Behind The Bots

Greg Kowal, founder of Fabbler, joins 'Behind the Bots' to discuss how his project harnesses the power of artificial intelligence (AI) to revolutionize story-driven game creation. Fabbler aims to democratize game development by enabling anyone to generate engaging narratives, characters, and interactive experiences using AI technology. Kowal shares his vision of empowering individual creators to make story games with minimal effort, much like how YouTube has transformed video content creation. By leveraging AI and machine learning, Fabbler streamlines the game development process, allowing creators to focus on their unique stories and ideas. The platform's AI-assisted tools help generate art assets, music, and voiceovers, making game creation more accessible than ever before. As artificial intelligence continues to advance, Kowal believes that Fabbler will play a significant role in shaping the future of gaming, where passionate creators can bring their stories to life without the need for extensive technical expertise. With the rise of AI and the growing interest in personalized, story-driven experiences, Fabular is poised to make a lasting impact on the gaming industry, empowering a new generation of game creators.

#ai #gaming #artificialintelligence #aitools  #contentcreation #contentcreator 


FABBLER AI

https://fabbler.ai/


FRY-AI.COM

https://www.fry-ai.com/subscribe
https://www.youtube.com/@TheFryAI
https://twitter.com/lazukars
https://twitter.com/

Creators & Guests

Host
Ryan Lazuka
The lighthearted Artificial intelligence Journalist. Building the easiest to read AI Email Newsletter Daily Twitter Threads about AI

What is Behind The Bots?

Join us as we delve into the fascinating world of Artificial Intelligence (AI) by interviewing the brightest minds and exploring cutting-edge projects. From innovative ideas to groundbreaking individuals, we're here to uncover the latest developments and thought-provoking discussions in the AI space.

Greg Kowal:

My name is Greg. I'm the founder of Faber. So my background in general is also in IT. So I've done some projects for large companies in Europe. One of the project was, like, really automated warehousing with robots.

Greg Kowal:

Those work for Ocadis like a competitor to Amazon, here in Europe, or at least they are trying to compete with them is in top 100 largest companies in UK. And, some cool projects that we've done there was, you have an app. You place your grocery shopping, and you get that order delivered to your doors in 15 minutes. I think this is not need in US because in US, you drive everywhere and you pick up your shopping, but in Europe, it was pretty, really cool projects to do. Yeah.

Greg Kowal:

People were super excited, you know, to do shopping and get this in 15 minutes, in in at your door. Yeah. So and then I've also done work in game dev, and here is where the this transition from IT into Faber comes. We have released a few story games. Those those were pretty successful games.

Greg Kowal:

We had, like, at the moment, almost 20,000,000 players worldwide, and making these games happen was really really hard. Yeah. In general, game dev is super hard industry. It takes you a year or 2 years to produce something that is later consumed in a week or 2 days, and you never can keep up with the content production. So in the middle of 2022, when first, stable diffusion 1 1.4 was released, we already knew a little bit about AI because of our previous experience for the IT companies, and we've realized that if stable diffusion 1.4 maybe wasn't great, but we could see that you could actually apply it to gaming and make it work.

Greg Kowal:

So we started first working on the, pipeline for creating characters, then we've created pipeline for creating scenes. And by the end of the year, we've created a proof of concept game that was entirely made by AI. Now I think this is very kind of achievable, but at that point, it was pretty kind of amazing to achieve this. And for some people, when we are showing this is actually a I made, they couldn't believe everything is a scam or and then ChargeGPT revolution started, technology started progressing, now I think no one will deny that you can make a game content in the game that is fully AI made. Yeah.

Huner Kallay:

Awesome. So when we're looking over it, it's dotai. So you got some depth on there and stuff like that. Can you just walk us through the stage that the tool is at right now and then maybe, you know, expand on that long term vision of where you see it going in the future, what you're trying to develop right now?

Greg Kowal:

Okay. Cool. So at the moment, you can use Faber to actually generate storytelling with minimal effort. So what you will be able to do is without experience, if I take my mom and I sit here in the front of the PC with the simple instructions, she will be able to generate a story with the characters, with the narrative, with the voiceovers and the music and export it as the video. But there is a little bit trick there.

Greg Kowal:

We actually record the video from the game engine. So what we do underneath is we actually build a game. And on the server, we record a gameplay as a video so that you can see you experience your story yourself. And here's where the the the vision is. So our vision is to actually enable people, like, democratize story game production.

Greg Kowal:

Yeah. So the storytelling module is the first phase. In the next steps, we want to add the gameplay modes. So really in the same fashion as you now create a story with minimal effort, you can generate games like a mystery games with choices or decisions, quiz games, and some, like, a textbook like gameplays. And we are actually, at the moment, also working with some companies.

Greg Kowal:

We can see that this process already works in some ways. So we have one company from Warsaw that took Faber, and previously, they were spending 6 weeks on producing a simple cutscenes for their puzzle game, and they already could automate that process from 6 weeks to a day. So I think that the this ultimate vision, what we want to achieve is is to take a random person from the street and make that person make a make story game, yeah, like, in in 15 minutes. And I think that this goes to kind of this reference, like, if you look at YouTube. So in 2010, if you go to a random person on the street and you ask that person, do you want to be a movie producer?

Greg Kowal:

They will have this mindset of, like, this is not really possible. You need to have all of this equipments, screenplays, cameras. There is some team required. But what YouTube has done, they've shifted mindset of people to to idea that every everyone can be a movie producer. Yes.

Greg Kowal:

So now you have 33% of Gen z, thinks that they will be the content producers. No one actually denies that you can actually start making movies like mister Beast does, and so this is very accessible. When today I will tell to any of you, like, come on. Let's make a game. This is still not very available to to people.

Greg Kowal:

You will say, okay. I need to have an artist. I need to have some writer. I need to have a programmer. I need to have all of these people to create a game.

Greg Kowal:

So the idea behind Fabler is that we want to solve that problem and make it accessible to everyone. And, ideally, we will trigger this individual creators in the future who will decide to make games. We can see that people really, really care about featuring themselves in games, which is very interesting. So, like, people nowadays have this their digital identity avatars. And when they create experiences, they typically ask, can I place myself in the game?

Greg Kowal:

We have such features coming. We actually already have them. They are not available. So you can upload your picture and create character that game character that looks like you. So you can create narrative with yourself and your kids in it or with your friends, and you can story tell.

Greg Kowal:

Plus, you can make that experiences interactive.

Ryan Lazuka:

That's awesome. So it's you're trying to make, like, the YouTube for gaming in a in a way, game creation.

Greg Kowal:

That's the that's the long term vision. Yeah. I think that there is long way to go because you need to create this like, there there is, I think, three reasons why people would want to create this. So what we experience now is when people use Fable, they tell us that it's magic magical. It's really cool.

Greg Kowal:

I really love it, but we need to give more meaning to them. Yeah. So first thing is make them earn cash. So we need to create economy where they if they create something, they can get an return from it. Second thing is let's make you famous.

Greg Kowal:

You've created a game and that game was released, was being for some people being famous is, you know, 400 people played it. Yeah. And they are super happy. If you go to Reddit, like, new YouTubers, space, you will see that people are so excited about even having 100 subscribers, 300 subscribers. They they they they appreciate that small small audience that follows them.

Greg Kowal:

So being famous. And the third thing is that to give them a sense of community. So make it more collaborative. So we're working on the feature where you can have the library of assets, and we envision that that we can have one creator create it's not even necessarily related to AI generated content, but having artists that will upload his own art and someone else will use that art and use AI to generate variants of that art. So if you let's say you are artist, you've created a character, you upload that character to library, and Faber creates the variants of that character that can be used in someone else's games.

Greg Kowal:

And then maybe we can go back to the first step, which is let's make a return to all of them. Yes. So everyone is everyone who uses Fabular is happy because they make a return. They have the sense of community, but also, they can be noticed. Yeah.

Greg Kowal:

I think people have this big need now today to be noticed by others. Yeah. So

Ryan Lazuka:

No. Very awesome. Now what so what would one of these games well, I guess a couple questions here is, are these games available right now in the beta program, or are you just doing sort of the story part of things right now and the gameplay is gonna come later?

Greg Kowal:

We a gameplay part will come later. We've already done, a work to prove of concept this and prove to ourselves that we can make it and we can. At the moment, we are taking it in kind of phase approach. So the storytelling is the first thing. We need to get it right.

Greg Kowal:

Yeah. So if we people can create meaningful stories, everyone can tell his own story with the tool, and we can have some game dev studios use it to actually tell stories for the games. So you are making a puzzle game. You are not expert in creating narrative, but then I can get FABELER, and FABEL will generate that. In general, in games, what you have, you have these 2 types of gameplay.

Greg Kowal:

You have real gameplay when you play a game, and then you have the meta gameplay that creates a need to play a game. Yeah. And and you have different types of meta gameplays, but you also have this very strong story meta gameplay. Yeah. So you let's say you play Witcher and you think you play it because you like to, you know, the gameplay mechanics, whatever, but there is also the story that you want to uncover.

Greg Kowal:

And that's the reason why you play the game in the long term, and it provides the provides the retention, let's say, mechanism so people are coming back and playing the game every day. So the story telling part is first. Once we get it right and we see that, what Faber produced is meaningful, We will start implementing the, decisions game place. And so we during that beta phase already, we discovered multiple needs among the creators. Yes.

Greg Kowal:

So for example, what I've mentioned earlier, initially, we our impression was that people will be really excited about generating things. Mhmm. But what it appears is that this repeated feedback that people want to actually upload their own images or have an image of the living room and transform it into game art. Yes. So so something that we discover is that people really want to implement their digital identity and their environment as digital kind of entities in that stories.

Greg Kowal:

Yeah. So this is what we are working now on.

Ryan Lazuka:

Yeah. It's it's really interesting because a lot of influencers out there just use social media to promote themselves, but I've never heard anyone saying, well, I'm gonna create a game of myself and use that to to get more influence or followers. So that's kind of interesting.

Greg Kowal:

And this is exactly what I wanted to what what I was initially saying is that people don't have this mindset. Same way in 2010, no one would ever consider themselves. I'm gonna like, obviously, it's different. You we always gonna have this professional studios making professional content, but we think that there is this space to empower. Like, you if you also, if you go to ready spaces, you will see people asking questions.

Greg Kowal:

Yeah. I'm I'm thinking of doing games after hours. Is it hard? Will I be able to do it? And the answers are, oh, man.

Greg Kowal:

That's gonna take like, if you spend 40 hours a week, you you probably can create something meaningful in a year, but don't, like, try to make it small. Like, small project. Yeah. So it's still in the same space making games in very similar space, movies production or movie content production was actually in 2010. Now with the saw Solar release from OpenAI, I think a lot of attention.

Greg Kowal:

So so so but I think it's always it's like with with everything that changes, you always have, space for people who will do it very professionally. I think that AI will never replace professional movie making because people will still want to see the this this this kind of spark of people in the movie. And once they they they will understand it's all AI, they will not be that interested. They need to have celebrities that they can follow, but it's not like it will be 1 or the other. Yeah.

Greg Kowal:

You're still gonna have big industry for the AI generated movies and probably the professional, normal, classic way of making movies will shrink, but it doesn't mean that it will disappear. And with the games, I don't think that there is any other future than having, you know, like individual creators making games at ease without having, you know, large teams of people, designers, outsourcing, etcetera. So so that that has to be the future. Yeah? And I'm sure that I'm sure that we can make it happen.

Greg Kowal:

Yeah?

Ryan Lazuka:

Definitely. Well, like, so what is one of these games look like? It sounds like you have you're already testing the games, behind the scenes. Like, what are the first games that you're you're gonna release look like, and what would you like them to be eventually

Greg Kowal:

are focusing on the games such in the in the story games. Marketing generally is huge, and one of the most popular gameplays especially for the female audience is a store like a story decision games. So where you play where you experience narrative, you make decisions, and that the decisions somehow change how the narrative goes. And the coolest thing about Faber is that because we control all of the narrative context for you, you can make that decisions very, very meaningful. So the problem in producing such games today is that to manage all of that tree of dependencies, is just hard.

Greg Kowal:

So what people do is they typically make decisions that are not really meaningful. They they change little in the game or they are faked. So in Fabler, you'll be able to create, like, very complex decision tree, and all of the decision can be meaningful because you have minimal effort to create this. 2nd gameplay that we also envision is quiz type gameplay where you kind of experience a narrative and you are being asked question and you can find whether the answer were right and wrong. We have some cohort of beta testers that come from the educational background.

Greg Kowal:

We can see that. So they look at Faber as the means to teach kids at school or or educate someone else. So this is another thing. And the future, like a future future, we envision that we could implement gameplays such as hidden object or point and click gameplays, like anything that has a story as backbone. Yeah.

Greg Kowal:

So we are not focusing on kind of game place where they're action oriented. It's all about the story and using that gameplay to tell your unique story to the audience.

Ryan Lazuka:

So it's a kinda is it kinda like a choose your own adventure game to begin with, like, most type of games?

Greg Kowal:

Yeah. Yeah. Yeah.

Ryan Lazuka:

Got it. Cool. But you could have a celebrity on there, an influencer, and then you you lead them on the journey, right, like, down the down the path for

Greg Kowal:

the Yeah. And, that is actually what we also found out during testing that the like, when you test with people, they do typically 2 things. They either place themselves in narrative, so they're creating themselves and creating story with themselves. And the second thing is they try to create celebrities, create stories with celebrities and known people and some alternatives. So definitely, this is what people want to do.

Greg Kowal:

They are kind of focused on the characters that are meaningful to them.

Ryan Lazuka:

Awesome. And can they like, if you if you're doing one of these adventures, can you record it so you can play it again? Or you can replay what you what the adventure was?

Greg Kowal:

Yeah. Sure.

Ryan Lazuka:

That's cool. So you could, like, you could play the game, record it, and then upload it to YouTube or or TikTok or something like that.

Greg Kowal:

Yeah. Yeah. Exactly. So with with the story with the story game production, we also are looking at this vertical of because you have influencer. An influencer influencer or content, let's say, creator on YouTube, and they have their primary content that they produce.

Greg Kowal:

So there's, like, a long form video or or or whatever other content, a Twitch stream. And then you have the these different means to promote that content. So we are also looking at this vertical. Can we take, content this long form content production, that this is the main source of income for the creator, but then he could utilize Faber to create a playable experiences that are actually connected to that primary content so that people who experience this content will say, okay. I want to find out this narrative is connected to that video.

Greg Kowal:

Let me see that video. Yeah. So I don't know if this podcast is the only thing you do, but you could imagine that if you produce some some long form content, Faber could be means to promote that long long form content in a little bit different way.

Ryan Lazuka:

We have another I have another channel. I do, like, 3 news stories a week on anything AI related. They're they're usually, like, 5 to 7 minute videos. So yeah. So how would that how would that work?

Ryan Lazuka:

Like, you say you have a video about, you know, an the a SoRA AI news story, right, about SoRA, how SoRA was just released by OpenAI. How would, FABELER fit into that?

Greg Kowal:

There's difference in terms of audience and needs. Yeah. So when we are talking about this promoting long form format so let's say we are talking to the person that is releasing a physical book. Mhmm. So what they would like to create actually to promote the book is, storytelling that gives some elements of the of that book.

Greg Kowal:

So you have, like, let's say book is about, you know, how to be more confident. Yeah?

Ryan Lazuka:

Okay.

Greg Kowal:

And the book itself is a book, but you need to have some means to promote it. Yeah. And, general kind of the difference here is is a little bit different that person that created book not necessarily wants to create a games. So what they would prefer to do is to upload the context. So you upload, you know, the contents of a book.

Greg Kowal:

Okay. And then you want Faber to generate some playable experiences for you in the background, share our draft with you so you can just approve it and share it with you with with with the audience so they can play it. Yeah. And if the people like the content, like, the occasional content in the story Mhmm. They might buy your book because they like the the gameplay.

Greg Kowal:

Yeah. And it's a little bit different and, because people who may typically, if you make this primer if you are making podcasts or long form video about AI, you actually want to focus on that on production of that content. Yeah. So here we can see that in final vision, we can get favor to, like, the 0 effort game production where you will just upload your video. We'll do that transcript of the video.

Greg Kowal:

We'll get the images of the video. We will, get the assets. We'll generate the variance of that assets, try to generate some draft of the story for you. We just get it on the email and say, yeah, I like it. Let's approve it, and let's share it.

Greg Kowal:

Yeah. So it's like it's really zero effort. Effort. Yeah. So it depends on who you talk to are different needs because if you have a person that really wants to self express with Faber, they will want to have a control.

Greg Kowal:

If you have a person that actually focuses on something different, they want Faber to do, like, the 0 effort production so that I can just share it with others and promote my main things. But it's still you've produced something, this primary content that was shared with Faber as context so that he can create those story interactive stories for you.

Ryan Lazuka:

Well, awesome. Like, it's such a different way of thinking about things. It's hard to get your head around when you first hear it. You know? Like, a game based on content that you upload and the and the example you're going over, a long form, video on YouTube.

Ryan Lazuka:

So it's crazy. And and a lot of content creators are looking to promote their stuff. You know, they've got a lot of good content, you know, whether they have a long form video and they wanna create a short out of it or they can create a game out of it. You know, they've already done the hard work. So if you just upload a transcript, which are the transcripts are easily, created on YouTube anyways, you you just copy that and put it in a Faber.

Ryan Lazuka:

It'll create a game for you. That's pretty pretty exciting, you know, to be able to do something like that.

Greg Kowal:

It all depends on the content. Yeah. And the content that we focus on, first. Yeah. So, like, if you have, let's say, you have these Roblox creators And the Roblox creators, create just you know, they record their gameplay in Roblox and they tell story.

Greg Kowal:

So now what they can do is they can cut those video into shorts, share it as shorts. Although they also could do is upload the video to Fabler. We will do transcription or copy transcription and take, you know, particular scenes, generate variants of that scenes, and put it together as a game that they could use to promote this primary content that they have. And very important thing is that if, let's say, we talk about Roblox creators, they also have their distinct digital identity. So those games should should feature their characters, yeah, their, you know, their unique avatar because other players know them, in Roblox by the way they look, like, in Roblox.

Greg Kowal:

Yeah.

Huner Kallay:

So how did you come up with this gaming idea? Do you have a background in, like you just like video games and thought of it this way? And

Greg Kowal:

No. Like, we are we have interesting mix here at Fabler because, we actually combine 2 worlds. Yeah. So my background is, like, this classic IT, and, long story short is I've released one small game, very short game. That was, I think, in the year 2 I've made it after hours.

Greg Kowal:

It took me 1 year to make this game. This is example. Yeah. I was when after hours working on a very simple game. It's called Enigma Spy Adventure.

Greg Kowal:

I think it's depublished now. So, you know, yeah, like, if you Google it, you will find the content on that game. It was the we we call the developer but pixel games. And I worked with my friend. It was, like, over a year after hours, and that was, like, probably 20 minutes of gameplay or something.

Greg Kowal:

But that game generated, like, in 1st 2 months, like, 1,000,000 plus something installs. So it's got pretty popular. Apparently, I I'm still getting emails from people, will you release the second part of the game? So that's and that was like years ago. Yeah?

Greg Kowal:

And for the other games other games that we've released later, we, people also keep emailing us like are you gonna do remake and so on. So like if you make a good story and the the there are also some other really cool things about making good stories. Yeah. So when we released a game called, lucid dream adventure, There was a story about disabled girl on the wheelchair that stepped into world of dreams where she was trying to save her mother because, like, it's a little bit I'm not gonna spoil it, yeah, if you're gonna play it, maybe. But in general, the story this is like a little bit like Alice in the Wonderland, but a little bit changed theme.

Greg Kowal:

Yeah. So you have disabled girl that struggles with problems at home and she steps into world of dreams and then in the world of dreams, she can walk and she's trying to save her mother. In the pro in in the process, she learns about her father and so on and so on. Yeah. So and, we got hundreds of emails from people saying that this game has made change to their life or that they suffered loss, and this game was so cool because it helped them to deal with this.

Greg Kowal:

And this is like you're some you're you're making a story. It's just a story. But and for you, it's like like, if you are if if you work on that game for a year or 2, it just becomes really kind of like another day in the office and you can start getting detached to the story and so on. It's very hard for you to assess whether story is good or not anymore because you spend so much time on it. But then when you see that some people were touched by just a story because it the game is more mainly about the story and it kind of help them to deal with some of their problems is really cool.

Greg Kowal:

And I think that, this will be also, I think, the the the biggest success of Fable if you could get people to tell their stories in the way that they could touch others and and make their life now a little bit better in some way or to make them so so coming back to your question, it's kind of like we are combining 2 worlds because by maybe small coincidence, we we we've released a game after hours that was pretty successful by its own. We've invested 0 in marketing and got, like, millions of installs. Then the follow-up games were also pretty successful. We won multiple awards. We got awards from Google.

Greg Kowal:

Google actually showcased our games in London in, during, like, indie showcase event. So that was really cool. We have very good relation with Google actually because of that. And then we I always had this mindset of optimizing things, and I could never kind of understand why it takes so long to make a game. It sounds so simple, and even viewers consider it simple.

Greg Kowal:

Yeah. Like I I have to tell you that this is pretty kind of it's hard to deal with when you get an, player who played the game and the game lasted 8 hours. It took you 2 years to make game and then he tells you, like, why it's so short? You know? Can't you make it longer?

Greg Kowal:

How long it took you to make it? Like, 2 weeks or 3 weeks? Like, really, this is the outcome because people don't understand. Making games is super hard. And when you have, it's not just like in general, our team is super kind of effective in making content.

Greg Kowal:

We were always super effective. Yeah? But then when you look at some companies such as I was talking to the, VP from Disney Gaming, I think they were having a game released in Asia, story game, Disney story game. So he told me that they have team or and it's a story game so it's like a linear. Yeah?

Greg Kowal:

Yeah. So they had a team of 60 working on a content for a year they they were designing content for a game for a year and that that content was consumed in 7 days. And he what he told us is that story games are really hard because the better content you produce, the the quicker it gets done by players. Yeah. And that's what we've experienced.

Greg Kowal:

Actually, we suffered on on that those platforms because in in Lucid Dream Adverture, the total gameplay was, like, 8 hours

Ryan Lazuka:

Mhmm.

Greg Kowal:

And people were consuming it on day 0. So playing 8 hours straight. And then what you get is those platforms actually promote games that are good over a week. So people, like but then we see people are finishing in a day. So the better the better game you produce, this and that's the problem in story games, that the story is once story is over, the game is over.

Greg Kowal:

When you make a game where you have some, like, repetitive mechanics, you play levels, and you have, like, 10,000 levels and you keep playing, keep playing, you can play it for a week, 2 weeks. Once you are bored, you will put it to a, like, candy candy crush. Yeah. You can play it forever. Yeah.

Greg Kowal:

But then when you you try to create a story around the game to make it really long and supply players with content for months is just, like, impossible. Like, even for the company that has 60 people, you know, onboard. And the companies, like, CD Projekt, they are making games for, like, 7 years, and then it's 40 hours of gameplay. Yeah. So

Ryan Lazuka:

Yeah. It's crazy. Like, your grand theft auto. Like, how many people work on that game? You know?

Ryan Lazuka:

Like, it's they spend just godly amounts of money to to make something like that. That's that's a whole another level. What for these games that you created look. The games that you created, did you you just made them for fun? Did you try to sell them as well or how did that go?

Greg Kowal:

No. Like, we've monetized the games and we made, let's say, decent return. Maybe that was not a return, mainly because the monetization, in the games we've released was more of unlocking the full story. So when you if you make a proper game where you can keep monetizing and grinding a player, you know, for months months months, they can spend, like, 100 of dollars, we could not apply such monetization strategy to to to such game, but we've made a, made a return and all of that return was actually invested in what we make now. Yeah.

Greg Kowal:

We also have investors, like, Angel Investors that joined us, but, like, we believe that this project is the biggest thing so so far that we've made. Yeah. And all of everything that we've done so so far will be very small.

Ryan Lazuka:

Yeah. We wish you the best of luck. It sounds like a really awesome project. What what is like, if someone wants to create a game, will they be able to monetize by selling the game on fabler fabler, dotai eventually? Or what's the what's the revenue stream for for for the creators?

Greg Kowal:

Yeah. So, in terms of revenue, it like, definitely, we have to make it the way so that creator feels that he's a part of the of the system. Yeah? I think that, we have to figure out it with players what works for them, or creators, what eventually works works well with them. I think that the YouTube way of monetizing is something that should also work for Fable.

Greg Kowal:

This is our current idea. But there is also this idea of if we can create the community of creators who create together and someone, let's say, creates art assets that he wants to monetize, there can be this additional revenue streams where the economy works. Yes. So you are you are, let's say, artist. You've created some art.

Greg Kowal:

You publish that art on favor. Someone has took the art or purchased that art from you. So then we will actually go for more like a Roblox like approach where you have some, currency inside of the game that you can spend or either exchange for the real money. But it's still, too early really to tell on what we will decide. But Yeah.

Greg Kowal:

Definitely, we want to we have to make it work for everyone. Yeah. So creators needs to feel that they're kind of making return and not we are making return on what they produce. I think this will never work, and now I'm trying to monetize the content without the creators. Yeah.

Greg Kowal:

So it will be revenue shared with them.

Ryan Lazuka:

What does the, what does the pricing look like? Once creators are able to create games on your platform, is it gonna be a monthly subscription? Is it is there a free tier? How does that work in terms of pricing?

Greg Kowal:

Yeah. So the one of the things that we have done from the start was is to produce the system so that we have the free tier so people can use it for free. So, we are actually relying on our own, models. So all of the art models and llm models are open source implementations that we have trained, and we are running them on our own infrastructure. And thanks to that, we can control the cost of the generation.

Greg Kowal:

And thanks to that, we can actually offer a free tier because if we rely, let's say, on any third party service that, that provide you with the art generation or text generation, that will just get super expensive. So, kind of the core strategy of ours is to be able to offer it to individual creators for free. But then, obviously, if you are a heavy creator and you are making return, we are planning to give to have some paid subscription plan.

Huner Kallay:

So where are you on the road map right now? I know we've kind of talked about, like, the pieces that you're putting together from image creation to music to all of this stuff. What does it look like right now and what are the current projects you're working on right now and what are the where are all the pieces that have to be put together before this is all said and done in a complete project?

Greg Kowal:

So at the moment, we are in the bet, beta phase of the the storytelling module. We are also working with few companies who can get game game dev company, like a game development companies who they already one company who already started using Fable for the met story meta content production in their game. We can, we can also we also have a company that actually does not intend to use Fabler as the, output in their game, but what it enabled them to do is to mock up the idea of the story really fast. Yeah. So what you do is you can actually generate instead of, you know, let's do a storyboard.

Greg Kowal:

Let's talk to our writer. We'll create, you know, the script, and then you and a concept designer will create concepts and the artist what they've done was sing single person sat down and generated the idea of what the narrative should be, show that narrative with the voice overs, music, and everything to the director and the director says, yeah. Yeah. I like it or I don't like it. Yeah.

Greg Kowal:

So it just improve that process. And at the moment, we are 100% focused on, making this story creation experience excellent. And there are a number of features that we are implementing based on the conversation. Yeah. So first is being able to upload, your own kind of reference images and generate game characters that look alike, those reference images.

Greg Kowal:

They've asked us about ability to edit narrative by hand, and the model is not, possible. So they want to actually, you know, generate the narrative and be able to edit it, change this narrative. And there is other smaller features such as, we want to upload our own music, we want to change them, we want to configure specific voice overs for characters, etcetera. So this is the features we are working this already to storytelling module. Once we'll complete that part of work, we will next step will be the decision, gameplay that we'll be implementing here.

Huner Kallay:

So how are you dealing with, obviously, this global crisis of copyright concerns with music, voices, and things like that. What kind of tools are you using on the back end and how are you dealing with those concerns?

Greg Kowal:

So we, like, in general, all of the models that we use are trained by ourselves and we are using synthetic data to do this. Yeah. So in general, it's kind of iterational process. Yeah. So we started with the base open source model.

Greg Kowal:

The question is what happens to the base found foundational open source models whether they will be killed or not? At the moment, I think everything is going in favor of AI, but we'll see. And what we really do is we have this at the moment now it's large. We have very big datasets because you take it in iterations. So first, you generate, you know, like, a small dataset.

Greg Kowal:

You train the model. You see what the outputs are. You generate larger. And then then you do permutations of that datasets because that doesn't work and so on and so on. And after 6 months, you have pretty good dataset.

Greg Kowal:

You train the model again. You see, okay, it doesn't work and you learn. And there is multiple different practices that you can do to improve it, improve the output. Yeah. I think that initially people were saying that training models on synthetic datasets won't work, but I don't think that's the case.

Greg Kowal:

Yeah. You can you can actually make it work. We are making it work, And I believe in very, in specific use cases, what we do is better than than some of the the the model strength on the on the human human art in general. So this is this is this is a lot of work. Yeah.

Greg Kowal:

In terms of what will happen to this, you know, in general AI content copyrights, I don't know. We just, you know, we bet on the AI being, you know, free to use. If the regulators will take it into direction where they will actually prevent use of the, I don't know, synthetic data to train the models. Like, we're gonna face that reality. I think that favor is also not just about AI generated content.

Greg Kowal:

This is interesting kind of learning from the beta tests that people not necessarily have to generate things. It's more about the ease to create the the the narrative and where it places all of the things together and kind of AI acts as the placeholders and later I will replace it. Yeah. So artists that, let's say, hate AI, there is this massive movement of artists not liking AI, I think that they could actually use favor to create games with their own art and compete with the AI art. Yeah.

Greg Kowal:

And I think that's gonna be very unique proposition because you're saying this this this game is like you have mechanical watches which are now super expensive but they used to be normal thing, you know, like 80 years ago or 70 years ago. So I think that the AI art will be here but the human made art will be just very special and people will appreciate that it's human human made. Yeah.

Ryan Lazuka:

Yeah. I think that that like you alluded to the very beginning of podcast is there's gonna be you know, for movies, for example, the human element is not gonna be taken away, but it also AI is gonna be there as well. So it's gonna be a hybrid approach, and and it could be the same way with artists. Right? Like, the artist can use AI AI to help them create their art.

Ryan Lazuka:

Right? But if they don't embrace it, they might get they might get left behind.

Greg Kowal:

What I wanted to add here is that I think that making making AI art is not just prompt. Yeah. Actually, if you now will go to the tools that enable you I think that Mejourney gives you this impression because they create a UX over making AIR that it seems like you are just creating a prompt. But if you go to open source tools that enable you to produce art, the like, we are working with AI art 247 at Sabler, and we cannot follow all of the technologies that are being released. We have I I have, like, a document with 1,000 different releases, add ons that you can use with the stable diffusion models, and things.

Greg Kowal:

It's just impossible to to to catch up with all of the developments, and everything that is being released in some way will help you to make better generation, better outputs. And, like, if, let's say, today you will decide, oh, let's build the character's pipeline for a game, that's not a prompt. You actually gonna have to do, like, some sort of, you know, AI engineering. Like, it's different. It's not the same like in this large companies I used to work for where you have the data science team that was working.

Greg Kowal:

It's a little bit different kind of engineering now, but you still we have to, you know, figure out the tooling, test this, check, you know, different metrics of what sampler works best, then, use some, you know, a detailer that will help you with the details. If you do full post characters, the eyes will give bad. So you need to find solution to that and there is, you know, some things that will work right for the scenes, but the lighting will be bad, so you need to so it's not like you just prompt and generate art. It actually like, I have one company that, I'm helping with the in general getting, up to speed with AI. So I think it's already, like, 2 months of them working on the on the pipeline that they could implement in their game.

Greg Kowal:

That was not one day. The guy just didn't come, okay, let's be prompted. Okay. Problem solved. It's actually engineering.

Greg Kowal:

Yeah? And I think that our like, obviously, I understand why people are are not happy with AI because it's it's it's about to take their income away, But at the same time, they could actually learn all of the tools and I promise them if they will spend 247 learning how to, learn all of this those tools, they won't be able to keep up. It's not you don't you don't have enough time during the week to learn all of that. Yeah. There's so much of things that you gonna have to learn to be expert that can later join, I don't know, the CD project, and do AI art for them is not a is not a prompt.

Greg Kowal:

Yeah. So

Ryan Lazuka:

It's so hard to keep up with just the AI news in general. So the there's new tools and and features and news all all the time, so it's it's impossible. Like, we deal with that. You know, we write a simple newsletter, and it's hard to keep up with everything. So I can't imagine what it is for gaming, but I think there's you bring up a good point.

Ryan Lazuka:

Like, there's tools for editing, for example, where, you know, 5 years ago, it would be very hard to take the background out of an image if you wanna take if you wanna just put in put your face on an image and have the background gone or change the background. Well, AI makes that very easy easy now, but it's not a prompt like you said. You have to actually go into a tool, you know, select things that you want removed or say, I want the background removed, and it does that for you. So it's like the it seems like these tools for gaming as well are are supercharged in the editor, and the AI makes those things easier within the editor, but you're not writing a prop you're not just writing props. Create me a game and, you know, voila, it's done.

Greg Kowal:

I will tell you what my kind of take on that is. I think that games made just with AI will be good. Mhmm. But games that are made assisted by AI but they are human made will be great. Like, to like, even you can, like, in general, like, you you are in AI so there is this agent kind of approach to to to to to dealing with AI where you have multiple agents talking to each other.

Greg Kowal:

So I think that, currently, the thing is that you always need a human input to make it good. So you cannot, like we've experienced this initially initially when we are trying to do like, obviously, things are changing and maybe this year we think things will change. But at the moment, I like, even if you if you if you try to create with state of the art OpenAI architecture where you will try to delegate something for, you know, 1 month to AI and then have that AI do meaningful things, it won't Because you're gonna have to direct it somewhere. Yes. So you still have a human somewhere that directs the output or tells that he's not happy.

Greg Kowal:

And I think that in general, even with what we do is the best things made will be when actually someone has some vision and uses favor as something that multiplies, makes it 100 100 times faster for him, but it's still his vision. Yeah. Obviously, you know, I was not expecting Sohra to be so good. So maybe we are something will happen soon. But at the moment, I think people are needed and the content that is directed by people who have vision, ideas, that would be the best content.

Greg Kowal:

Yeah. Like, even, like, if you start GPT today to do a PowerPoint, you will see that you're gonna have do a lot of prompting before you will get the idea that you want out of it. It's not like, like, create a PowerPoint amazing, you know, PowerPoint for me. And then what you're gonna see is, okay. I don't like this.

Greg Kowal:

I don't like this. And you keep interacting and you spend, like, 1 hour iterating. And chat GPT is just assistant helping you, not making it for you.

Ryan Lazuka:

Sometimes it might take you longer to do all those edits for an hour than just create the PowerPoint in the first

Greg Kowal:

place, you know. So Yeah. I I was I was thinking the same. Like, last night I was working on the PowerPoint.

Ryan Lazuka:

And I like, another very important thing that, you know, I've come to realize what, by doing YouTube videos is, you know, you can have all the data in the world and you can have all these LLMs spinning out PowerPoints or you have SOR make videos for you and this the videos look amazing. You know, it's not released to the general public yet, but the examples that we've seen are are pretty mind blowing. But you still need a story behind that. Right? Most of those stories, the good stories out there come from human human beings.

Ryan Lazuka:

It's hard for from my experience to have chat gpt tell a very good story, that I could use for a video, for example. So it's it's you could I guess my point is you can have all the the best data and the output in the world remote from from AI, but you need a human touch to tell a story behind it. And I think

Greg Kowal:

I I I believe the same. Yeah. I believe the same. And and to be honest with you, also, I think that like, at least in FABLAR, from the creator's perspective creators here, because the viewer is different thing. We need to get there, you know, to explain what the viewers like, what the creators create.

Greg Kowal:

We we can see that the creator is engaged where he actually interacts with the system. So when initially like, there is this new kind of vertical of that influencer created something and he wants to use that as input to create something. But, again, he's someone creator is happy when he creates. Like, and the good creators are good at creating that's why they are successful. Yeah.

Greg Kowal:

And they want to create. They they need to, like I I I think that this is very unlikely that creator who hates to create content is successful. He doesn't like. So you need to have this vibe, you know. Yeah.

Greg Kowal:

I want to create this video. I want to get in the front of camera, get in the go outside, record. You need to have a passion for it. Yeah? So now there is so much passion passion somewhere happens, and that passion is being fed into a system.

Greg Kowal:

Yeah? So, and when we initially our initial idea was initial initial January 2023 was to create a system that creates games on on on their on its own. Yeah. So you, like, give a prompt and keeps iterating. There was this agent.

Greg Kowal:

But when we were doing the initial tests with people, they were just engaged. They were just coming back to that initial prompt. Yeah. They were not even checking the story that was generated, but they were seeing the first kind of outputs and they were hitting back and re prompting and then sent them and re prompting. So they were, like, so we've noticed was okay.

Greg Kowal:

So they are not very even interested about the story, but they are more interested how it reacts to their inputs. So they need there must be control over what they create here because this is what is fun for them. Yeah. So creator wants to create. If you take it away from him, he will probably not come to your system to create because there is no point for him to create.

Greg Kowal:

Right. Obviously, if you have created a long form video and that video is the input and then so you've spent time all over creating that video and you fed that video into Fable and Fable created the things for you, that is also your input. Yeah? But at that point, yeah. So I think that people want to interact with those systems.

Greg Kowal:

It's like, you know, calculator makes things easier for you, like spreadsheet.

Ryan Lazuka:

Yeah. If you are a good creator and you're passionate about what you do and you do have a story created by OpenAI, for example, on chat gpt or claw.ai or any of these models, You read them and you can tell the stories. As being a good creator, you can tell the stories are kinda crap. Right? And, like and then you go back.

Ryan Lazuka:

You have to reprompt it and say change this and change that. So I think it's it's almost like a it's a human thing. Right? Part of us wants to be as lazy as possible and have have to do all this work for us. But the bottom line is the stories that do the best are the ones that have the passion behind them.

Ryan Lazuka:

And if you don't have that, your content's not gonna do well, whether it's AI created or or

Greg Kowal:

Yeah. On

Ryan Lazuka:

your own, you know.

Greg Kowal:

I I 100% agree. And also, I I like to look at the like, when you deal with AI, I like to kind of use this reference of if you have 3 people in the team, the smartest guys the smartest guys in, let's say, in the USA are in the room.

Ryan Lazuka:

Okay.

Greg Kowal:

We like and you get them to work together so you have the smartest minds. They will also kind of start arguing, no. You are wrong. I'm right. You know, this is how we should do it.

Greg Kowal:

So, like, even if you create the best AI model that knows everything, it's still not good enough. Because if you put 3 smartest people in the room, they will still have different perspectives. So there's no, like, people look at AI and they say, this is a singular source of truth. Not like like it should know every like, it's right about everything, but world is not black and white, you know, that's the problem. Yeah.

Greg Kowal:

That you Yeah. Three things at the same time can be true And here is the problem, yeah, that you need someone to decide, okay, I out of these 10 best things created by AI, this is the best thing for me and for my content and that other things are not really good in my context, in my very specific environment and so on. Yeah. So I even believe if we have kind of, you know, AG, like, you know, like, super AI that can, you know, do everything, still people will be needed because if you put 3 smartest people in the room, there is still there is still room for, you know, discussions, decision making, figuring out what is right. Yes.

Ryan Lazuka:

And all those it seems like AI is not good at subtleties. Right? Like, in a movie, there might be something moving on the screen, very subtle in the background that makes a big part of the story. Well, AI might not be the best at doing that, you know, even in Sorum, something like that. So those subtleties make a big difference.

Greg Kowal:

I I was, like, looking when, Sora was released. I was with my, with my wife in the room and I showed her this. We're just discussing that if you take Tarantino movie, like, scenes from Tarantino movie, it's gonna be very hard to replicate this because there are such a small things that play a difference that replicating, that's gonna be super hard. You can still use it to you know, for some scenes, but there are, like, very, very small things. Like, if you analyze the Tarantino is just example, but if you analyze this movie, there are very, very, very small details that make it a great movie.

Greg Kowal:

Yeah. Very small details.

Ryan Lazuka:

And that's be very hard for AI to replicate. Them.

Greg Kowal:

Yeah. But but what I think, Zuora, will do is it will empower the small guys. I mean, like, these casual creators to create something meaningful that can probably get them later a capital to create a professional movies. Yeah. Very professional movies.

Greg Kowal:

Yeah. Because I think that AI really, what it does is it empowers the people that want to learn and can probably gives gives like, let's say you have a person in, whatever very poor country that doesn't have our resources to create such content, and that person learns the tools and learns how to use SoRA, but not just SoRA. It learns different tools, how to edit it, how to align it. That person spent 6 months on creating best movie possible with the tools. Mhmm.

Greg Kowal:

So this is something that is great. Yeah? And I think that AI really will definitely, shrink revenue of that larger larger, you know, movie producers, but we'll also empower the small small producers that are not are now not capable. Yeah. Because if you cannot afford the even, you know, camera to record the movie, you will never create it.

Greg Kowal:

And with Soera, like, this is this is my idea in general. Yeah. That you could actually empower this this this people that are now without any power. Yeah.

Ryan Lazuka:

And those people might have, you know, back to the passion, those people probably have way more passion than the big big Hollywood studios do. So they're you know, the end result, I think, is a win win for everyone. It's a win for the creators and also the people viewing the content because the content most likely will be more interesting because the creators that created them created the content are passionate about it. You know? Yeah.

Ryan Lazuka:

Yeah. Now they're unable to make it because they don't need the money per se this time.

Greg Kowal:

I'm I'm really looking forward, you know, to seeing, you know, first movie that you can like. Because at the moment