Tonia Samsonova discusses the challenges that artists face with AI and copyright issues. She shares her solution of giving artists the ability to train their own AI algorithms, allowing for a more diverse and decentralised AI landscape.

Tonia Samsonova is a serial entrepreneur and founder with a successful eight-figure exit. Currently she is the founder of exactly.ai, a platform enabling artists to train their AI models and monetise their unique aesthetics by leveraging their audience. Before that, she was a product leader at Yandex, a search engine company, and led a COVID-19 response initiative with a daily audience of 50 million. Her team's work on this project was shortlisted for Cannes Lions and earned European awards for creative data usage. Before becoming an entrepreneur, Tonia was a journalist and news editor.

Takeaways

• Exactly.ai allows artists to train their own AI algorithms
• Artists need to protect themselves and weaponise their creativity
• Reactions from the artistic community vary, with some embracing AI and others expressing concerns AI allows artists to become curators and tastemakers, as they can mass-produce images and choose the best ones.
• The partnership between human brains and AI is what makes us unique and beautiful creatures.
• AI can elevate the average level of skills and open up new opportunities for collaboration.
• AI can help people with dysgraphia or language barriers overcome their challenges.
• AI models can be published and rented, enabling specialised tools and collaborations.
• It's important to manage expectations and not rely solely on AI for decision-making.
• The future world will still depend on education, curiosity, and moral values.

Links relevant to this episode:

• Tonia Samsonova | LinkedIn
• Exactly.ai

Thanks for listening, and stay curious!

//david

--

Tools we use and recommend:

Riverside FM - Our remote recording platform

Music Radio Creative - Our voiceover and audio engineering partner

Podcastpage - Podcast website hosting where we got started

00:01 - David Brown (Host)

Hey Tonia, how are you doing?

00:03 - Tonia Samsonova (Guest)

Good, how are you?

00:04 - David Brown (Host)

I'm very good, Tonia; I really wanted to talk to you on the show because I think originally I saw some posts on LinkedIn or something like that where you were talking a little bit about what exactly AI does, and I think it's really, really interesting. And, considering a lot of the challenges, I think it's really really interesting and considering a lot of the challenges, I think that artists in particular, have with how the current generative AI works, I think you have a really interesting solution, and I wanted to dig into that with you a little bit to see, kind of, how you came up with the idea and all that. But before we get started, maybe you just give a little bit of background: how did you get to the point that you thought of doing this idea and all that? But before we get started, maybe if you just give a little bit of background. So like, how did you, how did you get to the point that you, you know, thought of, thought of doing this?

00:51 - Tonia Samsonova (Guest)

Thanks, David. Thank you for having me, and uh, yeah, I would be happy to explore with you the challenges that creators are facing nowadays because I think they're quite big. There will be a bloodbath; it's coming.

01:04 - David Brown (Host)

You're absolutely right.

01:05 - Tonia Samsonova (Guest)

01:37 - David Brown (Host)

When.

01:38 - Tonia Samsonova (Guest)

02:28

I was pregnant, got birth with my first child. I saw the journey. I thought, wow, that's an amazing tool, super cool. But it's one and the same thing happening once again: tech companies taking data from the internet, putting in their technology and then selling access to it to everybody else. And then they promise that they will somehow compensate or reimburse everyone whose data was used. But that would never happen. That happened.

02:54

This story happened to the industry I was working on with. I was a journalist. These tech companies came. They really invented a very good algorithm, and the big data revolution happened. Once all the traditional media lost their advertising income, journalists lost their revenue. We're still trying to cope with this situation. We're not there yet. We can see what happens to the world where a very good industry doesn't operate. So now we have less freedom of press, less diversity of media and everything else, and you see what happened to the world. Now I thought well, well, well, I don't want the same thing to happen to art, what happened to journos. And then the solution is super simple. Instead of taking the data and training an algorithm, just give algorithms to artists. That's it. Make sure that any artist, no matter who he is, the background he has, can train their own AI, and then we will be living in the world with 20 million versions of AI, instead of like three big companies having three big AIs trained on the artist data of 20 million people.

04:03 - David Brown (Host)

Yeah.

04:04 - Tonia Samsonova (Guest)

Sorry, I promise not to speak, but I'm very emotional about that.

04:07 - David Brown (Host)

No, please do. That's exactly what this chat's about.

04:11 - Tonia Samsonova (Guest)

Yeah, I mean, like I promised not to you prior to the conversation, like a little background. It's like I promised not to pitch too much, but then when you talk about that, it's like what's coming naturally. I've been interviewing one of our illustrators today. I usually run a lot of interviews to understand what's going on. But she said her name is Barbara, and she said, look, I'm like musicians are at least protected by labels, so they have someone to go to to ask for help. All system illustrators are not protected by anyone. And they ask like, what are you going to do? She says, like, I tried to delete all of the illustrations that I've published in my Instagram account, and I wasn't able to do that, so I feel absolutely vulnerable on the road.

05:06 - David Brown (Host)

That's really interesting that you can't remove it.

05:12

I think it's a bug or something, it's a hidden feature, yeah, and it's probably at this point it's probably too late anyway, because everything will have already been ingested and trained. , you know, so, I, I, yeah, it's, and this is the thing, right, like you know, the, the barn door is open, and the horse is bolted already, and I think I mean the whole concept of it, though is super, super interesting, and the and the fact that you've been able to sort of deploy that in a meaningful way so that people actually can use it, I know well, I guess the obvious question for me then becomes is how do people know that their illustrations and their data are safe with you and what makes? Do you know what I mean, what makes your tool different, and how do they know that they're protected? And again, this isn't meant to be sort of a salesy type conversation, but that's the logical question, right, is how to? How do artists feel comfortable that they're protected from what you're doing?

06:18 - Tonia Samsonova (Guest)

I feel like, technology wise, we are not there yet where we can say that anybody, anybody, can feel protected. We're not there. As much as I want to say, go to exactly and you will be protected. It's like steps towards the protection, but you never. You can't feel safe about that. And first thing is, yes, it's scary if you lose, so you can't be scared and you can't protect yourself. Instead of protecting yourself, you should weaponize yourself. So it's too late to protect anything. So I think this whole attitude let's protect the artist is not something that should be happening. When Martians are invading your planet with super lasers, it's too late to protect yourself. You have to weaponize and have the same lasers, and that's when Martians are invading your planet with super lasers, it's too late to protect yourself.

07:09

You have to weaponise and have the same lasers, and that's what we're doing. So the biggest thing of protecting is you can train your own AI on exactly, and you know what it belongs not to me and not to exactly. It belongs to the person who's the owner of the data the owner of the data. The most important thing is that we're trying to make a market standard. Whoever owns the data owns the algorithm. That's the foundational idea of exactly. So if you come to us, if you upload your imagery, if you prove to us that you have copyright to this imagery, which means that it's produced by you or you're obtained a licence to, trade on the symmetry and you create an algorithm which belongs to you and everything that's created with this algorithm also belongs to you.

07:54

So it's like you inherit the images that are prompted. Images inherit the provenance and the copyright and the legal framework of the images that are uploaded. It doesn't give you protection, but it gives you a right to sell those images to the customers. Because you can't sell to the customer images produced with majority, they won't buy it because they're scared that they will have a lawsuit against you. Now you can use the technology to scale up your production.

08:20

The second thing is obviously the easiest thing that can be done. There, you publish your AI algorithm on Exactly. Somebody comes to Exactly prompts. This model creates images with this model, uploads it to Exactly. Now the new model is published, and then this person has stolen from you, and the way we protect you against that is that we always publish smaller-sized images that were used to train an AI on so that if you see that it's yours, but somebody whose name we show used your images, you will know for sure because currently, it's very difficult to understand which AI is trained on which data. Right, we make it absolutely transparent.

09:08

Then, the third thing is that when you train AI on exactly when you publish it under our T's and C's, you have to confirm to us that those images are yours and if it's not, then you're lying to us.

09:22

Then we are the victim of the crime alongside with the creator, who is actually, like, really a victim here, and then we partner with this person, and obviously, if you're an illustrator, it's very difficult and expensive for you to go to court and to basically to say something well, with us, it's a bit easier. So I mean, and this is the prevention thing because it's a bit easier. So I mean and this is the prevention thing because it's easy to steal from a child. It's quite difficult to steal from me. You know I'm feasting, so and that's how, on the legal side, to partner with those who are the most vulnerable here. Then there are some other things, but you need to know for sure if your image is published on the internet, I can make a screenshot of this image and somewhere or like of your Instagram, and they don't even have to be a big tech company train subversion of the model on your imagery and now I can put your images the same way as you are. So you are never protected.

10:23 - David Brown (Host)

Yeah, yeah, of course, of course, so, I guess. So does that? Do those models then run locally on someone's machine, or do you do you basically split that out on your end and then you say, okay, cause I guess what I was getting to with the last question was if you had one big model and then do you know what I mean? And everybody's putting images in and then they get a piece, but I understand what you're saying now that it creates a whole new instance of a, of a pre-trained model with that's ready to go, essentially waiting for some images to go into it, and then you build a its own model there, so it's a custom model for that person and then it's just dedicated to them. So is that something that sits on your servers, on your side, and you manage that, or is it like an app that you download and someone does it on their local machine?

11:18 - Tonia Samsonova (Guest)

It's currently on our side and we never thought that there will be a need to do it locally. There might be a need and we're open to that. For now, we're doing it on our side and I think it's cheaper for the user because we spend a lot of money to train an AI model for you and to generate images, but in terms of the framework, we can do it either way, to dig a bit into how AI works there. So, basically, what's going on on our end? We have a technology which is capable of understanding your style, and that's one thing. We have a technology that's capable of producing imagery from prompt. We never expose this technology that's capable of producing imagery from prompt to the user, to the end user, because it's useless, because you have other tools and for us I mean not because we're so ethical it's not how we generate revenue like you have DALI, you have Majority.

12:23

Their technology is much better. If you just need to prompt just a random model and to get average image, go there. That's not our business. Now we have this technology. When you upload your images, we create a small version of AI, repair this technology with your data set and the new piece belongs to you. Now this piece can be either exposed to external world, so you can publish it so that anyone can use it for a fee, or you can publish it so anyone can use it for free. It's up to you to decide. Or you can decide not to publish it, and some people don't want to publish their models.

13:01

Most of them they work for the clients, they write books or they do something. They keep it to themselves. The way we generate revenue is those builders of AI tools, of AI models and their data sets. They pay us a subscription fee and for a subscription fee, you can train your own AI and that belongs to you. And then, on top of that, you can either publish it and generate some extra revenue Instead of working as a freelancer for the company. You can just publish your model and get a passive income from people who want to prompt your model, or you can keep it to yourself, because maybe it's more lucrative to you to keep it to yourself interesting, and what sort of reaction have you had from the, from the artistic community?

13:46

Well, I think artistic community is in two camps now. It's against AI and pro-AI. And I know that in some ads agencies or creative agencies, anti-ai rules are so strict so they don't even use like meeting recording devices that transcript something and they can understand why they're doing it and they can understand the emotion they're doing it and they can understand the emotion. Obviously, people who use exactly they're more open and I think there are several reasons why people turn to exactly.

14:18

One of them is uh, we have an artist, a photographer, nancy olivier, and she's one of the earliest users of exactly andla, and she said apart from being a photographer, I'm a professional psychoanalyst and I know that if you're scared of something, in order not to traumatise you, there's something, you have to explore it. So I decided to learn who my enemy is, and then she figured out that there was Xavla. She was like against us, obviously in the beginning, but then she's our critic credit, which is like very supportive, and she's the foundational person to the community. The second night, a quote that I heard from one of the very popular french designer. He said well, I look, I work at the design studio. I have young children working with me.

15:06

He means like 19-20 year old people I knew what you meant, yeah and he says, like look, they can afford themselves not to learn what their eye is, but they're 42, I need to know if. In other words, otherwise I will not be able to be in the business. And I think this is the attitude of people who have a serious career already and who've been in the industry of design and art for decades. It's a very common thing that they're saying. One guy said like you know what I've started to do design and art before PCs were around? Yeah, and it was the same situation. And I just quickly realised either I learned how to use PC for my design and art or I'm out of the market. He said, like ages, I have my savings. With those savings I went and bought a PC computer and that was the first one. And they don't regret. So yeah, and that's one attitude.

16:03

The other one is well, I feel artists typically, there is a difference between artists and illustrators. Artists, that's one attitude. The other one is well, I feel artists typically, there is a difference between artists and illustrators. Artists are not scared of AI at all. So those people who do conceptual art, modern art, just copy me please, like. What's the point?

16:22

Whatever you want with my model and you can play with it. They are very happy to see how, like, their oil paintings are copied or something, because they never use it, or as a final product for them, with symmetry, search studies and a lot of things like that, and there was like there was no point in copying someone's art. This is so connected to the human being and there is like you can copy it. Please, yeah, makes 100% of it. It. It's not diluting the fact that, like with art, copying art is not diluting the art piece, because that's how art works yeah with illustrators is quite different.

17:00

Let's say you work for a brand and your job is to produce imagery in the brand style and you got paid to develop this style. Now the competitor to your brand from different country comes, scratches your data and the company, instead of hiring you, just trained the model on 10 images and now you lose your income. Well, that's the problem, then. That's the problem, and especially like illustrators. Well, most of them are doing really well. I know people who make like 200K per month because they're very, very popular.

17:36 - David Brown (Host)

Wow.

17:38 - Tonia Samsonova (Guest)

But it's like in the football, like top 1% is making a lot of money. The whole industry is quite a tough thing to to, to do. And also this varvar. And she said like I've developed my style within 15 years. For 15 years I was figuring out what's the best way for my characters to look like how should I draw a nose, how should I draw hair? 15 years I was thinking and expecting this. Now it's like this copy, like am I scared?

18:14 - David Brown (Host)

yes, I'm scared yeah, no, it's , it's interesting and and it's also interesting I guess no one's really talked about the difference between the different styles of artists, like illustrators versus photographers versus painters and that sort of thing. So it's interesting that there's a little bit of a difference there and I can totally see how that would be the case. I was writing some notes while you were talking because I didn't want to forget to come back to a couple of things. One of them was that and I've mentioned this on a previous show before, but I remember, I remember when, before digital cameras and there was the same discussion in photography about well, is a digital image the same as taking a photo on film? And there was a lot of consternation in the photography community, and there were, you know, there were a lot of arguments and many, many beers were spilled over. You know, is that really a photograph? And then the knock on from that was then you had Photoshop and then that was like, well, if you've done anything with Photoshop, it's not a photograph anymore. So we had to go through all of those iterations and I agree, I think AI as a tool is just another tool.

19:48

I mean, who does, who illustrates, let's say, harry Potter right, like that's the biggest probably series of books next to, like Lord of the Rings and maybe a couple of others in the world, and the person who started in doing the illustrations for that in the beginning and all that hand-drawn custom design and all of that. You know that is their style and that is the style of those books and for somebody to go in and then copy that that's a huge potential issue. But I can also see where someone who does that if they train a model on their style, and then somebody comes back and says I need some more art for Harry Potter and we've got this whole new series of things that we want to do can you like they could use it for inspiration? Maybe they might not use those final images, but they could use it for some inspiration or to look and see what does this look like? Or you know how it might, how something might work or not work, and I could see where that could be really really valuable.

20:50

It's like I was talking to Daniel Bedingfield the other day and you know, from a musical standpoint, he loves AI because it's helped him actually be more creative, because he can take crazy ideas and he can try it in 30 seconds and just say, oh well, that doesn't work, instead of spending hours working on something and then realising that it doesn't work.

21:12

So it's interesting, yeah, that you were talking about that, and this is leading to a question. The other thing that came to mind is and I have a friend who runs a gallery and she always talks about the fact that a lot of artists who get older and maybe now are disabled or they can't physically make the movements to do the art anymore, have found a lot of solace in using AI to try and be creative again, because they aren't able to do that. And so I was wondering if you had, have you had any stories or have you had any users that have come back and said you know, this is amazing, because I can't create anymore, but now I can because something happened maybe I don't know, it could be they got in a car accident or you know, or whatever and then they say, well, I've trained, trained it and now I can create stuff that looks like me I think I have two stories like that cool.

22:13 - Tonia Samsonova (Guest)

Okay with one.

22:14

I I'll be cautious of using names here, but of course one is uh, what is the story of a person who started to experience he was unable to move his hands essentially, and and then, at some point, uh, he was unable to move his hands essentially and then at some point, he was only. He wasn't even able to type, so he used this voice translator into the interface For him being able to produce images in his own style, as during the times when he was capable of drawing with his hands, something that basically prolonged his art career. The second story is a story of a artistic family and their mother sadly passed away and her son, an artist. He trained AI on his late mother's work and started to prompt it and, together with the sister, they were looking at those new images and he said for us, especially during the first months of grieving, it was something that we were doing during some evenings and it was as if we had conversations with her. Then he published this work on social media and the reaction was probably not.

23:47

It was like different opinions of people. One of them said that it's, it's back towards her legacy. The other said that it's a good thing and he said and what he said to me? Like you know, for me it's a continuous conversation with a person and I see how her vision is reflected into something new and I feel like, if it works for him, that's enough and nobody can judge. But then he said the funny thing when he then at some point he got annoyed with all these commentators on the facebook and and someone of the artists said, yeah, but it's not a real art and, uh, anyone can do that, and probably he didn't like the style of the commentator who said that and he said well, you know, it's very difficult to train your eye on your work because there is nothing to copy.

24:41

So that's exactly, yeah, and so and probably like, if we think about that in a broader context, artists previously. Obviously, new technology always changes a little bit the role of the artist, or at least shows it with a new facet, and also it allows different people to enter the art world. So if, like in 19, in 70, in some in the 17th century, if you can't draw, you can't make a portrait. So there are many people who maybe can do a great portrait and we now know from the photography is the skill of making a portrait is not about, like, making a very accurate likelihood, it's about capturing something different. Now we have photographers who might be very shitty in their sketches but very good in photography. Same goes with AI and I think AI allows the new type of artistic persona to enter the scene.

25:38

And this is an artist who is a curator. If you can, must produce images in your style. Your job is to decide which one is which. And the other artist told me, like I'm still the artist, because even if I have 200 images that generated under my prompt with my model, it's still me who is choosing which one will go and which one will go to the bit. So, artists today, because scarcity of the image becomes lower, becomes pacemakers and becomes the curators, and curatorship becomes a more important thing. But you're a creative director. You know it better than me and it's.

26:23 - David Brown (Host)

It's interesting that you say that, because that's exactly what daniel beddingfield said about music and he said you, you still have the trend makers. And it's, the trend makers are the ones and and that's you know, you're as musicians musicians are going to pivot a little bit because it is going to be easier in some circumstances to maybe to generate something, and so it's those people who can, who can pick the trends and pick the styles and understand and pull the stuff, it's the curators. So it's very interesting that that two people from two completely different fields have come to the same sort of conclusion as well. And I think you're absolutely right and I, I think there's still a lot of hangover. And again, this goes back to the same thing that we had in digital, you know, with photoshop, and that is, there's always this hangover, I think, of people who, like I, learned how to you know how to write code when I was in, like in the 80s, when, when computers very first came out, and you kind of think, oh well, I used to have to do this stuff by hand and it was really hard for me to do it.

27:31

So, because you're, it's easy for you to do it. That's somehow cheating, and I I think every generation has that. With the next generation that comes along, because the technology gets better and there's some sort of idea that because you didn't spend hours and hours slaving over something, that it's not as worthy and that's only because we slaved hours over stuff for years and so we want to feel like that was worth something. But but you know my son, who's in his teens, you know, when he gets to his 40s and 50s there'll be some other technology and he'll be looking back and going, yeah, but you know, we used to have to do this by hand and now it just gets done for you and it's not the same and it's like, but it is kind of the same. So I think that there's a little bit of that kind of carrying over as well, do you think?

28:21 - Tonia Samsonova (Guest)

I feel so, but I also feel that knowing how to do something from scratch is a very valuable skill. For example, they can know how to put plants into the kitchen garden and then to grow them and then to make soup from scratch, and I feel there is a lot of value into doing this type of work, because when you do it, it's another part of your brain that is thinking, and I think that when you're coding or when you're writing manually with your hand or your piece of paper, it's all about how you play with your own brain and with your own skill set. And human brains are very developed, very fun. In a way they're built. So, yeah, and I feel like this partnership with a human brain, with AI, that's what makes us super beautiful creatures, ideally, and I think that science fiction was always about how we take a human and how we put something in there, literally in their head, that will upgrade them. Like in Hitchhiker's Guide for the Galaxy, they would put a small fish.

29:43

And now you can learn and now you can talk and understand languages, but then take progress, Like we have this fish now which is a chat GPT-4.0, with a voice. I can speak in Russian to you and you will be able to understand it. It's just the frustration is that it's not within our body, it's somewhere external. But I feel that our attitude towards ai should still be like it's part of the human body which is not a human body. Imagine how much more scarier ai would be, would have helped me with it If, instead, if the form factor would be not laptop but injection of some liquid into your vein or a piece of tablet like in Matrix or some operation there where they have to install a chip into your brain. So we're lucky that we don't have it, but it doesn't mean that this external is a threat. It's still us, we developed it and let's just yeah.

30:49 - David Brown (Host)

And that's the problem Humans are always the problem.

30:56

I'm not sure computers will be the solution, but humans are always the problem because we're all flawed. We all have, you know, none of us are perfect. We all have our own biases, we all have different, but that's because we all had different experiences and that's what makes it great and amazing and fun. And you know it's that people are different. And you know it is that people are different. And, yes, we're going to get some weird average of everything in AI. You know, and that's part of the complaint I think that a lot of people have about it is that you get mediocre everything but but mediocre is better than a lot of people can do in a lot of different subjects.

31:39

all the time but mediocre is better than a lot of people can do in a lot of different subjects all the time, and so I think we will eventually raise, we will continue to raise the bar.

31:47 - Tonia Samsonova (Guest)

And I think what's happening now is that you can't be superb in everything I mean. I code not mediocre, I code like very bad.

32:00

Now with the help of AI. No, you probably. No, I'm worth it. Like, trust me, trust me. Now I can get to the mediocrity, but with some skills maybe I'm like very high performing, but with the others I'm below the bar. So AI helps me to be like like to elevate my average level and that's. I think that helps a lot of people and that helps to develop things. And I think one of the things that happened with people who have dysgraphia or dyslexia, for example, like 50 years ago, if you're a child and you don't know how to write, people will say you're stupid, you don't read books and education is closed as an opportunity for you and especially like you will never be able to join the job market. Then you have autocorrect, then you had Grammarly, now you have ChatGPT. So a lot of the people who struggled with writing, or a lot of the immigrants, like the whole, like yeah like.

33:01

I'm an immigrant, I will never be able to write in english the same way I write professionally in russian. And for me before, like fundraising, for example, it was a question whether I will be able to enter conversations with investors. Because, imagine, like, like, because when you're talking to a person and there is no problem with that, it's the country which is very welcoming to immigrants. But if you take any other country which is more hostile to different people, if a person is speaking the same language you're speaking, makes mistakes which I'm probably doing now then you feel less of this person. You can't appreciate that this person is probably professional in something else. You degrade the whole.

33:47 - David Brown (Host)

Yeah.

33:48 - Tonia Samsonova (Guest)

And now we have like so many immigrants and migrants in the world, and the attitude towards them was based on the way they use the language. Well, now they don't have this problem anymore.

34:00 - David Brown (Host)

Yeah, you're absolutely right. I hadn't really kind of thought about that before. And is it the same the other way in Russia? I mean, I think people are people and I think, yes, some places are worse than others. But is that a universal thing? Like if I, you know, because I'm a native English speaker if I moved to Russia and tried to speak russian and learn russian, I would get the same thing right. Like people would have a hard. They'd just be like oh well, he's an immigrant and he doesn't know how to speak russian, and I would. It would be the same thing there, or do you think they're? Are they more welcoming or less? Do you think?

34:33 - Tonia Samsonova (Guest)

well, I think it's like difficult now because russia's obviously at war and it's related to ukraine.

34:38

So it's not like hostility to immigrants is probably not the biggest problem that Russian society is facing now. But yeah, I think that, and I saw that many times, that when a person tries to speak Russian with somebody and Russian is not their first language, the learned language a person who only knows one language will always correct this person. And I was just like why would you correct this person? Like, this person is speaking like your language and that's like her language. But you always feel better about yourself when you're not making mistakes. Yeah, but anyways.

35:20 - David Brown (Host)

You're doing just fine, don't worry, it's all good. But no, it's always. It's interesting, and I always find the language discussion quite interesting as well, because, even though I speak English, I mean, I always make a joke of it but people ask me if I speak multiple languages and I say, well, I'm fluent in American and I'm fluent in English, and I'm not sure that I'm fluent in American and I'm fluent in English and I'm not sure that I'm fluent in either one anymore, because now it's all mixed up and I have no idea what I'm talking about. But anyway, so to go back to something as well, I'm curious to know your thoughts on where you think this goes. Where does it go next?

36:00

I mean, I assume that what we'll start to see and maybe this is something on your roadmap, I don't know, but I could see where you could take this model and you can say well, we've actually built this model around doing images. But we could do the same thing for authors to say give it all of your writing that you've ever done, and we can write stuff very specifically in your style and your voice. , that's, you know, protected from the main model. So you know, other people aren't getting access to it Like you have to spend the time to train it, but once you do, it's much more like you. , do you think that we're just going to see tons of specialist tools like that for all different aspects of our lives? Where does it go?

36:46 - Tonia Samsonova (Guest)

Well, for me, the vision is that usually I was always trying to hire the best people to work for my projects and most of the times if you try to hire somebody you really like as a freelancer, they're busy. That's how you know that like the person is doing his job.

37:06 - David Brown (Host)

That's how you know they're good is you can never get them.

37:09 - Tonia Samsonova (Guest)

Yeah, and at the same time, when people were trying to hire me for their products, they always were busy and I was always sad that I didn't participate in this and that, because sometimes you have expertise and you want to use it to help somebody and come to so many things that you want at the same time. So I believe that for me, the vision is that AI will open a new way of collaboration between people. I want to see people who have different, very, very cool skills to publish their AI models on exactly and yes, it's not only about artists for me and not only about designers. But since you want to do it like we've started so long ago, like pre-GPT, it was a different era, so it was very difficult to explain. I mean, professionals, the industry knew what it is, but if you will go outside, you think GPT, it was impossible. It wasn't possible to explain what.

38:07 - David Brown (Host)

AI is.

38:09 - Tonia Samsonova (Guest)

So for me, the vision is simple. Let's say you are a recruiter, or let's say you're going to a conference and you want to meet people there. There are 2, 000 participants in the conference. You have five slots. You have to choose which five conversations that you have you you're able to have, and what will you do? You will like reads through linkedin profiles of development people. Well, probably it's not your first conference. We will know which connections work well for you and what not.

38:38

Yeah, so instead of like doing it, of doing it manually and spending your own time, you can ask your personal AI to sort at least 500 guesses and sort out all of the rest, and then you will do this. Now let's say you're a recruiter and you're trying to hire the best product manager for your specific startup. You do the same. You go through LinkedIn and you look through those profiles. Now let's say you're a recruiter and you're publishing your model of how you pre-select the data set on exactly. I can now hire you or rent your model out, pay you a fee. That you will will say like. You will say like if I work personally with you, I have two conversations and do you agree? You will talk to me about that. Talk to me about this. You will like have another zoom meeting just to understand that they work like you spend a lot of time when you're working with a client like yeah talk about like and our directors never know what they want.

39:40

So you do the job, you send it to the. It's a lot of work. Essentially, when you work with a person, what you like, this person, this freelancer, spends 20 on the time on an actual job and 80 of the time to understand the client and to communicate the client. Okay, keep it for the number of hours that you have, but they will always be kept with your life, essentially because you're a human being. You will die at some point.

40:04

You won't be able to serve all the clients. Publish your modal that can do some certain thing and charge people who want to use this modal, but don't spend your valuable time on them and don't do mundane job there, while you can focus your creativity and your time on something that don't do mundane job there. Well, you can focus your creativity and your time on something that can't be done without you and then imagine how many fun collaborations we will have. Now you can work with the artists from Mexico whom you don't know, just because you like the model. If you want to test if it's working well, you don't need to ask you to make a test assignment, you can just prompt it and if you like it, you can just hire it.

40:43

It's a new way of interaction. Imagine you're watching TikTok, but instead of watching the videos, you can prompt those people's models. It's so much better in exploring each other, I mean, people will understand more about each other and will be able to collaborate. So it's like Fiverr on steroids, to put it simple. Like now, you have to go to Fiverr, choose a freelancer, pay for the freelancer, take a little amount of money, do mediocre quality work for you done. Nobody's happy, everyone wasted a lot of time. Do it on exactly.

41:26 - David Brown (Host)

That's pretty cool, that's pretty interesting, yeah, and that kind of gets around to. I'm trying to remember who it was, but somebody was talking about. You know, we're all going to have personal assistants and what we're going to end up with is our personal assistants. Like, my personal assistant will deal with your personal assistant to book the podcast recording, and we won't have to touch anything like literally, we'll just say, yeah, we want to do it, and we'll just tell our systems to go and do it for us and it'll just, it'll sort everything out on our behalf. And how does that?

42:00

You know, in the marketing world there's a lot of discussions about how do you move from a world where you're marketing to a human to now you're marketing to an AI, because now the AI is making some decisions on that person's behalf and so that person may never even see an ad for something or you know, or experience that discovery. So how does discovery happen when it's all AI and AI is just discovering other AI? So I think there's a really interesting wrinkle. That kind of goes along with that as well. It's like, literally, how does that happen in the background? And, yeah, it's interesting to think about. It's interesting to think about.

42:41

Well first of all about the advertising algorithms.

42:45 - Tonia Samsonova (Guest)

I used to work for the company who was optimising ads for users, so the job was to show you banners that you will click and you a naked human being with your naked, vulnerable brain, who's not protected by anything, is exposed to AI that optimises for you. Is that fair? Well, I feel like human brains are so fragile that you can't just do that, so I'm sure you would rather have your own AI that's facing the ad algorithm and then make selections. It's like being on Tinder, when you are chosen but you don't have a choice, because Tinder is two-sided algorithms. It has an algorithm of two sides.

43:27

44:29

Ai is a simple algorithm that's exposed to the data, learns the pattern and then generates something by multiplying a random number to this like other number and that's like don't have expectations and that's it.

44:45 - David Brown (Host)

Yeah, yeah, that's it. Yeah, no, no, you're right. And, , yeah, I'm conscious of time, so maybe one. One last thought from you is you said you've got four kids. So what, what type of a world do you see your kids moving into? And what do you think, like, how do you think AI is going to impact their world, that they're? You know that they're going to have to go out. They're going to have I don't know how old they are. I have a 17 year old, so he's just getting ready to go into work and they're going to be going into work soon and what type of environment do you think they're going to be moving into?

45:27 - Tonia Samsonova (Guest)

46:11

Now, look around you, look at the politics in Europe and people who vote for some kind of weird people, and those weird people says that that Nazis were not always right and like you said, you've been very lucky to and we probably are less lucky, but what I think is I feel like everything will still depend on the education that you have and the curiosity and moral qualities, and I don't feel like that there is a scar tissue that was going on, so it's definitely not something that we should prepare our kids today that our parents didn't do with us. I think the basic values and universal things like be kind, work hard, respect the community and just be curious and read a lot of books and play a lot of video games will always help.

47:18 - David Brown (Host)

I love it. That's an amazing place to start, Tonia. Thank me, thank you, thank me, thank you, thank me. I would like to thank you.

47:27 - Tonia Samsonova (Guest)

Actually, David, really thank you so much. It was like a very, very nice conversation. Thank you. Thank you for having me.

47:34 - David Brown (Host)

Thank you for coming on. I will have links in the show notes and everything, all of that to exactly AI so people can go and try it for themselves and hopefully, sign up and experiment and play around and, yeah, have a great afternoon and the rest of your day and it was, yeah, it was lovely, great conversation. I'm really looking forward to getting this one out.

47:56 - Tonia Samsonova (Guest)

Thank you, Dave. Thanks, have a lovely evening. Thanks for having me. Thanks, bye, bye.