Hi everyone and welcome to another episode of Living on Blockchain. Today we are speaking to Brandon. He’s the CEO of ArenaX Labs.
Brandon previously led an ML research at an investment firm where he focused on reinforcement learning, deep learning, sequence prediction. He has also created algorithms for trading in the Web3 space. Now he’s focused more on the intersection of AI and gaming and educating the world about machine learning.
They are tokenizing this platform and really bringing AI as well as gaming cohesively together. This is a little different than the rest of the conversations that you know you’ve heard so far. I can’t wait for you guys to hear this.
Let’s deep dive right in. Hi Brandon, thank you so much for making the time to speak to us today. How are you doing? Doing great, thank you so much for having me.
Really excited about this. Yeah, likewise. I’m so glad you could make the time.
Can you give us like a brief overview of ArenaX Labs and its mission in the intersection of AI and gaming? Sure, sure. So ArenaX Labs is basically an AI gaming company. We have a few missions.
At the core, all of them revolve around democratizing AI. Specifically, the two prongs that we’re using to tackle this is AI education. So we think it’s very important to educate the world.
It’s basically about how machine learning works and the other is empowering researchers, which I’m sure we can dive into a bit later. But those are the two big prongs. So we’re all about using gaming as a medium to deliver our objective.
All right, awesome. So you know, that is a little about ArenaX. Can you tell us a little about your background and how you got into this space? I’ve read up a little about you and you know it mentioned that you’ve done some work in the Web3 space as well before moving on to AI and the gaming intersection of it.
Can you give us a little, tell us a little about that perhaps? Sure. I will say I actually started in machine learning before Web3, but I can walk you through that. So basically, before ArenaX Labs, I used to be a quant building algorithmic trading strategies using machine learning.
So I did that for since 2016. And then it was around 2020 where I really realized that you can combine the two technologies. So I was kind of passively into blockchain around like the 2017 time, you know, like the end of the last bull market.
I started really looking into it, but then it kind of like went away my interest for a while. And then 2020 rolled around or 2019 to 2020 rolled around and I really got interested again, obviously, as what happens when price goes up. But I started really looking into the technology and that’s when I realized that these things called NFTs were very powerful, but not powerful speculative instruments as most people were using them for, just that they can containerize anything that’s unique and has value.
And that’s when the light bulb kind of went off where I was building these machine learning algorithms that I thought at the time were unique and had value. And so I thought, how great would it be if I can tokenize a machine learning model? And I would be able to basically realize the value of it. And specifically, I thought this was important, because I knew that there was a problem with credentialization in the machine learning industry.
Specifically, a lot of like employers, basically not look at people that didn’t have a PhD from certain universities. And so it left a lot of people that actually had the skills, not able to find a job that was suitable for them. And so I thought, how great would it be if these people, if they’re actually good, if they can make these models, and they can sell it.
And that’s when the light bulb kind of went off and I started exploring like, what are different mechanisms I can use to like enable this? Because just because you build this doesn’t mean people will use it. And that’s where gaming started coming into play, because everyone or not everyone, but most people really like playing games in some form or another. And games are also great of abstracting away complexity, which leads to the second point of what we’re trying to achieve, which is educate the world about AI.
And so it was this unique thing where we can achieve both objectives, or we think we can achieve both objectives through this like fun game where people can own and sell their own AI. Anyway, I was talking for a while, but that’s kind of like how the progression happened. Right.
So it’s been quite a journey for you. Like, you know, you’ve been around for a while. And you know, you could you tell us a little about what led you to, you know, really create ArenaX? What was that one pivotal moment that motivated that transaction, this transition, basically, because, you know, you moved from leading ML research at an investment firm, and then starting something like this? Was there any aha moment in this journey of yours? Yeah, yeah, it was.
So back at the fund, I was one of like the more technical people. And so whenever they wanted to learn about something technical, they would ask me and even though I was focused on machine learning research, it didn’t stop them from asking me about blockchain stuff. So I was like, Okay, I guess I have to learn about this blockchain stuff, like, more at a foundational level.
Like, so I would like, code up my own blockchains to really understand how they work. And that’s really when it happened when I was programming my own, like NFT from scratch. And I was like, wait a second, wait a second, these things are just pointers to images.
And they don’t have much utility, from my perspective. And so I was like, that’s when the aha moment was like, I can embed real utility in non fungible tokens. And I remember I went to my my boss, and I was like, I have this wild idea.
And my boss, then is now my co founder of arena x labs way. And so, yeah, he obviously thought it was a great idea. And yeah, that was basically the moment and then while still working at the investment fund, I was building out the prototype after work every single day.
I used all my vacation days just to take it off to work on this project. And then eventually I was like, Okay, I think we have something. And then at that point, I kind of left and started it.
Wow, okay. Well, that’s very, very interesting. So how do you see the role of AI evolving in the gaming industry? And what unique perspectives as arena x labs brings to the space? Yeah, that’s a great question.
So from what I see now, of course, there is probably a lot happening behind closed doors that I cannot see. But what I see is most people are using generative AI. In two respects, one to speed up game development, right to generate assets that they can either quickly prototype with, which is really cool, or to actually put in game, of course, they probably just like whip them up using the generative model, and then they make some tweaks to it.
And they put that in game. The other approach is increased personalization through these like large language models that each it basically allows each NPC to be unique to you and to evolve with your game experience uniquely. So I think those two are what most people are applying it to because those are the easiest to given the open models that we have right now.
Right. Where we fit in and where we provide a unique twist is we are doing neither of these. Rather, what we’re doing is we’re allowing users to train their own models.
So it’s almost like hyper personalization where you are involved not only with what your model ends up learning, but how it learns and you are there the whole process to actually teach it what to do and how to learn if that makes sense. So like if you imagine something like a chat GPT, the model is created. And even though there might be mechanisms to learn over time, you’re not involved in that process, right? At least not under the hood, right? Like you’re providing prompts and stuff and saying this is good, this is bad to help it tweak the model.
But that’s a different level versus what we’re providing where it’s like, no, you can actually tell the model how fast you want it to learn, which data you want to remove from the data set, all this stuff. And we gamify everything. So it’s not too burdensome on like new users that have never experienced machine learning before.
Does that make sense? Yeah, I think I know. So basically, your platform can be like a good gateway for people who are trying to build in the space of AI, like you, you know, building games in the space of AI. And you would encourage developers to perhaps with the right set of tools to build in a scalable way.
Does that sound correct? Have I grabbed the gist? Almost. So I would say I agree with the gateway part, but not just for building tools. It’s just, it’s the gateway for everyone that wants to dive deeper into machine learning.
And the way we see it is we would love to be the place that provides the full life cycle of someone starting their machine learning journey to someone going really deep into machine learning. And so AI arena, the game is the first stepping stone for people who have never seen machine learning before. And they started learning about these concepts like learning rate, epochs, batch size through a gamified experience.
And then later this year, we are going to launch a research platform where hopefully some of them will want to take the next step and actually program their own machine learning algorithms. And of course, we will provide templates, but that will essentially be a competition specifically for developers and engineers. And our hope is that we are able to convert some people through the game, the no code environment into this coding environment.
Does that make sense? Yeah, absolutely. I think that’s a wonderful initiative. You mentioned that basically the focus is on educating the world about ML.
And you know, you are taking a different approach perhaps to this particular statement, right? So can you tell us a little more, like you mentioned this one, a hackathon sort of a thing that you guys are going to be doing, right? If I can call it a hackathon, but what other initiatives are in place to make ML more accessible? Because again, for the layperson, it seems very, very esoteric and something very niche, right? So yeah, I mean, I wouldn’t say it’s a hackathon. Maybe it’s like a perpetual hackathon. It’s like yeah, like a forever competition and people can just submit models and see how they perform.
But I would say we’re trying to make it not esoteric. And I think for many people that hear about the concept, they’re like, okay, that’s niche. Like how many people really will play that? And like, will they actually learn anything? Like how much can you learn from playing a game, right? And I think most people would be surprised.
And there’s like a lot of anecdotes that I have. One of them is of like many, many people just learning about machine learning just through playing the game and have no machine learning prior experience. But I think one thing to understand, it’s a very powerful thing to gamify something.
Because, and I say this all the time, but people learn very complex gaming mechanics just to get a competitive edge in a game. And you’d be surprised, like things that are more complicated than machine learning training, but they learn it because they want to be the best at that game. And so by reframing machine learning training as a game, people don’t mind learning about this concept of a learning rate.
And now they quickly realize through our interface, like, oh, if I increase this, I can immediately see it has an impact in how fast the machine learning algorithm learns. And now they associate that in their mind. And that’s like a direct thing.
But there’s even indirect stuff that people pick up where it’s like, if I show the model more instances of something in this situation versus another one, it starts to skew how it behaves. And it more focuses on that situation, right? And people start intuiting this stuff about data collection, about like cleaning the data, all this stuff just by playing the game, which is like insane. And another anecdote that honestly, we see this over and over again, is that when people don’t, like if they haven’t played the game yet, and they hear the concept, a lot of people’s like, that does not sound fun.
Why would I want to spend my free time learning about machine learning? And I think, like, there’s like, we see this constantly in the data, because we collect all the data. But after a certain point of them playing, and once it clicks with them, like, they’re like, extremely, extremely into the game. It’s like this, like, weird, like, complete, like, there’s a certain stickiness to it, right? So you become, you know, become a part of the process.
And then you know, you want to continue perhaps becoming better. Exactly, exactly. And the thing is, it’s a never ending thing, because you can always make your AI better.
And so once people realize, oh, I did this, and it learned how to do it. And I can see it doing it in the match. It’s like, it’s, it’s like a dopamine rush, like, like none other.
And then you continually want to say, Okay, now I can do this in this situation, I can do that. And you just spend hours tweaking this thing. And before you know it, like, 12 hours are gone.
And so, yeah, and in the process, people are learning about machine learning, which is awesome. Yeah, I think that is a beautiful approach to perhaps increase education around ML. I think a good example to give would be, which is obviously not your competitor, but you know, what Duolingo kind of did with languages, I think that is some, you know, it can be a good analogy, you know, as to what you’re trying to do with ML.
No, I totally agree. And I really admire what, what Duolingo has done. I personally use it.
So yeah, it is. So as of now, like if somebody wanted to get involved in in the process of, you know, learning and moving forward, using your platform, can you tell us for our listeners, what would be like, what would be the flow like? How can they get involved? Sure. Right now, like, historically, we’ve had it like gated through a closed beta.
So they would have to kind of join our discord. And then there is a process in discord basically to get approved to test out the game. We are going to open up the game very soon, actually, to everybody for an open beta.
So we’re really excited about that. So at that point, the only thing that people will need to do is just go basically to the site, and just log in, and they should be good to go. And you can log in with like, whatever credentials, like we offer like a few different ones, like a Google account, Twitter, discord, or, or metamask, for example.
So yeah, they’ll just be able to log in and start playing. Awesome. So when do you expect to roll out this particular beta for everyone? Is that a milestone that is going to happen in perhaps the next six months? Oh, yeah, it’s likely next month.
Yeah, yeah, very, very exciting. We have a, yeah, likely going to be next month, we’re going to be hosting a big competition on chain. So yeah, definitely look out for that.
But we’re not just going to open it up for web three people, we’re going to be opening up for everybody, whether it’s web two or web three, to be able to, to come in and play the game. Although there are two kind of different versions of the game in terms of like, like, for example, the web two version obviously doesn’t have anything to do with like staking and earning rewards. But it’s the same underlying game where you can still learn about machine learning by playing.
Right? Okay. Can you tell us a little more about the web three tokenizing aspect, which is a part of the game? Yeah, absolutely. Absolutely.
So the core thing is that we are tokenizing these characters, which are powered by machine learning algorithms. And so essentially, and this is actually a really, really cool aspect about it, at least I think so. But the fact that you can, like these things are mutable, but they’re only mutable by you training them, right? So you can imagine when you first get at one of these, one of these AI fighters, it doesn’t know how to do anything, right? And you are able to basically transfer your skills onto this thing, right? Because it’s learning from you.
So the more and more you’re able to teach it something, theoretically, the more valuable it should become, at least to you, right? Like it’s starting to do stuff that it previously didn’t do. And this is what I was talking about before, where you’re actually providing utility into this tokenized asset, which historically did not exist for like, JPEG NFTs, you know what I mean? And so now it’s like, okay, I have this thing that’s tokenized, and I can keep kind of like changing it. And theoretically, increasing its value through me imparting my own skills on it and training it, which I thought was like, very, very fascinating.
Now, that’s the NFT side of things. Of course, we have also a token to administer reward, or this is an in-game token to basically reward users that are playing the game, but also use in various aspects of our game, like to buy items and stuff like that. I hope that answers your question.
Yeah, it does. This is very interesting. Like you said, this is a true intersection of, you know, Web3 AI, as well as gaming, it seems.
I can’t wait for, you know, to try it, basically. And while you were giving me the answer of join the squad already, I would love to try it out. This is super interesting, what you guys are building.
I appreciate it. I appreciate it. I think you’ll enjoy it.
Yeah, I do think so. And this is very unique, what you guys are building. Because I have, I interview and talk to a lot of founders, and I haven’t come across anybody who’s working perhaps in a similar way on a platform that is so cohesively bringing these three aspects together.
So, more power to you guys. I appreciate it. I think on that point, it’s certainly like a good thing, but it’s also a challenge when no one else is doing this, because it’s not well understood, you know.
And when it’s not well understood, people don’t really know what to think about it, you know what I mean? And so, it’s some of the communication about what we’re building is challenging at points, because people are like, what are you talking about? Like, that doesn’t make sense. But I think once there is one success case, I think a lot of other people will start doing it, and it’ll start making more sense to the general public, sorry. Yeah, I think that, you know, you’ve kind of put it and you’ve captured these two sectors in one sentence, that people don’t really fully understand perhaps AI and Web3 that well.
And that can be intimidating to them. And at times, obviously, it creates a certain amount of indifference as well. But now, because of noise, it’s hard to be indifferent.
And it’s better to perhaps get involved. Totally agree. Yeah.
So, you know, you guys aim to merge universal language of play with cutting edge technology. And that is what you’re doing with ArenaX. So, can you give me perhaps an example of how this fusion is manifested in your games and the platform itself? Yeah, I think, basically, just like what we’re building with AI Arena, right? Like, I think I mentioned before, since we want to educate the world about AI, we needed something that was not intimidating.
And we thought that was games. And so we’re merging, like you mentioned, like the power of games, because they are excellent of abstracting away complexity. Right.
And so we’re merging games with machine learning. And so that’s the technology portion, right? Like we’re merging these two in a way to provide this education for people. Yeah, that’s very exciting.
You’ve simplified it, you’ve gamified it, and you know, you’re putting it forward. So I think, I understand why you have such a raging community as well. But, you know, as somebody, you know, if somebody who is an entrepreneur, and who’s building in perhaps a similar space with gaming AI, it’s usually very difficult to, you know, because it’s a niche to really get a highly engaged community.
Without right now, financial incentivization is all the rage, at least it is during the bull run. How do you get a community of people who really care about this product? And what what is what is the kind of been your go to market strategy in getting an engaged community so far? Yeah, so first thing I’d have to say is that my co founder way is the one leading the charge here. So he has been doing a phenomenal job on this.
But and everyone on his team as well. Well, one thing I’ll say is we were kind of fortunate that we that we essentially started the process near the end of the last bull market. And so obviously people like it’s it’s no secret people like making money, right.
And so at that time, we announced our seed round with paradigm and framework. And so a lot of people came into our community, in anticipation of something, I don’t know, they expected some payoff or something for participating in our discord. Obviously, we do things very, very different than basically the rest of web three, we don’t do a lot of those, like, giveaways, like we don’t just give people rewards just for joining and stuff like that.
We do it very differently, like everything is based on meritocracy in our community. But nonetheless, it attracted a lot of people into our discord at that time. And luckily, lucky for us, this might sound weird, but we went into a bear market after that.
And a lot of our core community stuck with us through the bear market. And so they, because they played our game, and they really, they really loved it. And they just happened to stick with us throughout the whole time.
And now they are becoming basically, like the champions of our of our game and like really the voice to get more people in. And yeah, a large part is, is due to that, honestly, that we were building through the bear market, and we have a core set of community members that stuck with us and advocating for a game that has done wonders. But that’s not to say like, none of the other stuff is also important.
I think what what way is doing on the business development front and the marketing front is this phenomenal and a massive reason why we have the community that that we have now. Right? It’s brilliant. Absolutely.
So while you know, the potential of all of this is very, very exciting, adopting web three principles in the at the intersection of AI and gaming comes with its own set of challenges. What sort of obstacles have you faced so far? And what what kind of obstacles you anticipate, you know, in the widespread adoption of this, of this particular space? And how are you addressing it, perhaps? Yeah, the I want to build on something I previously said about it becoming a challenge that people don’t really understand what we’re building. And, like, specifically, like, so it’s one thing to understand what we’re building.
And let’s just say we get over that hurdle, like we’re, we somehow are able to communicate very well, like, this is what we’re building. And people are like, Okay, I get it. The next thing is making the game approachable.
And historically, it wasn’t, you needed a really high level of skill to play the game to a decent competency. Because it just, there’s like a lot of stuff in machine learning, that’s very difficult to overcome. Like, there’s this concept of catastrophic forgetting where you teach it one thing, and then the next training session, you teach it something else, but it forgets the previous thing, right.
And so these were like real machine learning challenges that gamers had to figure out how to solve, basically, which is not easy. And so the biggest challenge is the onboarding. Because like I mentioned before, once we saw that people are able to teach their AI something successfully, that’s where it gets very sticky.
And so how do we get from the point of them trying the game to that sticky point, and historically, there was a massive gap there. And so we had to feel like we had to figure out, okay, how do we close this gap, make it make the onboarding as smooth as possible and make it just like, almost like a smooth transition from onboarding to testing, to getting the basics to getting really good at the game. And so we’ve done a series of things to hopefully close this gap.
One of the big things we did was we made a simplified model that was able to learn more quickly. But it was able to learn a simpler subset of things in the game. That certainly helped a lot to get people that historically struggled to train to getting really good at the game.
But that didn’t solve everything, because still a lot of people, even from starting the game to that point, there was still a gap, right? So maybe that closed it halfway. And so we recently rolled out an in-game tutorial to basically teach them about the concept of training a fighter, right? And so we rolled it out, at least for some testing. It seems to be going well.
We’ll see when we roll it out to a lot of people in the open beta. But that has certainly been the biggest challenge, because this type of game is so new, right? No one really gamifies machine learning training. And so not only is the communication difficult, but also people playing the game was difficult historically, because they didn’t really understand these mechanics.
But we’re hoping to make it more mainstream going forward. Yeah, I think, as you mentioned, that one of the biggest challenges that you guys perhaps face is it would be difficult for a layperson to understand exactly what you’re building. And because you already are facing this challenge, and you’ll foresee this in the future as well, are you taking up specific content strategies as well to perhaps simplify what you guys are doing and creating a lot of content around it? Yeah, we’re trying a lot of different things.
So there was a series that I used to do on Fridays. It was called Get Good with Guppy. Guppy is my nickname online.
And basically, I would spend an hour live streaming in our Discord about how to do certain things in the game. And that worked great for core community members that already knew how to play the game, but they wanted to learn how to refine their strategy and do certain things. I really love those, because I would also get to talk about math and machine learning optimization in it.
So I would show them how to do something, and I would explain why it works. But I don’t think that is great for the general public. I don’t think the public cares about calculus and how it relates to games.
And so we’re trying different things to make the onboarding easier. We made different introductory videos, making short form content. But right now, it’s just a matter of trying different things and seeing what sticks.
I don’t think we have the number one solution yet, but we’re certainly trying things, and hopefully, we’ll find something that works great. Right. Yeah, it’s about iteration, I think.
Ultimately, even if you find something that does stick, after a while, it’ll change, and you’ll have to be dynamic with your approach. Exactly. Yeah.
These particular sectors and niches, they’re very dynamic in themselves. There’s something new happening every day, and you will have to be right in the eye of the storm to make sure that you guys are able to reach the requisite user base that you are potentially looking at. Absolutely.
Absolutely. Totally agree. So now moving a little macro and talking about the space in general, you’ve been a part of this sector for some time.
Are there any platforms that are to do with Web3 gaming, which are utilizing perhaps AI in the gaming space? Are there any other apps that you feel are really doing a wonderful job, and you want to take perhaps a few pages from their book? Not really. I’m not saying that just to say we don’t have competitors, because I think we do have a lot of competitors, but not in the traditional sense of they’re doing what we’re doing. It’s more competitors for mindshare.
But I think there’s certain aspects of different platforms, not necessarily like Web3 gaming or Web3 AI, but just more so in the Web2 space, to be honest, that we’re taking inspiration from. Obviously, our core game is a platform fighting game, so we’re taking a lot of inspiration from most of the platform fighters, but more broadly on the AI side, specifically, actually, Numeri was a huge inspiration for us. I don’t know if you’re, are you familiar with Numeri by any chance? No, you would have to enlighten me.
Okay. So Numeri is basically a data science competition. They administer their own token, NMR, and basically what they’re doing is they’re opening up a competition for data scientists to predict the stock market.
And then they kind of take these predictions and aggregate them and trade them in their own fund. But it was the concept of machine learning researchers and engineers competing against each other and earning this token, which is largely where I drew inspiration from. It’s essentially we’re applying it to a game, but that’s where I drew inspiration from for machine learning algorithms competing against each other.
Because at the end of the day, if someone were to say, my machine learning algorithm is more valuable than this other one, we need some way to administer this comparison and competitions are excellent way to do this. And so, yeah, we took a lot of inspiration from what Numeri is doing, although they didn’t apply it to gaming. They are sort of this intersection between machine learning competition and Web3.
Wow. Very, very exciting. Very interesting as well.
Brandon, I think I’ve completely lost track of time because what you’re building is so interesting and it comes with a lot of new concepts that I had to assimilate over this particular recording. So I completely lost track of time. It’s already a little over 40 minutes that we’ve been talking now.
I would love to, though, ask you two last questions before we wrap this up. One being, what is the kind of advice that you would like to give to or perhaps point folks in the right direction who are trying to perhaps get into AI and gaming? If not like advice, but if there are resources that you would recommend that they can look at? Yeah. So I think it depends how deep they want to go.
And are you specifically asking about people that want to use AI for building games or just in general AI and gaming? I think a little bit of both. It’s more interesting that if somebody is using AI for building games, then yeah, those resources would be more niche. But even just folks who are looking to add AI to the gaming platform that they have, are there good resources that you would recommend? Obviously, your platform being one, but right now it’s not open for public yet.
Yeah. Yeah. So I think I’ll answer the question more from the sense of people wanting to get into AI to aid in game development.
I think the low-hanging fruit type of advice I would give, which is not really, I don’t think adds value because everyone would give this, is basically just learn how to use these tools for prompting or just learn how to use like an API to connect to like an LLM that you can use in your app. But my advice is going to probably differ from that. So I am a massive advocate for learning how to do things from scratch.
And this is the advice I give everyone that does machine learning and they work for me, both now and previously when we were at the fund. But for example, I don’t even allow people to use a machine learning library unless they first show me that they can program the entire machine learning algorithm from scratch. And the reason for that is because you will never understand why something is not working unless you understand from its core, how it works.
And I think that’s very important for applying machine learning to gaming as well, because I can’t even tell you the amount of things we had to overcome just to even know how to apply ML to our specific gaming environment. It’s not like you just take a neural network, plug it into the game and it works. It will be a disaster.
And so you have to understand a lot of the idiosyncrasies about how the machine learning algorithm not only performs like inference, but how the training actually happens. Why is this thing called catastrophic forgetting happening? Why is my policy so skewed when I’m just playing the game normally because I don’t understand how basically human dexterity works and how that affects data collection, right? So there’s all these things that compound on each other. And if you don’t really understand the machine learning algorithm to its core, you will never be able to solve them.
And so my advice would be, and probably a lot of people don’t like hearing this because it’s like 10 times more work, but it’s to really go from the fundamentals, understand how the machine learning algorithm works at its core, and then you can start applying it. But that’s only if you want to do something novel, right? If you just want to apply open AIs like chat GPT to your game, you don’t need to do that. But if you want to truly do something novel, that’s certainly my recommendation.
Yeah, I think that those are very foundational recommendations that kind of go across all sectors, that if you really want to build something innovative in any space, then you really need to understand how the ins and outs of that particular space work. Absolutely. Right? You can’t really be doing just a simple copy paste.
That will not get you the results. And if you want to build something, you know, truly visionary in nature, then you have to get your hands dirty. There is no way around it.
Absolutely. Absolutely. So Brandon, now moving on to perhaps the last question, and this is a bit of a philosophical question.
I would love to get an answer or perspective from you. How do you balance out you know, the two aspects like creativity, human creativity, as well as the AI and ML aspect? While, you know, it has been utilized in this ever evolving gaming space, I understand that, you know, obviously, you can train these models. And that is essentially what you’re trying to do.
But there has to be a balance. Nevertheless, I think, what is your perspective on that? So I just want to make sure I understand the question fully. So by that, you mean, in this like, ever evolving, like, landscape where like, people are using AI more and more to create things? Are you saying human creativity starts to become less and less important? Because we rely on AI? Yeah.
And like, how do you balance it out? Because, you know, we are enabling humans to actively shape the future of AI together, right? That is what we are doing, at what you are building. So how do you strike a balance between human creativity and AI technology, say in your projects? And what is your perspective here? Yeah, so my view of creativity might differ from other people’s view of creativity. So I think personally, in our game, we are enabling a lot of creativity.
Let me expand on this. So I’ll try not to get very, very technical. But like, you can imagine, like, you drop an AI, like, in like, on a new planet, right? And let’s just say there’s like, like a treasure somewhere, and it has to find that treasure.
And so it has to start exploring to find that treasure. That’s what kind of machine learning training is, right? Where you have this thing that we call randomly initialized, so it doesn’t know how to do anything, you randomly drop it somewhere. And you have this treasure, which is the optimum, we call it the global optima.
And it has to find that. And this global optima is the place where if your model gets there, it learns exactly what you want it to learn, right? And it’s never, it’s straightforward path to get there. Because in machine learning, specifically neural network training, there’s this thing that we call it the lost landscape, it’s very non convex, which basically means there’s all these bumps, and these troughs, these hills.
And so your AI has to navigate through all this to get there. And a lot of times, it can be more an art than a science in order to teach your AI basically how to get there. And so the creativity in machine learning training comes into play where it’s like, okay, I’m in this situation now, what can I come up with, right? To enable my AI to navigate through this lost landscape to get to this global optima, which is the treasure, right? And so what you see in our game is people coming up with these really novel training strategies, whether it’s how they collect data, how they mix various things over sampling, and how they have changed the hyper parameters.
And you start getting really creative ways to come at the problem to teach your AI how to do something. I think a lot of people that aren’t in machine learning training just think it’s this black box, which kind of it is, but they think it’s just you let it run, and it does its thing. But there’s actually a lot of stuff involved behind the scenes.
And so what we’re hoping to do with our game is show people it’s actually really interesting, and you can try all these different things to teach your AI how to do something. It’s not just a straight path to get there. And so I personally think that’s where a lot of creativity comes in.
Of course, I also think there’s a lot of creativity in coming up with loss functions, and I think there’s a lot of creativity in mathematics as well. So a lot of people probably look at that and be like, this guy’s out of his mind, because most people associate creativity with more of the arts, right? But anyway, I know I was rambling for a long time. But yeah, I think the concept of creativity could change over time where it more so turns into how do I work with this AI machine learning algorithm, whatever, to achieve the objective that I want to achieve, right? Rather than it’s only me trying to do this, it’s now me and the AI trying to do this.
Does that make sense? Yeah, absolutely. I think what you’re trying to say is that AI is going to only complement perhaps human creativity and help, you know, give you another layer, perhaps, in the way you are approaching a problem statement. Yeah, yeah, I think for most people, yes.
But yeah, exactly. Right. Wonderful.
Thank you so much, Brandon, for taking out the time to speak to me. Before we wrap this up, do you have any last thoughts about Web3 AI gaming as sectors and, you know, anything for the entrepreneurs who might be building in this space? Any advice? So, obviously, on those sectors, and specifically the intersection of them, I’m obviously very optimistic about. Otherwise, we wouldn’t be building in these.
Maybe we would, if I just like pain. But no, I’m certainly very optimistic about all these sectors. And specifically for people building in these, any advice I would just say, like, it might be difficult at times, but certainly try to ignore the noise.
I think it’s very easy to get caught up in, especially like during like the bear market. And when you’re on Twitter, people are putting out FUD, it’s just important just to ignore it, right? Tune it out, and just build, just build. And eventually, people will come around when you have something to show for it.
I think that’s just the best advice I can give is just keep building. And yeah, ignore a lot of the garbage that’s out there on social media. Yeah, I think that’s very, you know, wonderful advice.
And that’s the way to go about it. Just keep building, keep your head down, keep building, build something substantial and market it well. The traction will come, the users will come, and the validation will come as well.
But it’s not going to happen unless, you know, you get your hands dirty and in the space, take some action, basically. Absolutely, absolutely. Wonderful.
Thank you so much, Brandon, for taking the time to speak to us today. This has been a wonderful conversation. What you guys are building is super fascinating.
I can’t wait to try it out. And once again, really grateful that you could make the time. Thank you.
It’s been a pleasure. Thanks so much for having me.