Max Child: Reinventing Gaming for the AI Era

25:42And, you know, I started playing around with it and asked it to play music and, you know, telling it to play Spotify from 15 feet away. And it like worked. and for anyone who had used Siri or any other voice system at the time, the leap forward of Alexa cannot really be overstated in how much better it was. I mean, it seems boring in retrospect, but being able to talk to a speaker from 10 feet away and say something in a pretty arbitrary way and have it actually do it was like maybe not an iPhone moment, but it was like a 60 % of an iPhone moment. It was like, oh my God. Like, this is like, this is nuts.

26:21Like voice recognition actually works finally. Like it, it didn't work a year ago and now it works. Um, and so James and I got really excited about that and we realized there was a developer platform on Alexa. And so we were like, why don't we take all these chat ideas we have for, uh, for Facebook messenger and start just making them voice versions and see if that's fun. Right. And so we did as a weekend hack project. I think the first game we launched was a spelling bee game, like the Scripps National Spelling Bee, which was a terrible idea because they were not very good at recognizing individual letters.

26:56I was going to say, I feel like it needs to have a very high degree of accuracy for that to work. Yeah. Yeah. Like B, C, D, P, T are all pretty hard for that thing to recognize, especially in 2015 or I guess it was 17 at the time. um and um so that was a bad idea but the second idea we launched was a storytelling game um kind of like a little bit of a heritage from our pet app but um it was a choose your own adventure interactive fiction medieval times kind of game called yes sire where you would like be the the lord of of a medieval castle and you would make decisions um and it was all based on yes no answers because we knew those would work like 100 of the time and we wrote you know we wrote all the content you know my brother actually helped out on some of the content because he was a aspiring screenwriter and that kind of was like in so far as we had ever had a hit that was like a hit like we got 10 000 users in like the first month or something on that with like no marketing or anything and we were like oh wow okay this is like the most successful thing we've ever done like we're like we have to do this we're like like uh like immediately like we are Alexa developers We're like, all right, it happens, right?

28:09And there probably weren't that many back then. So you're sort of early mover to a new platform. Right, exactly. Which is one of those things that startups in DCs always say, like, oh, you want to be first mover to a new platform. But let me just say, it sucks. It's not fun. Being an early mover to a new platform that no one believes in is a relentless process of being told that what you're working on is dumb. I mean, it is every investor thinks you're dumb. Every other founder thinks you're dumb. Like, you know, even the platform is sort of suspicious of like why people are even doing this. Like Alexa itself was like, why are why are people building these apps?

28:44Like, like, like, like it's it's just a process of like every day giving having conversations with people where like they're like, yeah, like, why are you doing this? Like, like, why do you why are you building Alexa games? Like, who gives a crap about this? Right. And, you know, we were just like, I don't know. It's fun. And we have a lot of users like we don't have like a good people are going to buy these. We believe people are going to buy these things. I'm sure that's what that was going through your head. Well, we were like, we were like, we think we can monetize these because, um, actually at that time we had gotten a couple investors, um, including one of them was, uh, one of the best game investors in the world, a guy named Gigi Levy Weiss.

29:23And, um, he has funded like 10 different$1 billion gaming companies. So he really is like one of the best gaming investors in the world. And he knows a lot about gaming, as you can imagine. and we would have these sessions with him where he would sit down and teach us like the ins and outs of gaming and and how to monetize games as well so we had talked to him enough where we were like you know there is kind of a formula to monetizing gaming you just have to like we just had to hold out for the platform to enable us to to make money on these games um and so we um we held out which took two years by the way uh and we built a bunch of other games and made our existing games better um and launched a few pretty successful concepts including like a music trivia game called song quiz which is like name that tune um a number of other storytelling games uh that did pretty well um we launched kind of like a couple like game show knockoffs like we launched a price is right one called price tag um which was you know guessing guessing prices um and a couple other like pretty crazy ideas but we had a i had a suite of games by like 2019 that we thought, okay, at least if any of these can monetize, like we'll have a real company basically.

30:35And, and then also screen devices started to launch for echoes, which was a big deal. Like, you know, you had a five to 10 inch screen on your Alexa. And so that let us make more immersive experiences because then you don't have to say literally everything out loud for the user to use the products. And, you know, simultaneously the whole kind of AI stack got better. I mean, speech recognition kept getting better. Natural language processing kept getting better. synthesis which we haven't really talked about but creating ai voices essentially um got better and better um where we could have characters in our games um and we were just getting better at designing games and figuring out how voice user interfaces worked even so um so we held out until 2019 started making a little bit of money in late 2019 and we're like oh my gosh like this this could be it like doing what like what are you selling with these are you selling levels you know song packs yeah we experimented with a few different things we experimented with um like in story games we experimented with um what they call like premium choices like um you know there was this whole wave of story games on mobile where you would make like the better choice in a in an interactive story like the cooler choice like you get the better sword or you get the better like outfit or whatever um and you could sell that for money actually um like people would pay a couple bucks to like have the cooler sword you know in order to have the game be more fun and then yeah and like music we would um we experimented with like more song packs that kind of thing um we eventually sort of figured out like a subscription business model which took a little bit where we were like giving you lots of additional stuff for like one monthly subscription um and that ended up being like where we still are today uh that ended up working pretty well because the games are pretty content driven.

32:28So like more stories is good when you're playing a story game and like more trivia questions is good when you're playing a trivia game. So like people were sort of conceptually on board with a subscription, our customers. So that is kind of, you know, how we got to where we are today, just building that subscription business and making those games better. And I guess, you know, by being on these platforms from day one and these platforms have gotten pretty big now. I mean, like I feel like most people I know have one of these speakers, whether it's an Alexa or Google home, um, you have a pretty big footprint now of users.

33:01Yeah. I mean, we have, we have many millions of monthly users. Um, and I mean, I think lifetime we have like over 30 or 40 million, like people who've tried one of our games or something. So you have like almost like 10 % of Americans have like played one of my games at some point, which is like pretty mind-blowing. Um, that is crazy. Yeah. Yeah. Song quiz is a major hit by the way. And it's not just because I know the founder. Um, no, my kids love it. Seriously. Thank you. Yeah. Awesome. Yeah. And, and, you know, in the sort of fast forward in the last couple of years, we've been able to do some game show, some real game shows.

33:33So we actually run the Jeopardy game, Wheel of Fortune and a few other ones. So those are also quite popular. Our storytelling games are still pretty popular. So, and to your point, Alexa and Google, I think are in like 45 % of US households or something coming up on 50. I mean, Alexa, just Alexa, I think, has like 200 plus million devices sold at this point in use. So they're pretty big. I mean, they're not the iPhone again, but they're pretty big. And I understand that you're also now starting to, you mentioned that you're on screen devices like the Alexa show and things like this. But now you're actually migrating to televisions, connected TVs, LG devices, Samsung devices.

34:13This is pretty fascinating to me, actually, because, you know, going back to the video game stuff that we talked about in the beginning, N64, PS2. I mean, from that era of gaming and this, you know, this still exists today, you have to buy these consoles. But the interesting thing about these connected TVs is like they're effectively gaming platforms now. And you're now providing content to these gaming platforms that people have by default when they buy a TV. Yeah. I mean, the amazing thing is basically every TV is a computer now, right? Right. And conveniently for me, almost like the majority of TVs are now voice powered computers, actually.

34:49Like almost, you know, the average TV in the United States has a voice remote where with a microphone. And I think 90 % of new TVs have microphones now. So kind of like this like backdoor technology shift has happened, which is that like every TV is essentially a giant Echo Show device like in every household in America. And with an app store. Yeah, with an app store, with like with the CPU, GPU. Like, I mean, it's like it's an actual computer. It's not a very good computer, but it's like it's good enough like to run whatever you need to run. And basically no one's really figured out how to like do anything other than play video on these things really like at scale.

35:28And even some of the gaming stuff that's happening there is what we call like cloud stream gaming, which is essentially like playing a video of a game running on someone else's computer and then like getting the latency short enough that it seems like you're playing a game. Oh, interesting. Yeah. The problem is that the CPUs in these TVs are pretty cheap. They're pretty small, right? So they can't actually run like a PlayStation 5 game. They just don't have the power. but you can run a playstation 5 game on an underground box an underground computer it's in virginia which is what we would call a server right and um and then just stream a video of that game over the air to the person playing the game and then when they do something on their controller it sends that signal all the way out to virginia and the thing happens on the game playing in virginia and then you get video streamed back of what you just happened in virginia and i say Virginia because that's where AWS's big server farm is but but it could be Oregon whatever it's not that many places um and amazingly we have fast enough communication systems that this like kind of works like it's still you know latency is a big issue for cloud stream games like a big issue especially with things that are like supposed to be super fast paced like a shooter game or something like that but it kind of works if latency isn't like life or death for you um Which is what people are doing on TVs right now.

36:52But what we're doing is something much more straightforward, which is we're essentially just running a voice game. You know, on Roku, you can play Jeopardy and it puts a big Jeopardy board up on the screen. And you can, you know, hit the voice remote and say, you know, give me U.S. Capitals for 200, Alex. And, you know, the clue comes up and you say, like, what is Sacramento? Right. And it's like playing Jeopardy. And we don't have this super heavy lift from a game, you know, know, rendering perspective. So we can actually run it like on your device or very close to your device and stream it very quickly.

37:26So that's really, really cool. You can actually play Jeopardy on your TV where you also watch Jeopardy. Yeah. How is it going? I think this is relatively new for you guys, right? Yeah. I mean, we really dug into it like nine or 10 months ago. And it's growing really fast. It's, I mean, it's by far the fastest growing part of our business. And I mean, that's not to say, you know, the sort of smart speaker business isn't growing fast because it is, but the TV thing's really blowing up. And I think like, you know, I just think we, we, we figured out how to make voice games on screen devices. And then, you know, a 50 to 80 inch screen just offers this really immersive canvas for voice games.

38:02So you can make them even better and voice as an interface is sort of intuitive and accessible to everyone. Whereas, you know, maybe if you're a gamer like us, whatever, you can figure out how to use the controller and, you know, play Elden Ring or whatever. But a lot of people can't, right? There's this huge, like what we would call like a casual gaming audience, which just wants to like, pick something up, play for fun, play with their family, like, you know, have a cool Saturday night and then like, put it away. And maybe they come back like a couple times a month. But it's not like a lifestyle.

38:31It's not eight hours a day. It's not, they don't have time to like level up on a controller, right? You know, maybe they have kids like I have a kid now, I'm not like playing a lot of controller-based video games anymore. And so we really believe there's this massive casual gaming opportunity on television that has not really been captured by anyone because all the gaming on TV that happens is on consoles and it's sort of what we would call more like hardcore gaming. And so to come back to the AI discussion, I just think that the whole voice AI stack enables this really natural form of communication, which I'm sure you've talked to lots of other folks about all these other things that they think they are going to do for people's lives.

39:10But I really think it's like the best way to control games because it's accessible to everyone from three years old, you know, all the way up to grandparents, right? And everyone knows how to talk. You and I are talking right now. I mean, like it's easy and fun and immersive, right? Yeah, it's a common interface that anyone can adopt. Through that lens, I mean, it's really clear that Volley has been an AI company all along. I think, and again, we're going to get into this, I think you're probably becoming even more of an AI company, but it really couldn't have existed without AI for the voice aspect, for the NLP aspect that we discussed early on.

39:53It's pretty fascinating how ahead of the time Volley was. But maybe not without, at least I didn't realize it. Well, I think we just really had this magical moment where we were talking to the Alexa back in 2017 and we're like, oh, this is like, when this works, this is the best way to use a computer, right? And the caveat was always the parentheses of like when this works, right? And so I think a lot of people were like, they had a bad experience with Siri or they had a bad experience with Alexa or it didn't do all the things they wanted to do. And like, those are totally valid concerns. But James and I always believed like, okay, well, if we can build games that work on these systems, like voice control is the most fun way to play games, especially like for a live multiplayer setting, you know, family game night, party game, you know, or just like, you know, in the background where you're like cooking or cleaning or whatever.

40:44Right. It's like, right. It's super fun to like yell out answers or like, or like yell at the medieval night to go do your bidding or whatever. Right. Like it's fun. It's just the way you actually interact with the world. And, and, you know, while we're all like on our little tiny, like phones in our hands all the time, like that is not the way we actually interact with the world, right? And, and I always think that technology like evolves to sort of suit human, like needs the way humans actually behave and are sort of like wired to behave. And, you know, we obviously spend every day, you know, using our hands to, you know, open doors and, and, you know, read books and, you know, cook things or whatever, but we spend a lot of our days talking and listening.

41:25And while the computers were not good at that yet, they got a lot better in the last five or six years. And I would say we're kind of turning the corner literally as we speak to them being about as good as a human at that, which is huge, obviously. Speaking of not being good enough yet, why is Siri not as good as ChatGPT? Why is Alexa and Google not as good as ChatGPT yet? If I could talk to ChatGPT on my Alexa device, I mean, think of how much better that would be. Why has that not happened yet? It feels like it should have happened eight months ago for starters these are big companies right so chat gpt launched 10 months ago right so or 11 11 months ago i don't know 10 or 11 months ago it was november late november so we're still 10 months into chat gpt so like they are big companies they don't move that quickly um amazon i think literally last week essentially announced that they're putting their own version of chat gpt into alexa um okay it's not gonna be like live for everyone tomorrow or whatever but it'll probably be live for everyone in you know six months or something right so So Amazon is putting their own version of ChatGPT into Alexa.

42:31I'm sure Google is going to replace Google Assistant with ChatGPT version of Google Assistant. I don't know what timeline they're going to do it on. Yeah, BART or Gemini or whatever it's called. It's kind of a new name. I feel like they have multiple. It's a new name by next month. You know, I don't know, whatever. Maybe it'll just be called Google Assistant, but the good version now. I don't know. And then Apple, you know, Apple always takes their time with these things, right? And I especially think Apple takes time with like web-based service initiatives, right? I mean, they're not, they are in the end, like a hardware and software company, like, you know, on a local device, not a, they're not truly like a cloud company in the way that Google and Amazon are, right?

43:13So I bet next year we'll get it, but Apple operates on one-year timescales. I think it'll probably be announced in June at their developer conference and it'll probably go live in September with the next version of the iPhone. would be my bet. I have no information on this, but that would just be my guess. I think they're struggling with the problem to give them like one modicum of credit versus chat GPT is there's a little bit of an innovators dilemma where chat GPT can just like do some really dumb stuff and it's okay. Like, because they'll fix it or it's a startup or whatever. Right. Like, you know, obviously we talk a lot about hallucination, like inventing facts, right?

43:49That's like Apple cannot conscience a Siri that's just making crap up. Right. Yeah, that's really dangerous for them, especially when their whole brand is based on trust and security. It just works, quote unquote, right? It just works. It has to work 100 % of the time. They can't have hallucination. Right. So they can't have hallucination. So that's a big problem. The other big problem is, you know, Apple and Amazon in particular and Google to some degree are the value in Siri and in Alexa is that they can do actual stuff for you, right? Like that you want to do. Not just like - Take actions. Yeah, they can take actions.

44:23They can't just, I mean, for what it's worth, like I love chat GPT, but it's like, what is it really good at? It's good at like cheating on homework and like making, you know, making up poetry and like writing your memo for you or, you know, summarizing some document or something. Right. Like, you know, or like, you know, sex bot chat or whatever is a popular use case of these chat bot interfaces. Right. So like those aren't really things that Alexa do for people, Alexa do for people or that Siri does for people. Right. Alexa turns on my lights to play Spotify. yeah play spotify does the weather does the weather it answers wikipedia questions for you you know but maybe that's because that's what it's been limited to right like totally if it had more capabilities i might use it for other things right oh yeah all i'm saying is they need the next version of these to ais to do that stuff still right right they want it to be as good as chat gpt but also when you say play taylor switch on spotify it needs to play taylor switch on spotify like you can't just like be like oh i think you're in the mood for like adele because I just hallucinated or whatever.

45:21Right. Like, you know, you're like, I can't like, and then the sort of like intersection between it has to do certain stuff, like really accurately. And, um, it has to be as creative and flexible and fun as chat GPT is, um, is a sort of intersection that I think all these companies are trying to navigate right now. And it's a little bit harder than just launching chat GPT where you can just do whatever you want and nobody cares if you just make crap up. Makes sense. So a little while ago in the conversation, you mentioned that, you know, your content, Volley's content is stories. It's new questions.

45:56It's song packs. And if you rewind to, you know, a year ago when sort of the generative AI meteor kind of struck Earth, you know, it was all about creating content, right? New content, new images, new stories, poems, as you said. And so it strikes me that the technology that generative AI affords a company like Volley is probably game changing for your company in that now you can basically have unlimited content for your games. Talk to us about this breakthrough and this realization, because I have to imagine it's completely changed the trajectory of Volley. Yeah, it's really crazy. I mean, I think we're still processing how powerful it is internally.

46:38it's almost like, I mean, again, like you said, it's like a meteor a little bit. It's like, whoa, okay, we need to like rethink the way we do almost everything around content, right? Yeah. And I think there's like, sort of three levels of how exciting this generative content thing is for a company like ours, that's that's really driven by content in a lot of ways, right? One is the sort of like, simple, obvious thing, which is like, okay, instead of writing prices, right descriptions, now we're going to have chat GPT writer prices, right descriptions, right? And we're still going to have a human review them because we want them to be accurate.

47:12And we want descriptions of what? Sorry. Like, it's like, um, okay. You, uh, you have a water bottle right off Amazon and you write a silly description about like this water bottle will change your day, especially if you're thirsty, like, you know, whatever, like, uh, something they might say about the bottle on, right? Like it's orange for no reason other than that was what was left on Amazon. You know, whatever. Ha ha ha. Right. A human would normally write that for you. A human being, a human, you know, has written all this stuff. I mean, I wrote some of them in the early days. Um, James did as well.

47:38and the crimson experience yeah exactly right really you know using my journalism uh background um and and now you're like okay well i can have chat gpg write those basically and then we can have a human like review them right we can you know because you want to make sure they're right you want to make sure they're funny you want to edit them there is still value in like editing right um in many cases um so that's like phase one which is like oh cool like this is enhancing the creativity of our content team for the sort of really obvious applications. That's like super exciting and I think very valuable, but maybe like only, you know, a small step forward from where we are today.

48:16I think like the next level up, like level two is essentially personalized content for every customer, right? So let's, you know, to use the price is right example, it's like we write a description in real time for Mike based on like what he, we think he likes or what he thinks is funny. And we like training on his past use, right? Or we like AB test, we AB test a hundred different descriptions, the same item, you know, generated in real time. And then we, you know, edit them down to the top five. And then we figure out what player profiles really like that certain type of item description and make them likely to continue playing or give us a good review or eventually like pay for something, right?

49:00And so there's this idea of like, instead of there being one item description for the water bottle, there is an item description of the water bottle for every single person who's placed the game, right? And that is expensive today with GPT, you know, running every single one of those queries differently for each person. But I think in three to five years, that's like obviously going to be the case. Like maybe it won't be 100 % personalized to every single person, but you will be running like essentially hundreds, if not thousands of simultaneous tests on, um, on what, you know, what we'd call copy, right.

49:31The writing and the content in the game. Um, and they're already about to do this on ads, like on Facebook and Google basically. Right. Because like anytime there's like money involved, you can really get people motivated pretty quickly and like better, better ad descriptions. Optimization. Yeah. Yeah. Optimization, better ad descriptions are worth a lot to Amazon and, and sorry, to Facebook and Google and Amazon and all those folks. So, um, that'll be sort of phase one of like massive personalization of ad copyright. Um, and I think, yeah, there's an interesting question as to whether like all content, like five years from now is personalized, essentially like everything you see on your TikTok feed, everything you see on your Twitter, everything you see on like Instagram, you know, probably not at the level of every TV show being personalized to you in five years, but maybe in 10 years, it is like, like, will there be the concept of like a single piece of content anymore?

50:22Or will we all get custom like movies and TV shows? And like, you know, like, I don't know, that's a little crazier idea. That might be a little farther out. But you could see a world in which every piece of entertainment that you see is personalized to you, right? And then I think the last level for volley is sort of an evolution of that personalization, which is conversation with AI characters, right? And it's that, you know, we built these games where you're a knight running a medieval castle, right? But in the end, we had to write everything in the game. And even when you interacted with a character like the Wicked Witch, you know, it would be like a screenplay, right?

51:02You would be interacting with her and maybe there were a few different forks you could go down. But like in the end, you were still kind of on rails, like on a train track, right? And now, you know, we can we can prompt GPT to imagine a Wicked Witch character in a storytelling experience. And you can just talk to the Wicked Witch and and, you know, you can ask her any question you could possibly think of and she will respond in the character of the Wicked Witch. Right. And, and we can give her like goals. We can say like, okay, like your goal is to like, make sure the customer understands that they need to go get this potion to like pass the next level or something.

51:41Right. Like, and so she'll keep like trying to bring the conversation back to like, ah, and you want to talk about potions? Like, you know, like, like, like, even if you're talking about something else. Right. Um, and that is sort of like, I mean, it's a variation on personalized content but i think setting apart the concept of characters of imaginary you know people or or animals or whatever that you can just chat with and ask essentially any question is also like completely life-changing for gaming and entertainment right um i mean it's hard to imagine that in gaming you know we call we have this concept of non-player characters so npcs like the things you would run into in a game that you can you can go on a mission with or talk to or get something from whatever it's hard to imagine like five years from now that every npc and like every game isn't essentially some you know there's some llm powering that because like there's a company dedicated to this yeah i know i know you guys yeah you guys invested in one of these companies building the tools there and like you know yeah maybe they're like the picks and shovels company there and we're like the we're the we're trying to dig for gold you know um so of these three levels and i guess level one being like creating copy around in-game out game elements level two being sort of personalized experiences level three being sort of uh in-game characters that are intelligent which of these things are you doing today and to sort of what degree i would say level one we're like pretty aggressively r &ding right now like where we're you know like i don't know if we have any copy in production that's that's ai generated but like i think we probably will in the next two to three you know one to two months let's say whatever i don't want to like engineering timelines always slip and everything but like yeah let's let's say before the end of this year we will almost certainly have like ai generated copy in production in a number of titles so that's pretty cool um level two the personalized one is sort of interesting because we're like thinking about well first of all it's really expensive if you really do it at scale right so you have to like come up with some sort of middle ground where you're like you can personalize content occasionally to really provide the user some like added benefit but um but you can't really do it all the time yet because it's too expensive with with how much these llms cost so that's one where we are we're experimenting with letting like users like generate content within an experience upon request that is personalized like one idea we have is like letting users generate trivia questions in a category of their choice right and that basically works and so I think we can do that we just need to get the UI, UX in the right place and make sure that the customer is getting real value out of some sort of user-generated content that's mediated by an AI so I won't guarantee that will happen this year but I would be shocked if we don't have that live in the next 6-12 months in some fashion and then characters we're definitely doing and I would expect to have live by the end of the year as well So, um, Oh, wow.

54:41So that you're okay. When you said level one, level two, level three, I assume level three is like furthest in the future, but it sounds like I wasn't saying the level of difficulty necessarily, I guess I was saying the level of like complexity of the interaction, I guess. And I think of characters as the most complex interaction, but it's also like in some ways, one of the easiest things to do because you don't have to structure the output that much. Like it just has to act like the character. Right. Right. I mean, we have some issues with it that I think companies like inworld are trying to solve for example like you know these things have issues tracking just like pieces of data that can't change like like like like if the user has like you know in a virtual pet game if the user has like five kibbles for for whatever reason these language models just will keep changing it to like four to six they're bad with numbers yeah they're really bad i've seen it like on chat gpt right like it very assuredly gives me the answer to a math problem and then when i tell it it's wrong it sort of assuredly corrects itself to another wrong answer.

55:36Which is pretty funny. Yeah, yeah. So we'll be like, the customer has five kibbles and they'll be like, cool, you have four kibbles left. And you're like, why did you do that? You're like, what? Like, they're not, I don't know. Anyway, it's not that level three is harder. It's just that it's more, it's a more complex customer interaction. Like in some ways, level two is the hardest because that expense around personalization is really high. What about just like, is this part of level one? Just say like, writing new Jeopardy questions. Yeah, I would say that's level one. Yeah, just writing new Jeopardy questions.

56:06And you're not doing stuff like that yet, just like creating content sort of generatively. We are doing that, not in Jeopardy. Jeopardy, you have to, you know, they have a very strong brand and IP. Right, right, right. Sorry, bad example. Yeah, that seems like the most, I'm sorry to cut you off. To me, that sounds like the biggest opportunity for efficiency gains, right? It is. You know, just taking like human powered writing and just handing that over to the AI. Really powerful. um i think in the end you end up doing a fair amount of editing because like you know we have i don't know justifiably high standards on the content that goes out into our games and so you know you end up with efficiency gains for sure but you don't end up with like ludicrous efficiency gains because you do kind of have to edit every piece and editing every piece takes some time so you know it's like a nice like 25 to 50 efficiency gain or something but it's not like a 10x or or something.

56:59So what do you end up editing? Like, what do you, is it making mistakes? Is it like the tone? What is it? Yeah. I mean, mistakes, tone, sometimes you can have trouble getting these LLMs to be sort of like edgy and funny. Like it depends if you prompt while you can, but I'm sure you've experienced with, with GPT that it's, it's kind of just like banal and boring sometimes. Like you're like, just like, I've tried to get it to write screenplays for me. I kept trying to get write me like a sequel to goodfellas and i just can't i couldn't really get it to do it you know it just keeps like it keeps turning the story back to this like very positive like you know oh and they were all friends in the end interaction or whatever like and you're like no like goodfellas is about like betrayal and murder and drugs and like you know it's like you know why were you doing this by the way was this just for fun like one day i'm like hey yeah this was just for fun yeah i don't know i think james and i were just like brainstorming about like oh what can it do and we're like screenplay and i was and james had mentioned to me that it was like having a lot of trouble being like dark and edgy and so i was like all right let's do like let's do a mafia movie right you know so um i wonder if that's intentional on open ai's part like they want to make it safe you know i think i think it it definitely is for safety reasons because i think everyone got pretty freaked out by like the the whole microsoft bing going crazy debacle and everything that was wild so now it's very like it's you know it's just a little bit boring it's a little bit safe it's a little bit like banal.

58:25I don't know. It just doesn't, it doesn't feel super original sometimes and it isn't obviously. But I think that'll improve over time. So you just have to edit stuff for that. How is this all impacting sort of your planning going forward around, you know, game production or new platforms or budgeting or hiring? Like, has this really sort of kind of infected your entire company and strategy moving forward? Yeah, definitely. It has infected our entire strategy moving forward. I still think like any company you're like well we had stuff we wanted to get done this year and we're still going to try to get that done like you know before gpt like hit you know hit us like a meteor i mean i think for almost any new concept we have going forward it's going to be completely um inflected by these tools right i just think that any new game or new mode in a new game or you know especially around storytelling character driven experiences um you know you just can't not like work with these tools and be excited about them.

59:22And we have we have a special internal team, you know, the task force that's just building, you know, new titles around these tools, or that's, you know, working to put these tools inside our existing titles in some fashion or another. So yeah, I mean, it's totally uprooted a lot of our planning for what products we're capable of building for next year. So, you know, and as I said, for the end of this year, in some cases, so yeah, I mean, I think it's like, it's, I mean, it's life changing, right? I mean, we haven't even talked about speech synthesis, which I think is particularly relevant for us.

59:57Right. I think that like, that's, that's the other piece beyond just, you know, language content. Um, we can now create voices that, you know, you can record your voice into a Roku microphone and with like 30 to 60 seconds of audio, we can synthesize a mic into a game. And, um, and, you know, there's like obvious easy applications of that like we could make you the host of jeopardy or whatever or we could like make you a character in an rpg game and like you know when you get into battle it's like talking like you or your friend goes home and plays against you and it and it talks like you right so um the idea that we can synthesize anyone's voice with fairly limited audio i think is like maybe not relevant to everyone else in in the ai space but is like really exciting for volley.

1:00:45Uh, and so I think that that's another one that's like totally, um, infecting all of our development plans because it's just a trip to like upload a minute of your voice and then have a synthesized version and fit like saying random stuff. So I've noticed that a lot of these TVs now also have cameras on them. And I wonder, do you, do you think about video capture as well? Yeah. I mean, that's one where I wish more TVs had cameras on them. Um, because I think we have a lot of great ideas around that, but that, I mean, I mentioned that like the majority of us smart TVs have microphones. I think it's still like under 10 % of smart TVs have cameras.

1:01:23Like, I don't know. It's, it's just not, it's not big enough yet. Um, but my hope is like three to five years from now it will be, or that people will start like the Apple TV will add a camera or something, or, or people will start using their phones as like FaceTime cameras on the TV because yeah, I think there's a lot of cool stuff you can do. I think like we haven't talked about AR, you know, AR, augmented reality stuff that much, but voice controlled augmented reality, I think is a really interesting canvas where you can say stuff like, you know, make a cat like dance across the screen, like make it have, you know, rainbow tail, like, you know, turn me into a, you know, an astronaut, you know, put me next to the moon, right?

1:02:05Like you can imagine that voice would be a really cool interface for, for augmented reality experiences. And I think that, I think that Apple actually mentioned in their vision pro demo that voice is going to be a key way you control the, their AR VR headset. So I'm pretty optimistic about voice control combined with generative tools combined with, with augmented reality as well. Totally. Again, like playing Jeopardy in AR would be sick. Like, I mean, it'd be amazing. I mean, or, you know, being in an RPG game and like walking into that, the wicked witch and like she looks like a witch and she talks like a witch.

1:02:40And it's like you're in the world with the witch. Right. I mean, that's going to be nuts. Shifting gears a little bit. You, in addition to Volley, have have become somewhat at the center of of the AI ecosystem in San Francisco. You and the journalist Eric Newcomer, who you mentioned earlier, put on this conference called Cerebral Valley. I think you're doing it again. We're doing our second one. Yeah. November. Talk a little bit about that. What's that been like as a project? What's it been like to be near all these awesome products and innovators happening right now? What does that feel like?

1:03:12Yeah. I mean, it's been really fun. I mean, so we, our office is in Hayes Valley. I, I'm, I'm sitting in Hayes Valley right now as we speak. San Francisco. San Francisco. Yeah. And there was like this meme on Twitter in like January from an investor called Amber Yang, where she said she was going around to all these hacker houses in Hayes Valley. And it was where all the, all the AI hackers were working. And she's like, oh, people should start calling it like cerebral Valley. There's so many AI startups here. And it was one of those things where it was like such a silly name that it would like sort of went through the looking glass and like became sort of a funny meme name.

1:03:49Like everyone sort of leaned into it, myself included on Twitter, just being like, yeah, like it's great to be like sipping my coffee in the heart of Cerebral Valley today. Right. Because it's like such a silly name. I mean, it's just so it's so nerdy. Right. Cerebral Valley. And so then Eric, yeah, James, Eric and I all sort of got together and we were talking about doing a conference because we have this giant office in Hayes Valley. and, you know, pitching out different ideas. And we were like, hey, like, we need to do an AI conference like right now. Like, this is it. Like, this is this is happening.

1:04:20You can just tell it is like the hottest, you know, hottest thing that has happened in Silicon Valley since the iPhone or whatever. Right. And so in like two months, we put together this whole conference. You know, Eric recruited, I think, you know, almost all the speakers, which were amazing. We got, you know, the CEOs of like Hugging Face and Replit and Stability and Databricks. And, you know, our team at Volley like ran the whole event. We held it in our office. We like designed all the logos and the swag and like put up the stage and all this stuff. So we put this event together in two months and it was a huge hit because like really no one had like nailed the AI conference in March.

1:04:59like you know which sounds like we were too too late already but we were we were kind of just perfectly timing it and um and actually like databricks bought a company that they met at the after party for wow 1.3 billion dollars oh that oh that happened at your conference yes wow they met at the after party yeah they met at the after party of cerebral valley uh one and uh ended up buying company for 1.3 billion dollars mosaic ml and and so we were like wow this is pretty wild. Like we really nailed this. And, and so I, you know, Eric, I think really wanted to do another one this year. Cause he's like, ah, so much has happened this year.

1:05:37And we're like, yeah, sure. Let's do it again. So, um, uh, so this time we're doing it again in mid November, we have, you know, Reid Hoffman, we have Mustafa Suleiman. Um, one person I'm excited about is I, I got Chris Latner to come who, um, is a, as a nerd famous person who invented the language Swift, uh, that runs on all iPhones and is now, um, running a big AI startup called modular where he's trying to reinvent the language that people use to, um, you know, program for AI, which is pretty exciting. Um, and so, yeah, it's going to be a really fun conference to be a killer lineup. Um, we have a lot bigger space this time.

1:06:10Last time it was really strictly limited to 200 people. This time it should be like twice the size. So where is it? It's going to be at the SF jazz center, which is literally like a block away. Um, and yeah, I mean, I think that like, you know, we're, we're obviously not a foundational model company, but we are building on this whole AI stack. And I think we're maybe the most successful AI-powered gaming company out there today and literally have, I think, the biggest office in Cerebral Valley. So we just thought it'd be fun to lean into it and do the conference. And so far, it's been working super well.

1:06:40So if people wanted to go to this thing, is it open to the public? And how can they do that? You have to apply. Yeah, it's CerebralValleySummit.com and you can apply. And we're trying really hard. No offense to the investor side of the table, but we're trying very hard to keep it balanced towards builders and founders and not just a bunch of VCs. We had lots of great VCs last time, including from Lightspeed and many other firms. But we're trying to have a little more balance insofar as possible. But yeah, and tickets are also massively discounted for founders. So that's a big incentive as well.

1:07:17Nice. What else are you excited about these days besides obviously volley and the conference you're putting on? What else is keeping you interested? I mean, you sound really busy. I know you have a young child at home as well. Yeah, I have a young child. That's that's all my free time. I mean, yeah, I mean, I don't know, spending spending as much time before and after work with my daughter as I possibly can, because it's super fun. I know you have kids as well. So I'm really enjoying that. My wife is a professor at Berkeley, and she's doing super great. So that's really exciting. I mean, but honestly, I spend like 90 90 % certainly all my non family time working, which is I don't know, I like my jobs.

1:07:53That's good. But it's it's really all I do. makes it easy uh is is volley hiring um yeah we're hiring we're hiring um especially for back-end engineering and a couple mobile roles as well um we are uh you know one of the things we're really excited about is connecting a mobile app to our smart tv experiences and so you can you know play jeopardy on your roku and then scan a qr code download an app and then share it with your friends you can go play jeopardy against your friend you know on the go in the car or we build that social graph and it can be shared between the smart TV and the phone. So, um, yeah, that's something I'm pretty excited about as well.

1:08:31Um, so yeah, uh, those are, those are the key roles we're hiring for. Um, and of course anyone who wants to build AI games, um, I mean, we're certainly leaning that way, uh, in terms of searching out folks who have excitement around, you know, generative gaming opportunities, cause that's a big thing we're investing for going forward, obviously. And if someone wants to try one of these games, what is the quickest and easiest way to do that? And which one should we try? I mean, if you have an Alexa, I would just say Alexa play Song Quiz or Alexa play Jeopardy or Yes Sire if you want the storytelling game.

1:09:04If you have a Roku or a Fire TV, you can search for Jeopardy or Volley. Both of those should work. Yeah, I mean that's the easiest way to play those games right now. And I mean also if you just ask your Alexa device like play a game or like I'm bored you will often end up at one of our games whether intentionally or not. So So yeah, I mean, those are all pretty easy ways to do it. Max, this has been so much fun as always. Thank you so much for doing it and looking forward to having you on again sometime. Thanks so much for your time. Thank you so much for having me. It was great. Thanks so much for listening to this episode of Generative Now.

1:09:41Hope you enjoyed my conversation with Max Child. If you like what you heard, please do us a favor and rate and review this podcast. And if you like what you heard, Also subscribe and follow us on Spotify, YouTube, Apple Podcasts, or wherever you get your podcasts. I am Michael Magnano. If you want to follow me, you can find me at Magnano on all the different platforms. If you want to follow Lightspeed, you can follow us at LightspeedVP on all the different platforms. And like always, we'll be back next week with another fascinating conversation. Generative Now is a production of Lightspeed and Pod People.

1:10:21Special thanks to our production team, which made it all possible. See you next time.