameplay and then you can enjoy that feeling of flow. That's why I like very old games because it's so easy. They're so simple to play and then you're in that sense of flow. So it was similar feeling because I already had the story. I didn't have to make it up. There was no chat GPT back then. So no one could write a story for me. All I had was like my story, my old images and a lot of motivation and a lot of free time because I was at home recovering like you know, couldn't do much. And then this magic happened and at first it was so exclusive. I remember there was this guest please Aaron jumping, jumping. Yeah, I remember like I was making it and it just felt like I could do this forever. I could never stop and it just I wanted to create more and I wanted to share more. It's also this feeling that you want to share everything. You don't feel possessive or like I want to hide all my skills and be the best at it. No, no gatekeeping, nothing. And we had such a wonderful community on Facebook at the time. It was a small group. I joined that group and everyone was sharing images and like writing nice things and sharing prompts. So it started really really well. And I think it's still going. It's been two years and the community is amazing.

Jon Radoff: So we've got Aaron Wolf saying you'll be on camera from LinkedIn and sorry if I get your name a little bit wrong. Arief Hussein on X. Loved to have you both on. Jump in. Yeah, um,

Kris Kashtanova: message. Hey stream. Do me a favor and guys message me because at least Aaron I can't directly

Kris Kashtanova: message you to send you the link but um you'll find me there. Oscar it. I'm going to message directly on LinkedIn. Perfect. Yeah. So Chris, while we're waiting for a couple of folks to join us

Jon Radoff: here and I think this will be awesome. You're the most popular guest so far in terms of attracting people who want to jump into the live conversation. So you've got a nice little fan community going

Kris Kashtanova: around you. Let's talk a little bit first about how has the technology changed over the last three

Jon Radoff: years since the time you did Zariah of the dawn. And then also I do want to come back to the copyright piece too because there was sort of a lot of fame around that and as I understand it some of the rules have actually evolved out of your experience working with the copyright. So we'll kind of put a pin in the copyright for the moment. Let's talk about the technology and the tools for us.

Kris Kashtanova: What's new in three years? So first of all I don't know like I know that a lot of people started quite recently and then there are also people who started at the same time as I did. And of course people who started at the same time I did they know the reference, they six fingers and all of that stuff and that you had to be a prompt engineer to write anything and make it walk

Kris Kashtanova: and you would spend hours to get anything decent really. Also there was a very big limitation in tools

Kris Kashtanova: and access to those tools. Like right now we have so many tools. The choice is absolutely insane and they keep coming. Every single day we have two or three models released, we get that. That was the time when there was no charge. The charge. The moment has not happened and imagine like every day something would happen and people would find new ways to use it because I think limitation always kind of breed creativity because you have to you want to make something but you have so little. Now I feel like you are only limited by your imagination and when people say that the tools aren't there yet I think it's they are getting there and if someone is skillful they can get pretty incredible results with it. I am friends with few people who are like filmmakers like professional filmmakers and of course they have a team of 50 people. It's not like they are making all this stuff by themselves. I make my all stuff by myself.

Jon Radoff: What would you like to show us Chris? Let's take a look at some things. Do you want me to show my stuff

Kris Kashtanova: or some stuff of other people? Anything you love? Anything that you think everybody would love. The Wheel and Drive. You got the share. I am going to share my screen again. I put together some stuff. I feel like probably one of the works I am most proud of. The video aspect of it, I was waiting this entire time. I actually made the video this year. It's quite recent. But the photographs I was taking for five years and all of those photographs are of me. I like dressing up. I like building prongs. I like hiding my camera on the tripod. It's just time I spend by myself. Sometimes my friends join me. Sometimes they help me take those shots. But those are just photos. Also I haven't really used a lot of Photoshop there mostly as they are. And then I animated them with Adobe 5 Live Video model. That also I felt special because it was something I worked on with the team. It was just so proud that I waited for so long that I could use my photos with a model that I participated in creating. I think about 30 seconds long it has

Kris Kashtanova: music in it. So just relax and enjoy. So yeah, this was fun to make.

Jon Radoff: And the tool you were using I guess was Adobe Firefly because that's what showed up in the end there.

Kris Kashtanova: Yes, I did. The thing is I had those bunch of photos and I have a lot of photographs. I was doing something called 365 Project which is like when you take a picture a day. So now I have like years of those photos so now I can animate. But we'll talk about it later because I see a reef here. Hello! Good to see you. How are you doing? Welcome to the program.

Jon Radoff: So feel free to speak up at any moment. And I think I saw we had Aaron but he dropped off so I'm

Kris Kashtanova: just going to go see if I can find him. Hopefully we'll have Aaron back any moment here.

Kris Kashtanova: And can anyone hear me? Yes sir.

Jon Radoff: Thank you. You're welcome to the program. What questions or comments do you have for Chris?

Kris Kashtanova: I hope you're doing great. My question is now we are stepping into the AI and the open I am seeing nowadays that it is not a plant, it is a kind of a theft. So how do we ask the AI generation specialists and this type of stuff to mention that this is not the kind of a theft but it's a process of continuation from the ideas that has been in the past to modify something like look better than the origin. Right I like what you said that it is ideas from the past that we can kind of bring them to life

Kris Kashtanova: and I correct did I hear you right? Yeah? Yes you hear me right. Do you have a question?

Kris Kashtanova: Just a thought. My question is how do we handle this type of stuff? Like I am sharing something far from its version on Instagram or Twitter. Some unknown person just jump in and start

Jon Radoff: questioning the authenticity. Right right. Are you for the question kind of people are questioning the authenticity of things made with generative AI because it sort of taps into all of this existing

Kris Kashtanova: structure of content? I have one thing that when they ask about authenticity they are also mentioning that model, the large language model that this app is built on are surfacing data from online libraries or the online publications and they without legitimizing those things that they have source those information or information taken from and when it is being built people are jumping in and then start typing some media well artifacts and then from the model and then

Jon Radoff: stop trying to get the question. So a common objection to generative AI, there is sort of the legal objection around copyright which we do want to talk a little bit about copyright and sort of Chris's personal journey through that but also you are raising not so much the legal question but just one of authenticity as you put it which is and it is sort of the question of when you create how much of it has to be sort of uniquely inspired versus how much taps into the cultural subconscious or in the case of generative AI just the pattern matching that occurs by searching overall of this content that exists on the internet already. Chris I'm sure you've

Kris Kashtanova: thought about this question. I think we wrote this question all the time. When I started using AI, I remember that when I was making that comic book Zara of the Dawn, there is one part that I wanted to honor my grandma and I wanted the character to look like my grandma and I remember I would use photographs of my grandma that I took of her before she passed away as a reference image and the generations didn't look like anything like her three years ago. Right now I can put your photos in video model and they actually make her move, they can make you blink your eye and it

Kris Kashtanova: looks very realistic, it looks like that could be her. So if I was making that comic book now,

Kris Kashtanova: I could make it look like her. There is no one else who looks like her. I look like her but I'm way much younger. I'll have to wait another 50 years to take photos of me and be part of that comic book but now I can actually use her as a character like like I was meant to at the beginning.

Kris Kashtanova: So I think there is an interesting question because some people are artists like me who use AI

Kris Kashtanova: they sometimes feel they need something extra, they need their own drawing or they need their own photograph or they need their own design and then they can use AI tools to maybe make move or make something else. While there are some artists I know and they're very successful, amazing people who only use text to image and then they are too sold at very famous places like Soda Beast and it seems like the person who gets it, they don't care as much how it was done as long as they connect with the artist. I think the artist is still very important because I'm pretty sure if I just put that picture on X without any reference everyone would look and be just like oh that's just some image but when it is the artist and when there is a story behind

Kris Kashtanova: it and the artist done it with intention then it can be very different and I think the meaning of

Kris Kashtanova: what it is to create art and what it is how to create art it can be so different. There are things like they shake the world and then you ask oh it's just a banana taped on the wall why is this art? Why is this better than text to image? But then there is context, art always has some sort of context and I think right now with AI because it's so new and the technology is new we are defining actually what is our thing tech. I know so many people on X when I see the generations I know it's them. I don't see the name I can just see an image and I know that that's they are I know that's no one else because they establish their own style so I feel like authenticity in a way that is it your voice is it your style is it recognizable is one thing another thing is being able to say how was it made? Where what model did you use your name on it? There are actually tools like content credential tools that you can search your image. It actually happened to me once I posted an image and I was so good and I was so happy because it was a long time ago and good images were very rare so I was so proud and then someone just took it, reposted without credit went viral with it and the only way for me to prove that was mine was to go to content authenticity site and show that I made it fortunately the person who did that didn't do it on purpose at all. He just honestly he saw it in the thread he didn't see my name we became great friends actually probably watching this right now so sometimes you know mistakes happen social media is so fast just having a place where you can kind of see I know that LinkedIn now uses the content credentials

Kris Kashtanova: as well you can click and you can see who made it how it was made so I think we are getting into

Kris Kashtanova: some sort of direction where I still think it shall be optional thing because it's like do you want to do you want to show I would definitely always want to show how my art was made but yeah that's that's probably hopefully I asked your question I know that I think it's actually amazing his generations get so many likes on Adobe 5 like gallery it's crazy like there is a gallery where people can submit their walk and then it's like a social network people can remix those and then they press like and recently I only discovered that people actually press likes on those I didn't see that and I saw a refs library and it's like it's like cool it has like hundreds of likes and they mix and stuff so thank you so much for for putting so much of you into that

Jon Radoff: I mean in article kind of that is related to this question a few years ago and I'll just say first that I I think this word we use creativity actually is a very poorly defined term anyway I think of creativity as actually the search for some kind of solution and people that we call creative are very good at navigating highly optimized paths into the universe of solutions you could kind of think of the universe as like this hyper parameter space where anything is possible but you can't just sort of brute force your way to a solution like an old chess game from 40 or 50 years ago you have to actually have good pathways towards solutions so that you can win and find interesting things so in this article you know I don't talk about art as much in this even though I wrote another article called the work of art me age of generative AI which which touches on that subject a little bit more but the interesting thing here is this article it's actually this research done by someone who has figured out how to discover physical laws not by brute force in your way through all of the possible physical laws but by finding optimized pathways and you can look at data sets and get to actually pretty interesting physical law definitions such as relativity for example can be unearthed through this process and the point of this is that when we think about creativity we often want to glamorize it or think of it as some kind of form of enlightenment as it only humans can do it but I'm actually very interested in this idea that humans and a machine working together can unearth really interesting patterns and solutions and approaches to art and the ability to apply known forms to new problems which generative AI happens to be particularly good at so I don't think it's as simple as just saying oh well there's all this content online that it mines and therefore you're just rehashing it particularly with these models as they've gotten bigger and bigger and bigger we have found that there are some really interesting emergent properties to this content especially by having the human in the loop being part of that prompting process because that actually makes it far more emergent because you're bringing in your own social experience or in life experience to the set of content that you end up creating from it so I don't know I don't I don't find it in authentic other than I think there are some interesting cases that we could talk around like the copyright aspect Chris let's maybe branch back

Kris Kashtanova: something yeah go ahead okay John I have no problem with generating

Kris Kashtanova: AI images but the problem is now there's I'm seeing some models that has the ability to do unsensor art like something the same model you can use for like far fly it will block you like away when you when the model senses that this is a and uh prompting for something non-conjunctive type of task like generating images perfect detects it and block you right away but now I'm seeing that some sort of AI models are after like a real unscursive and so so from something like that it will automatically generate it but uh they also mention that this is not safe for why but my question is why you give the luxury of unsensorship and then you are level those images like not safe for what why is that I think it's also it also has to do what are the use cases of the models because the use cases like I teach kids a lot trust me I need all the guard rails I can probably get because it's kids I have to make sure that whatever the model generates is safe and also I have to say my youngest student was two years old I his father and mother asked me to do it and I didn't believe I was like but he's two he probably is like how are we going to do it he's too small but the thing is that he was really smart and we made the comic book together we made the comic book with his parents and with him so for me to be able to generate images that not only I say from the point of view of us adults but also his two so they don't they have to be not very scary images and you know kids get scared by that so for me something like this I can understand it from a perspective as soon as you start teaching kids teaching younger people you can see why guide guard rails is a good idea because sometimes it's like AI still still hallucinates you don't want it to generate something either to scare you or inappropriate this is one use case then also businesses if if it's a meeting and someone makes an image you want it to be professional you want it to be like you know not a trouble so some companies are very careful because they want to make sure that the customers are happy that kids are happy especially if it's used for education and then of course there will always be open source models and people make stuff and something without guard rails and I think those things also great they exist and also great that we know that they exist because of course artists maybe they may want to generate a sculpture if you go to metropolitan museum of course you can generate all kinds of sculptures all kinds of art it's not as restrictive but then you know that that's where you are getting you probably need to be very careful not to have kids around your screen and like you know use it with like understanding that that model can generate whatever and stuff like this yeah I'm super interested how it's going to evolve because right now in terms of the law it's still evolving the guidelines are constantly evolving and there is something that we already have in the US and in other countries it is different I know that you you mentioned that you come from Bangladesh so probably the law situation is quite different and it's good to look in Sukup Red Law in your own country to see what the guidelines are there as well yeah

Jon Radoff: I will say that the genie is also out of the bottle because the thing that's interesting about art that's actually different than a lot of language models or at least the the very advanced frontier models

Kris Kashtanova: is that you can generate very pretty pretty great art off of a model that's capable of just running

Jon Radoff: locally on your machine so there's not really going to be any top down organization or two or three vendors that decide what is appropriate or not to do with a generative AI model with art language models that's becoming increasingly true as well like the language models you can run on a local device now are really really compelling I mean as good as what we consider chat GPT to have been when it first launched yeah run a frontier model on your machine but they're really compelling so we're in a world where open source models and open weight models they're going to be developed they're going to be on your device obviously that in my mind puts more of the responsibility on the user of those technology I don't I don't think we should be in a world where either we blame a big tech company for not having the right protections in place every time we miss use their technology or want to hire them as the policeman to decide what we get to create like we should have the agency as creators as coders whatever you use these technologies to be able to do whatever you want with it but then you have to decide if that's something you want to ship that's something you

Kris Kashtanova: want to be proud of yeah absolutely well to be quite honest the difference between now and the past is that in the past you could draw whatever image you want with pencil in your notebook no one could censor you but then it could be something wonderful and great or it could be something inappropriate well now the difference is that it's extremely fast and easy to do it and in

Kris Kashtanova: on scale and also like you don't have to like you you know what I mean you can you can just

Kris Kashtanova: find find a way to do it but I think in the past people also could find a way to do it it just now it becomes the way that we need to educate people especially parents what what can happen and how to you know to at least to know what is going on and as for models I want to mention something I was recently teaching a workshop at Fashion Institute of Technology here in New York

Kris Kashtanova: and they invited me because of my work at the Dobe and it was so fascinating all of them were

Kris Kashtanova: adult students and then they told me that actually they make their own models from scratch they put their own art and they make guns of course the art it makes it's not realistic it's not as realistic as it is now but because they are artists they actually make all kinds of abstract art with it and I was so fascinated because it's not just a model just fine tuned model it's a model some of them actually figured out how to make definitely not diffusion they don't have so much like GPU and stuff but still it could draw like very primitive stuff and they were so proud of it so I think the thing is that people will make models people will find you in models they will find ways but there is also a question of application what do you need it for like a gun model yeah I can I can train it on my art and I have I can have some very interesting abstract interesting art but then if I want to visualize product that probably or a comic book that probably is not going

Kris Kashtanova: to work because unless it's a very creative specific comic book so I am happy that there are so

Kris Kashtanova: many models and where I work now I actually advocated for having as many models as it's possible because we need to make sure that creatives have tools you have a choice you don't have to use AI at all you can draw everything you can take photos but you also have a vast toolkit of so many tools and then your creativity it just exponentially grows how you can mix it up I think what creatives do is just we mix up things that we have in front of us and now we just have more things more possibilities to mix more possibilities to collaborate with machine like you said and it's just amazing it's an amazing time for creativity I think Chris this touches this does kind of touch

Jon Radoff: on the copyright issue too because this this goes along with people's like legal objection yeah

Kris Kashtanova: the same question for for me also that this also raises a question of popularity should be because

Jon Radoff: yeah we hear you so Chris like can you take us through the like the a writ like you were the pioneer for copyright with Zarya of the dawn can you just talk a little to what happened then I think a lot of people might be familiar with it but what's happened since then and like how has this changed

Kris Kashtanova: you over over time okay so I'll open my I'll open my slides just in case someone doesn't know

Kris Kashtanova: I think it was 2022 it was 2022 August when I finished Zarya of the dawn and then the thing is I was sharing it I was sharing my comic book online my process everything and I remember people were constantly in discussion can this be copyrighted for me personally I never planned to sell it that wasn't the goal of this but I was just so curious what other copyright situation in the US um so I submitted just a copyright there is a website the copyright office has a website you can submit any of your works it doesn't have to be a it can be any work it can be a drawing it can be writing anything and then um and then I received copyright within a couple of days and honestly my first reaction was oh that's great they already know what it is I wasn't hiding the fact it's actually on the front page I did make this with me journey at the time and I thought everyone knew what me journey is so for me honestly there was absolutely no doubt that they knew what it is and it was copyrighted again there was no charge of it to Baghdad the story was written by me I did use a bit of retouching with Photoshop again Photoshop didn't have generated fill Baghdad so it was all by hand the retouching was by hand um and then uh those are few more pages like especially this second page there was a lot of work because Baghdad it just wasn't very good at generating like close-ups and features and I really wanted to make it as good as I possibly could um so I got that copyright and at first I had absolutely no doubt and then I think a lot of press wrote about it that the AI assisted comic book got copyright and a month later I got a message from not the message an email from the copyright office saying that they would like to ask me to write a very detailed letter how this work was done and they will decide if they if I keep the

Kris Kashtanova: copyright or if they somehow change it so uh then it took a long time I was already employed I was

Kris Kashtanova: employed with a startup um I was working in industry already I was super busy so the whole community was actually helping me out with like collecting all the images and making the document and uh submitting it to the copyright office so after we submitted the document by the way all of this you can find online everything we submitted um they said that they will give me a different copyright they will give me limited copyright my story is copyrighted and the arrangement of images is copyrighted while every image by itself is not what does that mean it means that for example if someone prints this exact page they can't they can't sell it because that is copyrighted but if say they change two images just change the order and print it then that's it they they can sell it they can do whatever they like because the images themselves aren't copyrighted it's only the arrangement and the story um did you be very honest with you at that point I only wanted to get clarity what is copyrightable and what is not so for me I understood that text to image is not enough even though some of these images had some editing but it wasn't a lot of editing it was maybe just adjusting

Kris Kashtanova: few pixels with like clone tool not a lot so then I got curious what happens if I

Kris Kashtanova: make a drawing I made this drawing by hand and then at the time you could use like it was quite hard you could use Laura to get a similar looking AI image so I submitted the second walk to see how much human how much human interaction do you need for that to be copyrighted and it took two years to respond and they did use my work in the copyright office guidelines so this is the copyright of guidelines it was published I think two months ago part two copyrightability and they are looking into my walk and they say that the drawing is copyrightable and like the output is copyrightable without any non-human expression and then they write that the work can be now registered but you need to show that you put a lot of human creativity into it to be copyrightable so

Jon Radoff: not only is to prompt things into existence yes yes prompted content could be a kind of copyright yeah

Kris Kashtanova: if I can ask a weird question like so if I if I crafted a prompt and I wrote the prompt and the prompt is my expression of wanting to invoke something to existence and I generated an image with the prompt and the image and the prompt were paired is that and would that qualify in your opinion as being like you know or this is a great question because what what the copyright office said for now is that every single case they will look case by case so I encourage you you can you can test it out it doesn't take a lot of time you can you can you do it online you put together that thing and you submit it and they will respond to you but what I know so far is after this copyright guidelines part two there are a few people who already posted but they works for copyrighted there is one I think called something about cheese and someone wrote that that's the first copyright and I mean it's the first copyright after this guidelines it was and it was very interesting what was copyrighted and then I know a few more people who copyrights it like there is a friend of mine she's an artist she draws flowers and she used a reference image from AI but then she drew it exactly the same so it's like if if you would put them side by side it's a very realistic drawing it's exactly like it looks the same that is copyrightable as well like if if if if their pixels are basically drawn and then those cases will keep collecting so to answer your question until there is someone who actually does that we will not know that's exactly what I was doing I was doing it before I got employed because now I'm employed I just don't really have

Kris Kashtanova: as much time to do that but I am very glad that they have some guidelines right now it's still an

Kris Kashtanova: ongoing process and it still will keep evolving as more people try to copyright their works and they show like human input so I mean it will be great if people actually try to copyright something and see what is copyrightable what is not like for me that was all about that it was just about getting clarity it's such a new thing and I'm a researcher I'm very curious what is what is going to happen again I'm not a lawyer myself so for me it's just none of us are lawyers we

Jon Radoff: should not be a lawyer see I'm telling you here is legal this is not legal advice absolutely not

Kris Kashtanova: no legal advice here they only advice they can give create stop

Jon Radoff: so is like we've got before generative AI which is very very new still we had more than a couple hundred years worth of case law of all kinds of copyright issues and what's considered truly copyrightable versus what isn't and we're at the very dawn of figuring out the rules around this I think maybe what we could conclude from your own experience Chris is just text prompting something in existence that's not enough but that plus some in the US in China it is enough

Kris Kashtanova: there are different laws I was very curious about reading different laws in different countries and apparently there are countries where it is enough also in Japan depending on how many iterations you make but again they will always look case by case it's such an important part of it that case by case thing that what worked for me may be very different if you use cases slightly different so what I know though is that that guidelines that if you have enough human human input and you can show it that thing can be copyrighted

Jon Radoff: well we're getting towards the top of our hour so we have this last question from Ryan but Ryan I think I think you're asking almost a legal question I wish we could answer that I think Chris has the right idea which is not a lot of cases have been tested yet there's really no harm that I'm aware of again not a lawyer so I can't actually say that 100 percent but I don't think there's any harm in creating something and then attempting to copyright it and see what the copyright office has to say about that and as more and more of these things happen the rules will fall into place as Chris mentions the rules are also going to vary from country to country like China has a different set of rules I don't know if I think of China the best exemplar of intellectual property stewardship in the world but they have a different approach to it where they seem to be willing to grant much more copyright ability to things that are just generated out balancing all of the stakeholders in this process is a really really hard problem and it isn't a problem as simple as oh we found all the patterns that represent these kinds of structures that are recurring in art anyway that artists would learn in school and just because computers learned it as well you can't use it because it was sourced up the internet like I don't think it's as simple as that it's a much more complicated much more nuanced problem than that we'll probably have to bring a lawyer on actually I think yeah that's a good idea if you invite if you invite

Kris Kashtanova: actual AI corporate lawyer that would be a very interesting conversation I just want to mention a few things before we say goodbye because I prepared I wanted to show you some creative examples of AI so I'm going to just do it real quick to do one more minute so all right okay let

Kris Kashtanova: me share with you the rest of my presentation so first of all what you can do is you can make tiny

Kris Kashtanova: video games with AI you can create all of your assets with AI and you can use any chat but you

Kris Kashtanova: want charge a pt clawed gemini whatever you like grog and just like I used Adobe 5 Lite to make

Kris Kashtanova: assets for it like graphic assets you can use all the items you have or you can even use your own designs drawings whatever it is to make it like for yourself or for your kid this is like a cool creative coding exercise because you can make simple games but now you can also make a bit more challenging games then you can make scripts for after effects you can make scripts for illustrator like this you can make 3d you can make 3d text and turn it into through AI to make it look really cool there is a Renaissance and typography right now because you can do so much with AI like this one I just used illustrator and made some text turned it into AI also there are lots of use cases with 3d and AI that you make a transform or you can make a 3d design of it so there are lots of use cases I

Kris Kashtanova: already showed you how I animate my photography visualized stuff so I hope you'll just give it a try

Kris Kashtanova: and share what you are making with AI because every day people come up with a cool creative use cases I love seeing them online so yeah hopefully I think that's the takeaway from this program if

Jon Radoff: you're an artist and you haven't used these tools yet in the last 3 years whatever your misgivings are you still have to try these tools and understand what they can accomplish and if you are not an artist you should try these things just to see how it could augment your own creativity and what you could do with it so do not ignore this the world is changing it's not going to be the same so learn about these technologies and try to apply them to your work how it is and definitely go with Chris's idea we've got 655 people watching live right now if one of you is an intellectual property lawyer come on the program we'll talk about the legal issues actually would love to get a couple of you taking on maybe opposite sides of some of these aspects of things I would be a really really fun conversation so reach out we'll make that happen Chris I want to thank you so much for taking the time with me today this has been amazing and I want to thank the audience for taking part in this your time's valuable and thanks Arif for joining the program live yeah thank you take

Kris Kashtanova: care everybody bye