How AI is Making Games Uniquely Yours (Ft. Marty Burgess, Lightning Forge Games)

0:01 Georgie Healy: Founders scale faster on Deel. Set up payroll for any country in minutes, hire anyone anywhere, get visas handled fast, and get back to building. Visit deel.com/dayone. That's D-E-E-L.com/dayone.

0:24 Marty Burgess: What actually happens if you're playing a game and we actually train an AI on exactly what you would do in a certain situation. And then let's say your internet drops out, like what happened to me, like, what, 20 minutes ago?

0:37 Georgie Healy: Mm-hmm.

0:37 Marty Burgess: Yay, NBN. And we can just pop an AI agent in place of you while we wait for you to reconnect.

0:44 Georgie Healy: Wow.

0:44 Marty Burgess: And your teammates and even your opponents don't notice the difference.

0:50 Georgie Healy: Hello and welcome to In the Blink of AI, where I talk to the brightest AI startups and innovators Each week, I'm Georgie Healy, and this week I am speaking to Marty Burgess, CEO and co-founder of AI startup Lightning Forge Games. LFG utilize the latest and greatest technologies across the blockchain and AI space to create new innovative gaming experiences. Marty has over 4,000 loyal followers on X, He's recorded just before hopping on a plane to showcase at the gaming industry's premier professional event in Silicon Valley. Say that 5 times quickly. But he's no snob. He is just as thrilled if your weapon of choice is Candy Crush or his grandmother's fave Sudoku. On the show, we talk about how AI can enhance the gaming experience and make it hyper-personalized for each individual player. Yeah. As well as where AI agents can best be utilized. I challenge you not to smile as you listen to this episode. Let's jump into it. Marty, hello. Thank you so much for joining in the Blink of AI. I'm so pumped for this chat. Let's start with just a quick explainer on what LFG is and what is your role there.

2:14 Marty Burgess: Yeah, absolutely. So I'm Marty. I'm the CEO and one of the co-founders at Lightning Forge Games. For fun, we call ourselves LFG. We call ourselves a blockchain and AI-powered gaming studio. So we seek to use the innovations and the incredible opportunities we see evolving in both the blockchain space and especially the AI space and use those to disrupt gaming as we know it.

2:37 Georgie Healy: Amazing. Look, your passion for gaming is so infectious, even as like a traditionally not a gamer myself, just accidentally grew up in a household of all women that didn't play games, but I want to know what inspired you. What was your favorite game growing up that got you on this journey?

2:56 Marty Burgess: That's a really awesome question. I've been an avid lifelong gamer. Probably the biggest thing for myself, so growing up, didn't have access to consoles and stuff like that. I was always a PC gamer. And when I was probably about 6 or 7, I first saw a game called Sid Meier's Pirates!

3:11 Georgie Healy: Ooh!

3:12 Marty Burgess: I've actually had the opportunity to shake Sid Meier's hand since then and thank him for this, because this is literally why I'm here today. At the time, like this is an era of like Super Mario Brothers or very simple linear, like just, you know, very the early era.

3:27 Georgie Healy: Sonic the Hedgehog, is that the same kind of? Yeah.

3:29 Marty Burgess: Yeah, okay. Or maybe even like, that's probably a bit beyond the sort of timeframe.

3:33 Georgie Healy: Really?

3:33 Marty Burgess: Yeah, yeah. And Sydney's Pirates, you play a pirate, you're on a ship, you can sail the Caribbean, you can do whatever it is you want to do in this game. You can literally just be a trader and just sail back and forth finding the best prices for goods and salts and things like that. You can take the pirate route and blow up Spanish galleons and look for the treasure fleet. You can literally— it was this beautiful— we now call them sandbox experiences, but it was this open-world sandbox where you literally just felt like you could do whatever it is you wanted to do in this game. And it literally just blew my mind because I'd never imagined an experience like this. In an era where games were very— like, think Mario, for example. You start at one end of the level, you run to the other end of the level, and that's it. And you can play Mario 100 times, and 1-1, that level plays exactly the same every single time, to the point where we now have speedrunners who have literally perfected that level. To the best of our knowledge, there is no way to finish that level faster than speedrunners are able to do consistently now. Like, that— it is a solved puzzle, to put it that way. And then to see this other game where it's just 'Every single time I play, something different happens. I can't plan for what's going to happen. I've got to adapt on the fly.' It just, like I said, I've had a lifelong obsession with gaming from that moment. And yeah, obviously feel very, very privileged and very fortunate to have the opportunity to work in the gaming space as well, and hopefully get to share a little bit of that joy that I feel when I see these incredible experiences with other people out there in the world.

5:11 Georgie Healy: Wow, that's a perfect lead to my next question. What kind of experiences do you think you would like to create through AI?

5:21 Marty Burgess: So AI is, I mean, obviously it's a huge focal point for LFG in what we're choosing to do. A lot of the studios right now are using AI in various ways. You go back a year ago to GDC where the idea for founding LFG first started. One of the frustrations I had is I would talk to studios and they're like, oh yeah, we're an AI studio because we use ChatGPT, or we know what Midjourney is, clearly we're an AI studio. It's like, no, no, no. What actually happens if you take AI and you bake it into the core of the experience and you build something that couldn't be made without AI? There's nothing wrong with using AI tools and technologies to assist the development process. That's great. It's bringing production costs down. It's making life easier for developers, artists, all that sort of stuff. It's great to do that. It's something the gaming industry desperately needs right now. But what we're really trying to do is go, right, What actually happens if you're playing a game and we actually train an AI on exactly what you would do in a certain situation? And then let's say your internet drops out, like what happened to me, like, what, 20 minutes ago? Because yay, NBN. And we can just pop an AI agent in place of you while we wait for you to reconnect.

6:31 Georgie Healy: Wow.

6:32 Marty Burgess: And your teammates and even your opponents don't notice the difference. What happens if we take that same AI agent and maybe you're offline for, you know, you're asleep, you're at work, something like that, and we let you play, your AI agent play with your friends, we can tell people that, you know, we're doing this. So, you know, you and I can go play, you know, a game together when I'm not available. Or perhaps you get to play with one of your favorite influencers, or you get to play with, you know—

6:58 Georgie Healy: Wow.

6:59 Marty Burgess: One of the top pros in the world because we can train AI using a variety of different training methods to act and feel and behave like a human would behave in different situations.

7:10 Georgie Healy: This is mind-blowing. Like, I had Steve Hind from Laura Keet. He was talking about chatbots that behaved very human to the point where we didn't know that they were chatbots. And it was incredible because the experience is so positive. But I never thought about agents in gaming. Even I know that people get very stressed out if they have to pause a game, leave a game, forfeit a game. I'm curious if you'd use AI in any other ways, or this is the sole focus for you guys at the moment, is the agent for replacing players or being used in place of players.

7:47 Marty Burgess: So with LFG, where all of the AI tools and technologies we're building, we're not just building them for ourselves, we're going to make them available in a SaaS-style product called Saturn AI. Because as we literally say in like the first page of our pitch deck, what we've built is just too fucking cool to keep to ourselves. Apologies for that. You can bleep that if you want.

8:06 Georgie Healy: No, we won't bleep it.

8:08 Marty Burgess: Okay, cool. Easy enough. I mean, my studio's name is Let's Fucking Go.

8:12 Georgie Healy: Yeah, yeah, you're on brand. Yeah, exactly.

8:15 Marty Burgess: So we want to basically take what we're building, and like I said, AI agents as a core gameplay mechanic is step 1 of the process. Once you've got that built within the game, and again, it's up to the individual game or studio if they even tell the players that they're doing this. We have the attitude for our games that players know that we're doing this. It's more a case of our games are more like playing Football Manager, where you are training your army of heroes and then you can choose to play alongside them, or you can even just send them off to go play while you're at work. Let them do all the boring stuff, grind out the battle pass and unlock achievements, stuff like that. And then you get to do what you want to focus on when you get to sit down and play.

8:54 Georgie Healy: More strategic type stuff, Marty, or?

8:56 Marty Burgess: I mean, it's up to you.

8:58 Georgie Healy: Whatever you like. One of the crazy examples that we didn't even you can think of.

9:02 Marty Burgess: We've been doing little mock tournaments within the studio to test out a lot of the stuff we're building, and some of our developers would use AI to play through the first couple of rounds of the tournament. Then when it gets to the finals, they would jump in and take over at that point. So it's really like you're building this team of AI agent-powered heroes— heroes in our case, but it can be cars, planes, goblins, whatever you want, obviously, depending on the game. So for us, AI agents as a core game mechanic is step 1. And then we're looking at how we can then use AI tools beyond that. For example, once you've got the ability to deploy AI agents within a game, one of the biggest frustrations that you will have for a lot of multiplayer games especially, imagine you wanted to jump into a game at 2 o'clock in the morning, but you need, maybe it's a battle royale, so you need 50 or 100 players to have a really great game. But we're in Australia, it's 2 o'clock in the morning, and there's only 20 other people online at that particular time. So why don't we, after say 5 or 10 seconds, just fill the match with AI agents trained off players who are all around your skill level so it feels exactly like you had 100 players online ready to go, but the matches start in 5 seconds?

10:13 Georgie Healy: That's brilliant.

10:14 Marty Burgess: Like, I know for some League of Legends players in particular, if you're at a really high ELO, so very competitive player, Sometimes you have to wait literally hours to find a match. Or, and this is always a trade-off, you could put them into a match with lower ranked players, but they're not really going to have a great time because it's not that fun after a while just beating up on newbies who have no idea what they're doing. It's a terrible experience for the newer players just getting wailed on by some semi-pro or professional player. I remember there was a talk given by one of the developers who worked on Gears of War 2. This is again, Showing my age here, I had hair when I started working in games. If you like having hair, don't work in games.

10:54 Georgie Healy: Simple as that. It's the headphones that do it.

10:57 Marty Burgess: Yeah, exactly. But what they found was 80% of players who jumped into a multiplayer game and lost their first match never went back into multiplayer. 80% of their players gone if they didn't win that first match.

11:09 Georgie Healy: I find that relatable. I wouldn't go back.

11:13 Marty Burgess: If you went bowling and you just completely suck, you're not incentivized to go back because you've had that negative experience.

11:20 Georgie Healy: Me and all sports. Yeah.

11:23 Marty Burgess: If you go bowling and your first ball is a strike and everyone's cheering, you're going to feel great.

11:27 Georgie Healy: I love bowling.

11:28 Marty Burgess: And you're going to— there you go. It's as simple as— this is not unique to gaming. This is just human psychology. But if we're able to level the playing field, if we're able to make sure it doesn't matter, like you can jump online as a new player, as a professional player at 2 o'clock in the morning, you know, in a lesser populated server. And bang, straight away you get a great match. Now, when I say great match, you might still lose, but I'm hoping if we can get the balance right, the matchmaking, and really nail it, I'm hoping it was a nail-biter. I'm hoping there was back and forth. I'm hoping it went down to the wire. Like, that's the sort of experiences that we see as being able to deliver using the power of AI. But like I said, it all starts with that core AI agent as a game, like core game mechanic. And then we can build and expand on that in so many interesting ways.

12:18 Georgie Healy: Fascinating. This next section of the show I've kind of titled AI and Gaming 101. I've got some terminology that I don't know the answer to, and I'm hoping you can help me and the listeners understand what this stuff is. What is perfect game balance, and do you need AI to achieve that?

12:38 Marty Burgess: God, how long do we have?

12:41 Georgie Healy: 5 minutes.

12:42 Marty Burgess: 5 minutes. Okay, cool. I'll bring it right down from the Perfect game balance is, one, it's very game dependent. It's very dependent on the player taxonomies that you are bringing into your ecosystem. It's very dependent on how long they've been there, their level of expertise, their experience with other games in the genre, a whole bunch of different things. But at a super high level, perfect game balance means that a player is able to join, they are able to, in a relatively quick amount of time, have a positive experience that ticks the boxes of what that particular person is looking for in your game. Now notice how I didn't say win or lose, because certain players like Bartle stacks on it. We've got this whole bunch of different ways to break down player taxonomies, and Bartle's is one of the best ones, especially for multiplayer gaming, where we look at certain player types. Some we call killers, and they're the diehard, they must win at all costs. So they're going to study they're gonna, you know, they're gonna eke out every tiny thing they can do to have the best chance of winning the game. But some of your players are social players. They are there— think games like Roblox, for example. Think what we're seeing quite a bit in games like, say, Fortnite, where, you know, people aren't playing, you know, Roblox to win. It's their primary form of communication with their friends.

14:03 Georgie Healy: Mm-hmm.

14:03 Marty Burgess: They are joining for the social experience. They're, you know, maybe they're streaming, maybe they're creating like YouTube or TikTok content. You know, if you think of a game, maybe like The Sims as a really great example. Most people aren't playing The Sims to win, but they're looking to either have a great social experience, maybe they're looking to explore, maybe these are people that want to see every bit of content that is in the game. These are the people that will explore your whole game world. They'll chat to every single character. They will know more about the backstory than you potentially know. They may take it even further and they write fan fiction, they cosplay at events. So perfect game balance isn't just a win or lose question. It is really a case of you identifying what certain players are looking for in the game and then trying to tailor an experience that gives them that as close to that experience as possible.

14:59 Georgie Healy: Sounds so complex, Marty.

15:01 Marty Burgess: It is.

15:02 Georgie Healy: A lot of data. How do you do it? Does it need AI, or we can get away with it without it?

15:08 Marty Burgess: I mean, as an industry, we've done pretty well since Pong, so we're getting there. Game balance and game design has come a long way. You think of what things were like back in, like, you know, say the '90s, like the arcade era, where games were literally designed to try to get as much money out of you as possible. So they were deliberately hard, or you'd get into situations where you just couldn't win. Thankfully, we've learned and we've moved past that as an industry. Where AI really opens up new possibilities, one, as you said, there's a lot of data just for us to process. If you have a game where you might have, say, millions of daily active users, they're all going in, they're all doing different things. We capture as much data about their play patterns, their behavior. You know, hey, we've got a bunch of players who go into the shop and they look at this sword, but they don't buy it. Is the sword too expensive? Does the art not match what that particular person is looking for? Again, not everyone is looking for the most powerful-looking sword. Maybe they're looking for the coolest sword that goes best with the outfits that they're wearing in the game.

16:07 Georgie Healy: Mm-hmm.

16:08 Marty Burgess: So there is so much data flowing into the system. AI is a brilliant way of trying to process it and convert zeros and ones into something which is able to be absorbed by human beings. If you can run all that data through an AI system and then have an LLM trained off that data where a designer can just go, hey, ask a very natural language question and get natural language responses, that removes the barrier. All of a sudden, to be a game designer, you no longer have to be a data scientist.

16:42 Georgie Healy: Mm-hmm.

16:42 Marty Burgess: And those are two weird things that you would never think were intertwined, but some of the best designers know Excel better than they know game development tools because of just the raw amount of data we've had to process. The other side to this as well, where AI— and again, this is very much the direction currently that LFG is leaning into— AI allows us to, in real-time or semi-real-time, tailor custom experiences for different players, different guilds, different cohorts of players. If we're trying to build events or build scenarios, dungeons, whatever happens to be for different players. Unless I've got millions of designers, programmers, artists, it's an impossible task. As an industry right now, we're seeing the fallout of just the exploding cost of producing video games and producing high-quality content. Games, I think the current estimate I've seen for GTA 6 when that comes out is that would have cost over $1 billion to produce.

17:40 Georgie Healy: I even know that's Grand Theft Auto, so major.

17:46 Marty Burgess: GTA 5, one of the most successful games of all time. The sequel is probably one of the most highly anticipated games of all time.

17:54 Georgie Healy: Yes, everyone keeps talking about when it's going to be released, right?

17:56 Marty Burgess: Rockstar, brilliant marketing move, not to derail, but they dropped an announcement for their trailer and the announcement video got like tens of millions of views. So just the fan base is rabid. But if it takes $1 billion to make that, well, one, I don't know about you, I don't have $1 billion lying around. Maybe if LFG goes well, who knows?

18:19 Georgie Healy: We'll hit you up.

18:21 Marty Burgess: Well, just the rampant cost of development, of production, the human hours it takes. AAA gaming especially, most games take 5 to 10 years from initial storyboard to to commercial release. Then you've got live ops where you keep building and supporting the game beyond that point in time as well. So all of a sudden, you're trying to support a game that was started development on 10-year-old technology with a team that probably has completely rolled over. It's a whole new team now compared to the people that started the project. It's insane.

18:50 Georgie Healy: Wow.

18:51 Marty Burgess: Using the tools that we have from AI, if we are able to make that content quicker, easier, cheaper, that's a huge benefit for everyone involved, if we're able to tailor that content more to specific— if we could get to a world where I can build, you know, a level in a game designed for Georgie, you know.

19:11 Georgie Healy: Even me. Yeah, that would be amazing.

19:13 Marty Burgess: If we can really break it down where, and at the individual player level, you know, we're able to create tailored content suited to that player, that's the dream. Because at the moment, you know, sure, we might do an event which, you know, targets new players, existing players, and the pros. But that's still 3 buckets. Imagine if we can add more buckets into that mix and we can really sort of create that in semi-real-time or in real-time, that unique content. That's what's so exciting about the potential of AI.

19:45 Georgie Healy: Yeah, all jokes aside, I have told numerous people that I feel like I've really missed out on something that I would've it would've suited my personality. I think I would've been a really massive gamer and it's just due to the circumstances of my childhood that I didn't get into gaming. And so I often say I feel like I've missed the boat and it's quite sad, right? But I didn't know that AI could really open up opportunities in a way where it's worth exploring those markets if you can reduce the cost of production.

20:18 Marty Burgess: But here's the beautiful thing. I don't know you that well, but I'm going to argue that you are a gamer. You just don't know it.

20:23 Georgie Healy: That makes me feel really happy.

20:25 Marty Burgess: Have you ever played a Sudoku? Have you ever done a crossword? Have you ever grabbed a deck of cards and played Patience?

20:31 Georgie Healy: Yeah.

20:31 Marty Burgess: Guess what?

20:32 Georgie Healy: Uno.

20:33 Marty Burgess: Guess what? You're a gamer. We've got to break past this thing where gamers are, you know, the 16-year-old teenagers locked away in the basement with the Xbox. Like, you know, one of my favorite games right now is competitive Sudoku. I just can't get enough of it. It's great. I'll also play other more traditional gaming, but my grandmother is 93 years old, does the crossword puzzle in the paper every single day. She is a gamer as far as I'm concerned. Now imagine if using that simple example, imagine if using the power of AI, I was able to create a custom tailored crossword puzzle using facts that I know about her life, her experiences, things that she's interested in. And that's a unique puzzle that only she gets to experience unless she chooses to share it with other people. Come on, that's cool.

21:20 Georgie Healy: That is super cool.

21:22 Marty Burgess: And that's a very simplified level. That's what we're hoping to achieve en masse using the power of AI.

21:29 Georgie Healy: My game would have dragons and like, it'd definitely be old school fantasy, medieval concepts woven in. What would your, like, would it be a pirate game like your old school childhood relived, Marty?

21:42 Marty Burgess: What would it have in it? This is the dangerous question. Every game designer has their magnum opus. They have their dream game they would love to make. Yay! I don't think I would ever have the courage to try to make a pirate game because I don't think it could ever live up to what, like, the nostalgic memories I have as a child in my head. I don't think I could ever actually get there. One of the big things, and again, we're not doing this directly with LFG, but you'll see shades of influence, I guess, carrying through. One of the big things for myself, and this has been swinging in my head for a very long time, where I hate the fact that to play certain games, I have to go upstairs into my office and sit down in front of my computer with a keyboard and mouse and play it. I have a nice TV downstairs with an Xbox plugged in, but some games are only on one or the other. I would love to build as many as I can, and our first title we're building does exactly— hopefully will do exactly this. Mm-hmm. I would love to build games where you can play them on whatever device you want to play on. That's step 1. But I would love if the game gave a customized, tailored experience based on the device that you're playing on. So, for example, if you take, like, use the Pirate game as an example here, you know, if I'm on a computer, great, I've got a nice big screen, I've got a keyboard and mouse. You know, maybe this is where I do my trading. This is where I organize for my fleet of AI-powered ships. To go out, go to certain markets, buy goods, bring them back, sell them, stuff like that. It's very suited for this platform. When I'm downstairs on the Xbox, well, okay, there I want an action game. I want to play something like really fun and fast-paced. So maybe there I'm a swashbuckling pirate, like capturing, you know, Spanish galleons and, you know, swinging on ropes and stuff like that. Like something very suited for that platform. But then, okay, well, this is the most powerful gaming platform on the planet right now.

23:36 Georgie Healy: You're showing up your phone for those listening on the podcast.

23:38 Marty Burgess: Oh, sorry. Yeah, yeah, mobile. Yeah, sorry, I've got to remember. I can see you, but—

23:42 Georgie Healy: We've got both multimedia options.

23:45 Marty Burgess: Thankfully, I've always been told I have the face for radio, so this is great.

23:49 Georgie Healy: Why do you think I started? I don't have to wear makeup. It's great.

23:53 Marty Burgess: But mobile, like, so now you've got a device that's always on you, that's always active, that, you know, you can just very quickly, like a gaming session on mobile can be 5 seconds. So maybe there, if we had, say, an LLM trained in the, you know, talk like a pirate one, which would be fun, but it's trained on what your crew and your team are doing. Maybe I start getting SMSes on my phone going, hey, just letting you know, we got to the town to buy the sugar that you want us to buy, but they've got a sale on salt at the moment. Do you want us to buy that? And I can just text back from my messaging app.

24:26 Georgie Healy: Ah.

24:27 Marty Burgess: And that will interpret it. And that's affecting the game that is currently going on. That I could then choose to interact with from any device, like that tailored experience based on the device that you're on. I think that has huge potential ramifications for the experience you can deliver to the users.

24:47 Georgie Healy: AI making that world building and really feeling like you are living that life because it is starting to overlap with your everyday. How incredible. I read a lot of fantasy and, you know, a really good book, you kind of every now and then, like, will get the emotions that you're feeling due to that character. And you kind of wish that you could kind of feel that more. And I feel like gaming is a great way to really have this alternative universe that you could be part of.

25:23 Marty Burgess: Imagine, you know, again, I work in games, everything in my head has to come back to gaming in some way, shape, or form. But you say novel, I think interactive novel. Interactive fiction is a huge genre in gaming. It's a very valid genre. Imagine if you could actually be part of that story. Imagine if, you know, we've seen like, remember the old Choose Your Own Adventure books?

25:46 Georgie Healy: I loved those books, yes. They were great. Go to chapter 8 if you want to follow this journey.

25:51 Marty Burgess: But still, it's like, oh, you know, you walk down a hallway, there's a door on the left, door on the— it's, yeah, it's, you know, do you do the door? But maybe I want to turn around and walk backwards.

25:59 Georgie Healy: Ah.

26:00 Marty Burgess: Maybe I want to scream. Maybe I want to check the painting on the wall. Interactive fiction, calling it now, AI is going to completely change the way we build those experiences because it's less about you presenting, you know, a linear or a branching path to your players and more about you give them the, the way we look at AI as a studio, we built the sandbox, which is the game. The AI is the toy. So if you can build that sandbox in an interactive story, then the way that you could choose to absorb that content— and everyone gets a slightly different experience as well, and that's beautiful. So when you're having the water cooler talk on the Monday morning, it's not a case of, "Oh, did you see what Daenerys did in Game of Thrones on the weekend?" Yeah. It's like, "Oh, for you, they did that, but for me, they did something completely different." Not my Daenerys. Exactly. Like, you know, everyone has their own tailored experience within the game or within the story.

27:01 Georgie Healy: You've talked about the incredible possibilities in gaming through AI, but it does make me wonder in the back of my mind, woof, that's a lot of compute. How do you rationalize that? As you're building your own business in AI and gaming?

27:19 Marty Burgess: Absolutely, it's a very valid question. And again, being the fact that we're a studio that works in both the blockchain space and the AI space, I guess we've got a bit of an advantage that this is not something which is completely new to us. Because if you go back, say, 5 years, this was a big concern around a lot of people mining crypto and stuff like that. Back when we were using GPU compute for, say, proof of proof of work and stuff like that. The way I look at it is what it really comes down to is developers, engineers, the people building these tools and technologies, getting as efficient as possible in the ways that we are working. We're already seeing this in the world of LLMs. We're seeing more refined models that use a fraction of the compute of some of their larger models, but still generating the same or sometimes a better outcome. So optimization is just a huge part of what needs to be happening in the space. And, you know, to be fair, we're trying to focus on that very heavily as a studio because if we can optimize, that brings our costs down and, you know, we can scale more aggressively and deliver this experience for more people. So it makes a lot of sense from both sides of the fence.

28:22 Georgie Healy: Don't give any secret sauce away, but when you say optimization, like, where can one optimize in this gaming space?

28:30 Marty Burgess: There's so many different ways because obviously At a fundamental level, if you can have a model which runs more efficiently or doesn't require high-end GPU, if you can have something that can run on the end user's device, for example, it doesn't need a server farm and stuff like that. That's a big part of it.

28:46 Georgie Healy: Sure.

28:46 Marty Burgess: But also using the interactive story example, if I can, instead of just asking the model, "Hey, tell me what happens because they went left instead of right," have it go down that path a little bit further, have it generate several potential options down the road. And then, you know, potentially if you do go down that path, which again, the AI is probably going to get it right because it's been studying what you've done up to that point in time. You know, I've removed several different actual generations in the first place. So, you know, I can exponentially reduce the actual compute requirements just through more efficient design in how I'm choosing to utilize AI.

29:24 Georgie Healy: I have another term that I'd love you to break down for me that's apparently very hot in gaming right now. What's a dynamic world, and is it achievable?

29:34 Marty Burgess: Okay, that's a big one.

29:36 Georgie Healy: Another 5-minute easy question for you.

29:39 Marty Burgess: Exactly, yeah. We could have done this all day. This is great.

29:41 Georgie Healy: Making this so easy.

29:43 Marty Burgess: So dynamic world is kind of like the Turing test, but for the real world, to think of it in that sort of way. So the Turing test, for anyone who may not be aware, was a thought experiment that was basically, could you communicate with a computer and have it communicate back in a way where you could not tell apart from a human. Now, to be fair, computers beat the Turing test well before the recent— I mean, AI has been around for a very long time, but the recent proliferation of AI and LLMs and stuff like that. So this is nothing new. A dynamic world is taking that one step further. So this is where you're in, say, a VR headset, and I'm sure part of your brain would know you're sitting on a couch with a VR headset on. That part is fine. But being able to look around and see a world which is being— you're not supposed to use the name of it in what you're saying, so I can't say being generated dynamically.

30:33 Georgie Healy: Mm-hmm.

30:33 Marty Burgess: But seeing a living, breathing world which is not being driven by an overarching rule set, it is being driven, I guess, bottom up, where each entity in that world, I would like to think, and Now we're getting into fate, karma, and all sorts of crazy discussion topics here. I love it. I personally don't think there is an overarching plan driving my activities.

30:59 Georgie Healy: You're not in The Matrix, Marty. No? Maybe.

31:03 Marty Burgess: Maybe. But imagine a game world where a character in that isn't being told, like, okay, you walk between these two points on the map, and if something happens, this is how you respond to it. You're more giving them 'Hey, you're in this world,' and then they are able to determine, 'Okay, cool, what do I need? Well, I need food. How do I get food? I need a job. How do I get a job? I go to the job market,' and that's more the progression that this character has within the game. And then if you're able to scale that out from there, you're creating this very real— And real in the way that it has flaws, it has imperfections. The real world is not Perfect. And for better or for worse, a lot of gaming worlds, because they are very structured and very programmed, feel wrong. They don't pass the sniff test. They feel very orchestrated. And the real world doesn't do that. Everything changes. Everything's very dynamic. Everything's very fresh. Everything's— I could walk outside right now, get hit by a bus. We're not able to predict what is going to happen. I don't know what question you're going to ask next. I can try to guess, but I'm probably going to get it wrong, and that's wonderful in so many ways. Dynamic worlds, dynamic content, it's a whole potential new— I don't say genre because it's not necessarily a genre into itself, but it is a genre-breaking development if we're able to get to that point. Because you think of You think of Mario, to go back to that very simple example where speedrunners have solved that first level of Mario. But if every Goomba in that game is acting the way they choose to act on any given day, now we're getting something interesting. Now it's not predictable. It's not going to always play out. Maybe it does play out the same, but not in the exact same way. It opens up a very interesting world.

32:57 Georgie Healy: And it's so fascinating when you think of dynamic worlds. And thank you for the explanation. I had no clue what it was. But like, maybe some people are like, games are my happy, safe space where I feel like, you know, I know where it's going and that's a good thing. My real life is too chaotic. It'll be interesting to see who takes it up. Or am I missing something? Do you think everyone would prefer the dynamic world?

33:28 Marty Burgess: I'll be honest, I can't answer that because we've never gotten to a point where we've built one good enough. Where players can actually have that choice in the first place.

33:36 Georgie Healy: Yeah.

33:36 Marty Burgess: In the same way that there are different people looking for different experiences in a game, I would say that there's probably going to be some players who love this idea and some who are repulsed by it. But that's fine. There is enough different studios building enough different amazing games out there that hopefully you will never be in a world where it becomes like an A/B choice. You can always choose to play the kind of experiences that you are looking for. We're just trying to add a whole new flavor of gaming into that ecosystem.

34:10 Georgie Healy: I'm having so much fun with this, but I'll get a move on. This is the last question from this section. How do you feel about AI playing Pokémon now? And apparently Mario. Is it a good barometer for model quality or no?

34:24 Marty Burgess: Yes and no. So let's probably expand on that a little bit. Otherwise, this is a really bad way to end this section.

34:31 Georgie Healy: Do tell.

34:32 Marty Burgess: For everyone listening and watching. Okay, so a game like, say, Mario or even Pokémon, in a lot of ways, they are deterministic experiences. As I said, Mario is a great game where if you can memorize certain inputs, you will beat every level in a perfect time. And we're getting very close to humans being able to reproduce what we call TAS runs or tool-assisted speedruns. So humans being able to match what a computer can theoretically is the perfect way to play the game.

34:57 Georgie Healy: Cool.

34:58 Marty Burgess: Pokémon, a little bit different because obviously there's an element of RNG randomness that can happen in the game. So a model has to now actually adapt for that on the fly. It's not just like rope learning, it now has to actually adapt to different situations. So in terms of where I see AI going, Pokémon is by far the better example of that. It's kind of like when you're studying. If you just memorize the answers to an exam, cool, that's one way of passing the exam. That's the Mario example. But the second someone asks you something which is not that exact question, you just, you fall over. The Pokémon example is more you actually having a deeper understanding of the content in the exam. So when you get thrown a curveball, you can hopefully deal with it. Or more importantly, and this is where AI is so wonderful, by failing, you can then learn from that mistake. And that's like, I don't like thinking of models as like you train it and it's done. True AI, especially in the gaming space, needs to be constantly being trained and retrained because you want the AI to be learning from the experiences that it has. With the games that we're building, we see that training AI is a core part of the player experience where they choose to go and train their heroes and go, hey, you know, here's a new strategy I worked out, go learn how to do that better. Because by doing that, now you're developing this game where, you know, if I play it today and I play it in a month's time, the experience in a month is completely different to what I get today because the AIs have learned, evolved, and grown and developed new strategies, new tactics in the same way that humans would do that as well. Again, the way that we play Mario today is not the same as we played it 20 years ago. We've learned new tactics and new strategies along the way.

36:37 Georgie Healy: Yeah.

36:38 Marty Burgess: If you can get to a world where AI is helping you shift and evolve your metagame, so what's actually happening, the tactics, the strategies, how people approach it, that becomes such an interesting, fun way to experience games in a whole new way. And it's something you can apply not even just to new games being built, but you can go back and use them on things like Pokémon, like Mario and stuff like that. And see how AI can approach these existing puzzles but in new innovative ways.

37:11 Georgie Healy: Absolutely fascinating. My husband watches people play games on YouTube. I'm sure there's a better term for that. Do you think people will watch AI play games as well?

37:22 Marty Burgess: Yeah, it's a big part of what we're— so we do design documents for everything we create, and for Gladiator Mayhem, our first game, one of the first things in that design doc is this has to be a game which is as fun to watch as it is to play.

37:36 Georgie Healy: Mm-hmm.

37:37 Marty Burgess: Because if it's not fun to watch it, then cool, you might train your AI, but you're not going to enjoy watching the result of that. So, you know, we're putting time and effort into things that in a normal game we wouldn't have to worry about. Cinematic cameras, replay modes, AI-generated highlight reels with one-tap sharing to TikTok and your favorite streaming platforms, or, you know, doing AI-powered narration or announcers. All things that you don't necessarily need in a traditional game. But for us, we want to make the experience of watching it— and again, hopefully we get people who choose to stream that experience and share that with their communities as well. Organic growth is a thing. But yeah, if we can make that just as fun as playing the game itself— you know, the example I always fall back to, myself, a bunch of mates will come around on a Friday night, we'll throw FIFA up on the TV and we'll put Barcelona, Man United, and we just turn it into a drinking game because that's a beautiful game. Sorry, FC 25 it's called now. They changed the name for reasons. But yeah, like sitting back watching it, having it as a fun thing, it's like putting the footy on but just in a nerdy digital format. And I would love to be in a world where maybe someone on the bus to school or on their morning commute on the train, doesn't necessarily feel like playing one of our games, but fire it up, send the AI off to do a couple of battles, and just watch how it does. Or maybe watch other people's battles playing out, or maybe there's a tournament running, watch highlight reels that again we put together using AI, where the AI will study a match and pick out the most interesting moments, move the camera dynamically.

39:18 Georgie Healy: Okay, I am putting it into to the universe that you're going to give me an invite when you do this with the AI playing the games. Okay, big party at yours.

39:28 Marty Burgess: If any of your listeners happen to be in San Francisco, next week is Game Developers Conference. It's the biggest gathering of game developers for the calendar year. There's actually an Australian showcase on the Monday night. It's at a place called Buzzworks in San Francisco. Please feel free to reach out to me on social media. I'll happily give you a VIP ticket. And it will be the first opportunity anyone of the public will have to play our flagship title, Gladiator Mayhem.

39:55 Georgie Healy: So the date and the time in San Francisco again?

39:59 Marty Burgess: Next Monday. Monday.

40:01 Georgie Healy: So the 17th.

40:02 Marty Burgess: 17th, yep. BuzzWorks San Francisco. I think all the normal tickets are sold out, but just reach out, I'll give you a VIP ticket. Because yeah, like I said, this is the first time the public will ever get to get hands-on with Gladiator Mayhem.

40:15 Georgie Healy: You have quite an active community on Twitter, but I think LinkedIn's probably the best way to reach out to you.

40:22 Marty Burgess: I try to keep an eye on all of them.

40:24 Georgie Healy: You got fire.

40:25 Marty Burgess: You know, Twitter, sorry, I still call it Twitter, not X. Sorry, X. X, yes, sorry, sorry, Elon.

40:31 Georgie Healy: Sorry, Elon.

40:32 Marty Burgess: LinkedIn, I try to be professional. Twitter, I don't try to be professional.

40:35 Georgie Healy: Same.

40:36 Marty Burgess: Yeah, pretty much. So yeah, you know, I'm on Instagram too, but you probably won't find me there. 'cause it's all just photos of my cats.

40:43 Georgie Healy: Oh, that's a good sell. That's a good sell right there. We're gonna put links to maybe all except the cats in the show notes, guys. I am so jealous. I will not be in San Fran on Monday, but that is huge.

40:57 Marty Burgess: We have a full camera crew coming, so we'll definitely send you some footage. And yeah, it's gonna be a big week.

41:02 Georgie Healy: I'm gonna put that on my Instagram as well, any footage you send me, okay?

41:06 Marty Burgess: Nice, awesome.

41:07 Georgie Healy: So that everyone can see it. We're at my favorite part of the show, the rapid fire questions. Title in progress every single week. I'm like, these aren't rapid fire. They're just spicy hot takes. Are you ready, Marty? Are you ready for the spiciest part of the show?

41:23 Marty Burgess: As we say, LFG, let's go.

41:25 Georgie Healy: Let's fucking go. So who is the gaming GOAT? I've listed Nintendo, Electronic Arts, and Microsoft.

41:35 Marty Burgess: Other? God, any answer I give you, I'm going to offend like so many people I know.

41:40 Georgie Healy: There goes your following.

41:42 Marty Burgess: Yeah, pretty much. Bang. It was fun.

41:45 Georgie Healy: Your current fave on Monday the 10th of March.

41:49 Marty Burgess: Monday the 10th of March, my partner's birthday. Happy birthday if she listens.

41:53 Georgie Healy: Happy birthday, shout out.

41:56 Marty Burgess: Okay, I— God, Sony is doing amazing stuff in terms of just beautiful console-based experiences. I love the stuff that's happening in the mobile space at the moment, especially from Turkey. There's so many amazing mobile studios. You look at games like Royal Match and stuff like that, which is absolutely killing it. There's so much innovation happening in the mastery genre, which is crazy to see. We're seeing story-based experiences like Gone Home, one of my favorite games of all time, is just this beautiful narrated walking experience, which is always going to be great. But there's, you know, we're seeing co-op stuff coming out now. Nintendo obviously are always going to be like just the GOAT, especially in terms of their particular like brand and their particular style of gaming. Yeah, I don't think you can ever pick one because there's so many different facets of what, you know, what makes the GOAT in so many different ways. And, you know, even— Okay, I'll bring it back a bit closer to home. Just shout out to Australian studios. The guys at Chaos Theory doing amazing educational games. The guys over at SMG making beautiful— they did Moving Out 2. They've got some stuff coming out on Steam, No Way Home, Sping, stuff like that, which is amazing indie stuff. Australia-made Mello, Australia-made Flappy Bird. Yeah, not Flappy Bird, Angry Bird.

43:20 Georgie Healy: Yeah.

43:20 Marty Burgess: No, the bird one.

43:21 Georgie Healy: Anyway. There's a bird one.

43:23 Marty Burgess: Yeah.

43:23 Georgie Healy: Uh, we'll find it. Thank you, that is a really beautiful answer. It sounds like it's a community that's not competitive if you can see the joy and the benefit in many different houses.

43:35 Marty Burgess: It's a close-knit community. I'd like to think we also try to help each other as much as we can. There's enough gamers to go around, basically. So yeah.

43:44 Georgie Healy: We touched on this earlier, the perfect age to get into gaming.

43:48 Marty Burgess: Every age.

43:49 Georgie Healy: That is the politically correct answer. What's the actual perfect age?

43:55 Marty Burgess: The best age is whatever age where you find joy in a game, in an experience, because it doesn't have to be, you know, your— like I said, you know, the stereotypical 18 to 25 male playing a, you know, Call of Duty game. It can be, you know, a younger, you know, primary school age kid playing Uno with their family and using it to learn to count and learn colors and things like that. It can be my grandmother in her 90s doing the daily crossword puzzle. Gaming is not exclusive. We shouldn't gatekeep it in any way, shape, or form. And the way you choose to enjoy games is up to you. No one should ever be in a position to tell you how to enjoy a game.

44:33 Georgie Healy: Mm-hmm.

44:34 Marty Burgess: It should be up to you to find joy in whatever sort of game that you particularly like to play. And there are so many different experiences available in so many different formats. It's on your phone, it's on your computer. There's so many free-to-play games now, which, you know, is beautiful in its own way because it's not, you know, literally the barrier of entry is you just need to have access to the internet to download a game, and that's amazing in its own way.

44:59 Georgie Healy: So— I think that's fair. Like, and as you were saying that, I think of all these, um, like boomers that play— what's the jewel game where you match up jewels and they explode.

45:11 Marty Burgess: Oh, Bejeweled.

45:11 Georgie Healy: Bejeweled. Candy Crush. Candy Crush.

45:14 Marty Burgess: Royal Match.

45:14 Georgie Healy: It gives them so much joy. Who am I to say that that's not a real game?

45:20 Marty Burgess: Like, I know some people who are like level 10,000 in Candy Crush. You know, they put more hours into that game than most people put into like a AAA title. And if they find joy in that, then that's beautiful in its own way.

45:32 Georgie Healy: I would love to end it there, but I've got more questions.

45:36 Marty Burgess: Let's go.

45:37 Georgie Healy: They're just too good. I can't not ask them. We had Alex Valente from Redactive on the show He found his CTO from a very active group chat. I think it was World of Warcraft. If you had to hire an avid gamer or an AI expert for LFG and you had to pick, is it the gamer-obsessed employee or is it the AI aficionado?

46:02 Marty Burgess: All right, I'm gonna give, I'm not gonna dodge the question, but I'm gonna give a slight—

46:06 Georgie Healy: You're gonna dodge it, aren't you, Marty?

46:07 Marty Burgess: A little bit.

46:07 Georgie Healy: You so are, I see that you are.

46:09 Marty Burgess: So there's two things that I think of when I think of this. I am going to pick either the gamer or the AI expert, but I'm going to pick whichever one is more passionate about gaming or AI. Because at the end of the day, like, AI is such a new field. Like, we're all learning as we go. Gaming changes all the time. We are all constantly learning and evolving what we do. You know, any skills that you may have from 5 years ago are horribly outdated. The person who is the most passionate, driven, Like, I want the person who is listening to the podcast on the drive to work because they want to get some more, some, you know, current insights and find out what's happening, what other people are doing. I want the person that, you know, reaches out on LinkedIn going, hey, I love what you're doing, can you explain why you're doing this? And I will always take the time to try and answer and try and help the, the industry or the community grow where we can. The other side to it is the layover test. If this is a person I'd be happy, you know, a 3-hour layover on a flight somewhere.

47:04 Georgie Healy: Yes.

47:04 Marty Burgess: If I'm happy sitting in a hotel or an airport bar, having a drink, having an interesting conversation with that person. I think we always forget that work, even in the AI space, in the gaming space, it's done by humans, it's done by people. And building a team where everyone feels psychologically safe, building a team where everyone feels heard, building a team where you feel excited and energized to just be in a room with them, I can tell you right now, both from our other co-founders, from the team that we've built, I, in most of our meetings, I am the dumbest person in that room, and I love every second of it because what I get to learn and grow and develop new skills. But like, yeah, if you're the smartest person in the room, find a new room. When you're like hiring, I want passion. I want, you know, people that I am just, you know, I can't wait to see what they can actually do. you know, build.

47:58 Georgie Healy: I love it. You've, you've basically named why I started this podcast too. I just want to talk to really smart people like yourself, Marty. Look, thank you so much for joining the show. This has been— I don't know where the time has gone. I have had genuinely so much fun. I've learned so much, but you've got a huge next less than a week coming up. I would love to give you a chance to shout out to ask the listeners something that they need to know about LFG and how they can support or help or get involved.

48:30 Marty Burgess: Yeah, obviously that's really awesome. So yeah, as I said, big week for us coming up. We're gearing up for the Game Developers Conference over in San Francisco. We all fly out, I think, on the day that this is going to air. So if you're going to send me mean comments on LinkedIn, I won't respond for a few hours because I'll be in the air somewhere. But yeah, for us, honestly, the best thing anyone can do, and we I get this question a lot, like, how do we support you? It's following us on social media, following us on Twitter, jumping in our Discord, having conversations, talking about the things that we're doing. As I said, we're very, very close to having a test version of Gladiator Mayhem start going out for the community to get involved with. Play the game, try it out, give us your feedback, whether it's good or bad. Tell us what you like, what you dislike. That aspect of support is criminally underrated. And you have no idea how motivating it is to have someone go, "Hey, I played your game. I thought it was awesome." That's just mainline adrenaline right there. That's as simple as it gets. We do this, you know, one, it's pretty fun. We get to have some weird meetings where we're talking about like, "Hey, you know, when I hit the goblin with the baseball bat, how far should he fly off the screen?" And everyone's getting a whiteboard out and they're like, "Okay, okay, how do we do it?" These are serious conversations that we have. Especially on the gaming side. No one works in gaming to make money. We could all make a lot more working at banks, working at financial institutions, building login pages for websites or whatever it happens to be. We work in games because the joy that you feel when you get to create something amazing and put it out into the world and have people enjoy what you've built, that's an experience that just I would not give up for all the tea in China. That's, yeah, so like I said, follow us on socials, connect on LinkedIn. When you see us post about GDC and stuff like that, throw a comment, throw a like, reshare it or whatever LinkedIn calls it. I'm supposed to ask for you to like, subscribe, hit the notification bell.

50:32 Georgie Healy: Do all of it, guys.

50:34 Marty Burgess: Do all the things. But yeah, that in itself is more powerful than literally anything else.

50:43 Georgie Healy: Look, I will be doing all of those things. This has been an absolute privilege, Marty. Thank you so much for joining the show. Such an amazing time for you guys at LFG. I can't wait to see what happens after Monday and have the best rest of your week.

50:57 Marty Burgess: Thank you so much. LFG!

50:59 Georgie Healy: LFG! Let's go! Listening to In the Blink of AI. You can check out the show notes for anything discussed in this week's episode, and we will be back next week. This podcast was produced by Day One with music by Dan Hansen and visual artwork by Sophie Tyrell. If you loved the episode, please tell your mates, and I love AI news. Please share your thoughts and suggestions to georginarosehealy@gmail.com.