Fireside Chat: How AI is Shaking up Game Development in 2024, Craig Chapple SM & Tom Pigott, Ludo.ai

[Music] [Applause] [Music] next up in our uh AI track we have a fireside chat um how how AI is shaking up game development in 2024 um so for that I would like to welcome to the stage cig Chapelle from steel media and from Ludo AI Tom pigot pigot sorry Tom pigot hey everyone how are we all doing uh thanks again thanks so much for attending the F side chat and we're

going to talk about how AI is shaking up game development in 2024 uh we'll discuss practical applications of AI tools and go production how Studios can use them right now so I'm Craig Chapel I'm the head of content the head of content at pocketgamer.biz and with me today I've got the the Ludo AI uh founder and CEO Tom pigot uh and you've been working the space for many years uh as in the AI space I think for many years but uh yeah if you could just introduce your yourself and what you're up to thanks for being here everyone my name is Tom pigot uh founder of Ludo AI I've been in the games industry for about six years and we founded Ludo four years ago uh more uh out of necessity um than otherwise we were in uh game content Creation in both VR space and in Mobile gaming uh and like a lot of Studios were having our challenges and that's when I actually met my co-founder uh who is really the technical brains behind uh

Ludo and has a PHD and Ai and we began to think about what models were available uh that could be utilized both for ourselves and ultimately what we developed uh for game studios to give an advantage When developing the the games as they exist now so you're actually in kind of the The Trenches of development and this is the tool that you kind of came up with to be super useful for that absolutely as I said yeah uh necessity is the mother of Adventure and we were we were challenged and struggling and and uh we'll talk about yeah the platform that we've developed so I guess my my first question really is like how's the world of AI changed over the last few years kind of since you found at Ludo and then how has that impacted your company and and the work I think um well certainly when we first launched our platform in January of 21 we had we had founded the company in a year prior in 2020

um the models that were available um were fairly Limited in terms of open-source models um there was uh gpt2 and so uh and the even the image generating models were not that significant or not that advanced um at that point so I think absolutely the biggest change from what we said is is when we used to come to um conferences like this it was explaining a what a was going to be you know and will it have an impact which we believe it did and um you know what would be the Practical abilities to use it um since obviously with uh open Ai and chpt uh coming out a lot of that has gone from doubt and skepticism to okay uh what impact will this have either for my job or the industry and uh so a lot of that let's call it controlling or educational aspect of um explaining AI has been taken care of and now it's really okay

this is clearly going to be here to stay uh broadly speaking not just in gaming but across what we believe many Industries um and so how do we get smart about it so it's not just the latest buzzword like metaverse or whatever yeah uh we don't believe so no we we bet early on it and built a team uh of AI exper experts um and we just happen to be focused on gaming and that's sort of how we differentiate ourselves is that we are a AI company first that has a deep vertical f focus on game development problems and challenges so where is AI and kind of specifically generative AI already used in production uh and where do you think it's proving most useful right now obviously AI has always been used by developers but in developments um well we started uh when we built the the platform at the whole entire beginning of the game creative process right and that is you know I want to start a game maybe I have an idea or Colonel of an idea or want to

take an existing game and uh make a variation of it so we with the models that uh we originally trained were're more on the ideation aspect of it which is a you know concept creation ideation the brainstorming uh gamestorming uh phase that was probably the lowest hanging fruit uh that we could utilize and and uh was also a real need um the majority of our customers are in the mobile uh in PC um uh space and so they're having to create especially in the casual and Hyper casual back several years ago having to create a great number of variations and prototypes um rapidly so that was a pretty straightforward um opportunity for us to provide a tool that did that and then subsequently we um got into uh game design uh documents um image generation as the tools became uh more robust and uh available for us to to um TR train on

so what are the benefits of kind of using those tools that using that kind of tool versus not using it like why do it what are the benefits yeah versus not using it um I guess we look at it from a perspective and and having started a number of uh companies in my career including back in the early Parts early days of the internet in the in the mid 90s um we just think that maybe unlike as you said some Bud buzzwords which I can't predict how well they will do or not but AI is is here uh impacting and and cutting across multiple Industries um already um you know it's been around for a long time but as you said I think it's been a combination really in the past um several years of of the data that's become available the computational processing uh that has become available and the the robustness of the models that are open source uh primarily that have been release that

has allowed this sort of convergence of of uh different companies that can uh work on it and apply it now so to answer your question uh in a long withed way uh from a development game developer standpoint um it's here in getting stronger every day in terms of the models and robust whether you like it or not and so we just sort of look at it from a practical standpoint it's like sort of you know I'm being somewhat fous but sort of like ignoring the internet you know you know I mean you can but um when it came about in the '90s and but uh eventually you know there's going to be an impact to you and so that's what we always when we speak with game studios uh who have questions about hey where can I apply this or um we say start small and start with you know obvious um challenges you're having and and just get smart on it you know try a bunch of different tools try a bunch of different um uh models and um and and see how it works so is it about like yeah you're constantly having to to test

these tools and make sure it fits into your I guess production pipeline rather than you can't just plug it in and it works correct yeah definitely not no it's it's uh human plus AI as we say in terms of uh the creative uh process um it's like you know what think of the impact of the personal computer U when that was uh being widely propagated you know it's it allows you to do your job in many ways uh a lot more quickly a lot more powerfully a lot more options um but did it take away you know your particular job most of the time certainly not other jobs sort of migrated towards something that they could uh use more effectively but um you know when spreadsheets came around um you know it just supercharged a lot of the work and that's what we think these tools when they're finally tuned um uh and they always keep improving because more and more models are coming out literally every month um it's the pace is that fast um so we're trying to provide a platform a

truster platform where you don't have to figure out which is the best you know you can come and and not and you also don't have to bring in an AI team to your studio which is a big concern right is well if you don't have the expertise in house how do you keep up and and so I think you have to go under the um presumption that at the end of the day um AI tools will be um very widespread and and Common Place um and that everyone will have access to the tools and the models themselves so then what what's you know then what do you do and how do you build a company out of that you Imaging like you know things are moving so fast so how do you reliably put these tools into production when there could be a major development within three months and then you kind of have to change how maybe you have to change how you do things yeah it's I mean it's it's a challenge I mean for us we're the ones who are actually trying to build the tools let alone from game studios perspective but for we're in constant touch with uh the different um

different developers of the llms and and the different models and and what we're what we're trying to do and provide for game studios is to provide an integrated platform where instead of going to seven different sites with different various AI tools um you have a you have uh one platform that offers um integrated approach in terms of how you develop your game design document your research um your marketing image Generation video generation Etc as a studio though yeah you just um you're not going to ever be on top of it the pace of research and the papers that are being produced and the models being just it's just accelerating actually my co-founder has said he's been doing this for 15 years not in gaming but in AI research and and um AI applications uh he says he's never seen the the pace of which um uh papers are being produced on breakthroughs uh around the world guess

it's just so much significant investment going on in the space that there's always something new to learn for everyone absolutely yeah it's um I wanted to go a bit deeper on there you talking a bit about ideation and some of the stuff that you do like can you go can you explain that a bit more like how that works actually for developers like what's the process for them sure so we like to say um you know start with Ludo uh if you're a game developer and most I'll to be clear U you know we presume that a lot of triaa Studios Studios have already have inous teams and and uh you know have the the bandwidth to be able to hire even their own machine learning Specialists so our focus is really been always on the the Indie Gamers Indie developers uh Studios from you know a solo um Dev to allf to you know 50 people um and so if they have a new idea or they want to start something maybe it's in a new genre uh you know we like to thank you start with Ludo you're able to come and have do some market research

understand some of the trends all of this is tuned with our AI models um understand you know even some of the Popular Mechanics or things that are that are trending and be able to ideate from that uh create a number of different variations which will lead to uh beginning to flesh out uh some of your design documents um and Ludo can help with gaps in that um and in the future will'll be able to actually uh provide proactive um uh ideas for how you would make it more successful and then generating you know all the things we've seen um in terms of images basic video and ultimately to be able to have something that gets you to production so we're we're trying to cover all aspects of pre-production which we believe certainly given this environment uh you know difficult environment that we're in um you just can't uh throw out random prototypes as easily as you could say three years ago

you've got to think through and and really plan it especially if you're working with Publishers uh so it's more important than ever I think uh to have a very solid planning process in place and we're talking about like kind of pre-production but are there any other areas that you think Ai and generative AI tools are going to be particularly useful for developers uh absolutely so I think if we talk about we're talking about what's coming or what I get firstly what's now and then what's coming uh well certainly you know and you've heard some of the other speakers today um AI tools are absolutely being used in in postlaunch um monetization optimization that's not an area we focus on but uh you know there's there pretty much all aspects of I mean the only thing we say that AI tools literally and might I don't think I'm going to be disputed here but uh can do for you it can't generate the game by itself um and that's a frequent almost daily question we get is hey I thought could generate this AI you know by pushing a button and

the and you'd have an AI generated game uh at some point will there be an AI groundup uh game engine built you know quite possibly in the in the years to come which would change things dramatically and the types of games that are would be able to be uh created would be you know probably unlike what anything we have now so it'll be interesting I think I wasn't there just was it Google or something that that was showing that they had like uh it was creating like a simple game from generative AI sort of assets I guess so and there's been Platformers that have been trained on that and that's a good example and a good question of of of uh you know you can have a very very very narrow use case of doing something like that um but let's call it a general uh AI game engine is not there yet the physics involved is is a super big challenge I think people are absolutely working on and maybe some cell startups and and probably some of the big players but um yeah I remember writing about that like over 10 years ago when I was

at develop this AI called Angelina that was creating uh platformer levels as it was fascinating um so what kind you kind of touched upon this a minute ago but what kind of companies do you think will start using these these tools is it large Publishers small startups midsize and know everyone you know is it more advantous to some than others well I think your last uh question is the one that for us makes sense is it more advantageous U again given how challenging the current environment is to actually successfully build and launch a game and get it you know heard above the noise um I think Indie Studios or smaller companies who Embrace uh these types of tools um will uh will allow you know I think was even this previous speaker mentioned you know it democratizes things a bit um it levels the playing field in a sense because for without having to create your own in-house models your own in-house uh types of tools you get the power and the robustness of of a lot of

what AAA Studios maybe have always had um and so I I do think that with user generated content um people embracing uh tools that allow them to turbocharge their whole pre-production ultimately production process um they'll be the first to benefit um you just have to presume the larger Studios have that in-house capability but it's also as everyone says it's harder to turn you know the battleship around and some of the those ones in terms of the cultural mindset you know Indies have always been the first to embrace um you know new technologies and and um and have been more Nimble in that regard uh I guess it' be remiss if we didn't talk about you know some of the legal aspects of of using generative AI so do you think it's challenging to use for Studios due to those potential like copyright infringements like how can companies watch out for that with the tools that they use yeah and I think there's everyone has their own you know perspective and opinion on it I mean for us um clearly the technology has

outraced the legal framework or the legal framework has not had a chance to uh come up we we follow it closely the bigger players the bigger uh issues but whether in the EU or in the US or elsewhere and um uh you know we're taking open source large language models um uh in terms of the copyright issues all of that is just remains to seen how that's going to be played out but um yeah it's it's you know even speaking with uh some lawyers recently it's it's just a gray area that ultimately they'll have some impact but I think I would highly doubt that this would put everything back in the bag so to speak in terms of so many companies so many and they'll just re you know reach some type of compromise but I mean we we've trained on millions of different data points uh and to say one of that one of that piece of data is is direct you know copy of something that's out there is it's you know I think it's

a challenging yeah I think there's there's a few things moving legally in that regard I think there's a court case in the US with George AR Martin about open ai's use of I guess his his books characters and books yeah and in the EU there's uh I think there was just like new can't remember what it was regulations or guidance or something around AI but it was more focused on the uh the apocalyptic aspects with a tiny bit about generative yeah yeah so I I I can't I don't I don't know if it's our place here to um delve into the the legal aspects of all of it but definitely definitely fascinating so uh does anyone have any uh questions we we've got a couple minutes yeah we got a question I really can't see anybody lights out there pretty bling hi there uh very interesting that's great thank thanks to um sorry can you give us a little more insight you you mentioned that you're just on what large l models that you're using so you said you use

and you mentioned you open source what what what what model does lud use uh well we have a number of different ones I would probably have my uh CTO and co-founder come up or talk to you later if you want the specifics on it but most of the um popular large language models for both image generation uh video generation and uh text Generations are what we've um started with and then we fine-tuned on on our some of our own models which are much smaller in terms of parameters gotcha great thanks um I would one one thing I would say in terms of um where things are going to be going um for and areas we're worked on but they're certainly everyone everyone's working on them as well um I think uh the AI tools will become much more proactive uh in terms of um like an assistant you know I talk about co-pilot and they talk about various assistants and I think think once that gets figured

out that'll be very powerful and transformational in terms of um in our own case if you came to the to the platform to the little platform and it says hey you know what can I help you work on today and and have you considered aspects of this or you know based on you know the training um you might want to supplement with uh certain Concepts mechanics Etc so I think the the uh proactive aspect of these tools will be extremely powerful and then the other obvious one will be um video and Sora has already been come out with you know opening eye with Sora and um but when it gets to the point where uh you're you're going to be able to maybe test uh various ad creatives do you know various playables potentially that's going to be super impactful in terms of uh creating either image to video video to video or text to video um uh you know if you could generate 30 different uh short videos in a matter of of a minutes that's that changes things quite a bit yeah that's yeah like we said like the

the pace of the technology what's possible is is is kind of crazy um I think that might be it we got another question yeah I got a question so you kind of touched on it there um the way I see it is Ludo can get us G from in a pre-production process straight through to basically MVP um and that's with imagery video and and pretty much everything you need game descriptors etc etc um so if you can do that can you do that with um your your adverts because in I do a lot of work in adtech could you see game uh games developers being able to use Ludo to create different creatives with lud's platform yeah great question um I guess short answer is yes I think ultimately and I think that is a big challenge adtech there is a great opportunity for Innovation uh right because of right now as for playable ads or other uh game ads um you know you're it's a very manual

process um and creating those variations whether it's brand specific or variations of a of existing game and so we would expect that the once the AI tools are stable um and trained particularly in the video you could get to a scenario where um you could create a number of Vari uh a number of playables um to test out uh to see you know from a protoy standpoint or which which ads work best so um that's a big it's an area that we're looking at but yeah adtech is a huge for games that's a huge area of opportunity definitely good question all right thank you very much I really appreciate that please give a round of applause thank [Applause] [Music] you

[Applause] [Music]