Panel: Hype Or A Real Force For Games – Where Are We With AI In 2025?

[Music] a panel now um that will take us for the next 40 minutes and we'll do some questions at the end of that the panel is hype or a real forcing game where are we with AI in 2025 so I would like to well we have uh Liam from omia guy from Charisma a Vulcan from Leia AI Vanessa from Scenario Oz who's a freelance consultant and we've got our um and Andy from fitu and I think Liam is going to be the moderator so why don't you come on up and introduce your panel and get started so round of applause for our expert panel thank you hello everyone welcome uh so we are going to be about hype real

Force for games where are we with AI in 2025 we have a great panel with people with different perspectives expertise going to tell us a lot about what is happening where we're going with AI in the games industry we'll let everyone very quickly tell you a bit about themselves uh Vulcan do you want to go first sure hi I'm Vulcan co-founder CE of ler um as I think if you were here earlier you might have heard about yon speak who's IP of engineering but basically we're building an AI powered Creative Suite for game development can you hear me yes my name is O I've been in AI since 2017 spent the first five six years on more machine learning uh oriented stuff like Trum prediction marketing optimization RTV prediction boards that play games things like that last one year was with Leonardo AI doing imag generation for games and other Industries and now I'm advisor to some AI startups and some gaming companies on how to use Ai

and uh morning everyone I'm guy from Charisma AI uh charisma's been going for about 8 years now powering conversational NPCs in games and uh arguably we launched the first generative AI powered uh video game uh in the world maybe there we go hi everyone I'm Vanessa Vice President of Sales at sario I've been in games for about 10 years now U mostly on the bizdev and sales side so a non technical perspective in how I've seen AI evolve in the gaming World on yourself thanks uh hi everyone I'm Andy Carell uh I a mobile industry veteran I've been in Mobile for about 25 years I made some some of the very early mobile games including a game called space impact which the older people in the audience might remember from the Nokia 3310 uh these days I'm running a full stack growth agency uh helping apps and games to uh to reach their full potential uh working with uh uh yeah

companies like supercell wuga Adobe and the NBA amongst others uh yeah we're called feature thanks okay great thanks everyone and my name is Liam Dean I'm a principal analyst covering games Tech at omia H just this month we ran a survey that found that more than 30% of game developers are using AI right now but what let dig a bit into what that means exactly I mean can you tell us bokan kind of from your perspective Ive some kind of concrete use cases where developers are already seeing value from AI yeah absolutely um so I think we are we we thought hey like AI would be great for game development and over time what we seeing seeing is it's not just game development but it's like marketing and live Ops and all sorts of content creation needs can be met with AI but you know we've been doing this for over you know a year and a half and in the beginning you know the the base models weren't very good the results weren't very good it required a lot of customizations and you know attention to

detail to get some good results but the promise was there and I think the promis has kind of played out significantly with the advancements in some of these base models like say flux that came out recently especially with the recent base models around video and 3D Generation Now what we're seeing is you know we've done some case studies with a lot of our customers they're seeing for example like for a puzzle game one level used to be generated in say 15 days by a team of four now it takes three days that's like a 5x Improvement in productivity just for you know inapp you know a game development and you know in-game assets and that's not even talking about how much faster they can ship say a liveops campaign for Halloween or Christmas or how many more experiments they can do uh with different marketing assets and this is still very early days like there's so much more automation yet to be done yeah of course you're you're in a similar space as well Vanessa similar similar to Vulcan so uh we have seen really customers taking things not just for ideation but direct to production so

the assets that they're producing they're not just using them for testing or for ideation these are 80% hit rate and going straight to Market so it's not even testing anymore it's not new it's live it's out there and some of the games that you're playing today likely have some of the technology in scenario involved we're working with inog games we're working with unity directly to be embedded so it's not hype it's now and it's real and and these these tests aren't tests anymore they are production ready assets so they're the best and K you're in a little bit of a different space um I mean do you have kind of a different perspective on the way that developers are making use of the the tools that you're providing well I think um I'm quite Punchy on this and I think you know out of your uh report which said I think it was 32% of people were using uh AI I think that's 68% of people companies which are in uh in trouble and you know we all know that the uh the games industry has challenges around content generation uh the stats around content

generation are you know it's uh 90% growth in content generation in a game since 2017 and yet actually only uh two times um the amount of revenue is generated out of that so there's a problem around it and my view is that you should be using this both in terms of the um uh the production efficiencies you're going to get what we' talked about what we know but if you look then at uh Matthew Bull's recent report which you all should uh you know part of what he's saying is that this is looking at creating a new genre creating new Innovations creating something we couldn't do before and to that point I think you know and Andy will have experience of this this to me is like the same inflection point we had with mobile you know when mobile came into the games industry there were people who adopted it the vast majority didn't and that generated an an enormous opportunity for people and I think we're going to see the same thing with AI so my question almost like I don't see 32% as a good number I see it as a I'm looking at 68% of the responders who didn't say saying guys you got to catch

up otherwise you'll fall behind when you said AI is it like all just generative AI or machine learning H generative AI specifically yes so probably 100% almost I would assume exactly yeah yeah um I mean what would be your perspective on this Sandy yeah would just chime in um you know to to uh guys point about uh inflection points uh we see on the marketing side there's there's already a huge Paradigm Shift um you know which is available to those early adopters like we've we've built some tech um which started out just as kind of like a Labs project I mean we we do a lot of tech enabled um service stuff we're building Tech to kind of just improve our our services but we also spun out some stuff which which a bit more experimental and we've seen we saw from that like wow this is like super impactful so what it what it's basically doing is in it's really turning the whole kind of experimental process on its head and inad in of kind

of having a marginal cost which is pretty high for running every additional experiment um you know to drive for example conversion rate conversion rate increases in in the App Stores uh from impression to download um we're seeing with generative AI Automation and reinforcement learning um we can generate almost an infinite number of experiments run them parallelized running up up to 500 experiments simultaneously which is way more than any ASO team can can do manually um and it's iterating very quickly towards like quick results well so it's the Paradigm Shift there is right really that it's it's reducing the marginal cost of each additional experiment to to to zero um yeah so it's a yeah it's it's a huge shift so I guess an interesting question that I might throw out to all of you is see we are seeing real productivity gains I mean but is this is this something that's really going to help the economics the industry or is it just going to create a a race where you know everyone is now producing more and you know we have haven't gone any more

profitable than we were or or does do we have something that kind of helps us break that cycle in some way maybe I can quickly jump in I think uh just like with any Tech Revolution there's going to be a transition but what I believe is going to happen is AI is going to automate a lot of the repetitive kind of non-creative aspects of game development but then that unlocks a lot of creativity at the higher end like ways that we can't even imagine right now like say we're just this morning talking about new genres imagine like a new AI powered genre just like strategy games or simulation games like a new type of game experience is possible with AI it's not possible today but it will definitely be possible in the future and these are the types of experiences I think that game developers can create in a way that actually unlocks more value like ultimately you know we're talking about things like LTV and CAC and user acquisition but the ultimate goal is to create a game that users enjoy like if they enjoy it then the rest kind of comes along with that and I think that is going to be possible in unique ways with AI that aren't really possible today with you know older technology something that I of my friends brought up with me that I

thought was really interesting is you know is if AI is able to reduce the cost of Studios from production to release would that eliminate not eliminate let's not be crazy but would that reduce the amount of ads required in games because for them as a player if the company doesn't have so many overheads maybe they don't need to rely on ads for revenue and that would make the player experience better um which was a really interesting perspective that I hadn't thought of um in terms of how AI might help the consumer um enjoy games more as well and I'll tell you a specific use case which I quite like which is that if you're building a game imagine you're building a small village and you're building a bar in that Village that's going to cost you a certain amount of money it's going to take a certain amount of time and what the player is probably going to do is run through that Village stop up the bar and then disappear off total game play time probably about four or five minutes or you start to put conversational conversational gameplay into it where they're sitting down at the bar they're shooting the breeze they're negotiating they're playing a poker game that time is now about an hour hour that's an hour so an incremental

you've just gained about 55 minutes of game time with negligible cost deeper engagement all through conversational AI with NPCs or whatever it may be so by changing the game playay style a little bit putting these into the thought process you're uh you're increasing your game time and your you know emotional interaction as well so that to me is something you couldn't do before because you would you technically speaking you'd be doing sort of JavaScript or something but the new uh natural language processing technologies that been around for a while now uh are I think changing the game in that sense literally sorry to jump in twice this quick example maybe to kind of paint a clearer picture and all in in your minds there's a recent game super simple where you have to convince an AI to let you in through a door right and it's it's basically like a super simple llm and it's with this system prompt but imagine that's the type of experience that's possible where you have to convince a machine to do something and if you do

that via conversation via whatever you need to do then unlocks say a level or a door or a chest so that's the type of experience that's possible now that wasn't possible before so I guess we kind of have in a way there are almost two conversations about Ai and games one is how is this going to let you do tasks you already do more efficiently and the other is how is it going to allow you to do something you could never have done before I mean to what extent do you you know which of those is the most important strand I mean from where we stand right now of course we can always say the big thing in the future is going to be you know what whatever um future you can imagine but like is that where we should be focusing our energy right now on a completely different vision for games or is this the time to focus on kind of just uh more of the brass tax applications well just I think I do think we're at an inflection point and

it's and so I think we need it you know all the again reading Matthew Bal I don't want to go back to it too much but but it was quite influential when I was reading every slide of Matthew Ball's new report it's saying it's and it's yeah it's just 200 or so it's it's saying we have an issue and actually a lot of the issue is around the amount of money that we put into games it's the game players are moving you know towards video there are a lot of really important shifts H happening and it needs to change and something fundamental needs to change AI is a fundamental technology it's not a nice to have it's not some sort of thing that's come in and then going to disappear next year it's fundamental so if you look at the there is a fundamental need for the games industry to change and there is fundamental technology that's available to the games industry to help it do it so my sense is is there's a have to not a nice to have uh yeah I think uh I think the question you know is is you know which which should you should it focus on like I I think you can do both simultaneously

like what we're seeing already on the marketing and growth side of things is that AI technology can bring double digit increases in efficiency of marketing right that you can either say okay that means we can spend you know 20% more on marketing get you know with and get more users into the game or you can say well we we we get this efficiency uh and we'll invest that in better gameplay experiences and and invest in in building out you know sort of more um more engaging yeah experiences and and capabilities within the game development itself I I don't think it's an either or right so you can choose where for your particular company where you want to to focus that can I pose a question to the audience just a quick show of hands for those who are creating games how many of you are using AI now if any all right that's good that's good that's positive I think that shows the sentiment that I mean given the question of the title of the talk you know does it feel like hype or is it is it real I

mean I mean a follow-up question you're using it are you seeing the benefits that were promised to you when you started to use it yes show of hands yeah significantly fewer hands I think interesting very very interesting and I guess you know I have only been with the company for a week scenario and to read through the case studies of what the company has been able to accomplish so far and the sentiments of our current customers who continue to come back and to utilize the product and to see such benefit um my urge to you is to keep exploring because there are definitely there's definitely something out there for everyone that can enhance your workflow cut your production time and save costs and every AI is in a blanket um example for every company but there's definitely something out there that can help so if you're not liking what you're using now um keep exploring because it exists um but I mean okay so let's talk about cost cutting then I mean realistically

when we talk about cutting costs are we talking about cutting jobs you know this is obviously something a lot of people are concerned about now a lot of developers are concerned about are they right to be yeah let me jump it I think when a game Studio thinks about hey like here's cost savings here or like productivity gains here their initial reaction is not to cut cost but see to see if it can increase Revenue because I think University whenever we speak to like anyone small large working with some of the largest Studios out there their intuition is not to say oh I'm going to save X you know a million dollars a year by cutting my Workforce no they're going to say what we're seeing repeatedly is I have this amazing say art team I have this amazing development team how can I maximize the revenue I'm getting out of them how can I maximize the number of games I'm shipping how can I ship them fast imagine like one Studio makes a billion dollars a year if they ship a game one month faster that values $100 million to them that Revenue is a lot higher than

whatever cost that they have that they can cut with AI so they will always be thinking how do I maximize Revenue before they can think how do I cut costs and just like with any kind of tech revolution in the past what usually happens is the productivity gains result in more economic activity not less which means overall like there's more Revenue to be made more money to be made in games so therefore if you're focusing on cost cutting and your competitors are focusing on increasing Revenue the cost cutting ones might not necessarily be the the top ones remaining in a few years and would that be your experience as well Al in the companies you work with that they focus more on Revenue maximization than cost cutting I agree like I haven't heard the phrase of like cutting Workforce yet yeah yeah I've got a a concrete example from from mobile games like what we see is typically like a a large studio might might develop 10 to 12 games in a year um and maybe only two or three of those would get out of soft launch right now what we're seeing on the marketing side like deploying like

automated um App Store optimization automated uh bidding optimization for Apple search ads for example or the UA channels um that could mean the difference that's bring in efficiency that could get two or three extra games out of soft launch and into full production into scaling where they're suddenly making a significant amount of Revenue um which indeed like you know that so that that saving there is like that's there's a cost saving there which actually nothing to do with cutting jobs it's actually like bringing bringing a lot more money into the into the company and actually keeping game production teams more busy because they're they're actually going to be them running those games for you know for the next 6 to 12 months through their life cycle yeah and for what it worth this is consistent with the developer surveying we've done at on that unfortunately a lot of the developers we survey have been affected by layoffs at their companies uh over the last year or so but only 1% of those have been connected to AI so that that's not really something that people are seeing that job losses because people are being replaced by AI so but it it's

interesting that so much of the conversation is around cost cutting when when we actually get into it uh you know it doesn't seem to be reflected so much in the you know in the actual practice of application of AI in games I mean what do you think that's I think it's a scapegoat too like imagine you you said like only 1% are AI related I think people might be job cutting for various reasons for everything like in the economy outside of AI and AI is an easy skep Cod right now right well I think this is where we have a an answer we have a solution which is that to a point slightly Liam which is we can do something new and why would you do something new well you do something new because at the moment uh only 6% of game playay time is for new games 6% so if you're in the in the space of launching a new game only 6% of that gameplay time is going to go to

your everything else is going to go to existing franchises so you got to have cut through You' got to do something new and that to me is where we start to be able to bring new technology into the space to innovate with the game play and do something that's going to get cut through and then that 6% can go up by the way that's high that was 23 22 21 and 24 were were lower than that so if we're looking to innovate actually I do think that you know things like we did the game this generative game I mentioned the beginning called Pro uh project Electric Sheep it's on Steam so 3D Dream simulator you tell it your dream and we generate that dream so we generate all the assets all the buildings the terrain the soundscape the characters over 50 characters they're all talking about that dream it's a sort of casual casual game but the point is that it's this combination of generative and procedural AI as we move in uh that that that allowed us to design that from the ground up and then build it and that has an element to cut through which is which is fun it's experimentation yeah I that's interesting can we focus in on

these sort of more runtime use cases of AI CU I think we've been speaking now you know mostly about AI in production uh but I know you've been doing a lot of work kind of implementing AI at runtime I mean this is some of the most ex things people are most excited about AI but there also a lot of challenges there in terms of the the compute cost latency I mean and how how do you see kind of um or first of all how how serious do you think these challenges are and how close do you think we are to solving them I think we're I think we we've already solved them actually because I I my view generally if you're looking at this in a production perspective there are two ways to solve it either you've got a tech solution which is throw more compute or throw more money at it or or a creative one which is we'll don't do that in the first place think of a different sort of way around it so I think we are you know with with with project Electric Sheep what we did was we uh we put in a business model that allowed a you know bit of free play and then you pay for it um the latency you

know is going away as an issue for a lot of different pieces and as I said the the solution that we put in was this combination of generative and procedural and that way it's it's a it's a balance that you can do the two but the key thing is you start with a creative idea that allows you to have something that is almost infinite as a um as a gameplay concept and if you think about us sitting here we're already in an infinite space in the sense that you guys could ask ask a question a minute I've got no idea what it is it's infinite you know so in that in that sense in a gameplay sense what we were trying to do was was sort of show a couple of examples of how that may look like in the future but the models are there the models are there it just it create it it it needs a creative mindset shift slightly to understand what you can do with it and oh please go ahead sorry I was also going to mention beyond the say the the 2D 3D content there's also the story aspect of this like CU a lot of times what really hooks

players to a game is the story like if you look at some of the top grossing mobile games you know they're say puzzle games they like the the actual mechanics of the game are simple but then the developer puts a really interesting story around it all the way from the acquisition of the user all the way to different levels that they ship and then the story the characters that's kind of what really connects with the kind of like the human playing the game and there's also a very interesting opportunity there with AI beyond the hey these are really beautiful visuals or these are really kind of fast generated images that are super style consistent that's one aspect but the other aspect is how do you actually carry a story throughout a user's life cycle how do you actually acquire them and the interesting piece there is how do you personalize it that really kind of resonates with them like there's different types of games some are more quirky like Angry Birds some are maybe more serious or fun and these kind of quirkiness or the fun aspects plus maybe some you know user specific information could really personalize that game for for individual users and we believe

that's a massive opportunity to increase like the the user engagement LTV all the metrics that of course you all care about so I guess the other elevant in the room here that we haven't spoken about yet is you know the various ethical objections to the use of generative AI I think we know that this is something a lot of game developers feel very strongly about I mean do do you feel that you know these objections carry forces something you think about when you uh kind of speak to potential you know Studios and potential customers how do you reassure them on this who wants to go first I can start so well um so we we're just completing a year-long project at the moment which is around narrative generative narratives with Channel 4 with adman animations it's very focused on story and we were fortunate enough to win uh innovate UK funding for that of just over a million pounds which was fantastic on that

project we have a full-time digital ethicist from start through and she is we have I have we weekly one on once with her she's at all the exact meetings and what she does very specifically and I recommend this process is consequence scan so she is looking at every single piece of work that we do and working and working out what the intended or unintended consequences may or may not be that doesn't mean she has full control of the project it means she's raising what the potential issues are and then we work it out so it's a methodology that I recommend but you know I it's it's an it's an important piece and especially you know these days as things moving so fast it's important but that that's how we uh approached it on that particular project well I mean maybe just to press on that a little bit more um there are some more fundamental objections as well so for instance a lot of people feel you know the use of scrape data is uh um you know do is that something that well

first of all you know do you encounter push back on that point and secondly do you see kind of any practical way of addressing it well yeah I mean I just it's it's always been it's been something I've been tracking since about 2017 and writing publicly about going guys this is happening this is not cool and probably it's illegal and we need to think about it so it's good that that debate is now happening there are there is a uh there are so many strong views on both sides and ultimately it's going to be a political uh you know decision that comes down a legal political decision the fact is in the UK and EU we've got really good copyright laws they're not vague they're not uncertain well all of that they're really solid and and and they the the applicability into data sets is is absolute it's perfect but there is another view which is very Tech driven the US which is we need to change those so it's a political

decision we do the best that we can as a small company and maybe I might add sorry please go ahead no I was just so when I joined scenario it was one of the things that I think I really had to think about as someone joining a company based in AI you know is the business that I'm joining going to be again we we talk spoke about taking jobs but also when it comes to private IP and and games people don't want their data out there they don't want it going into some sort of Learning System where other people will be able to access and and IP infringement and there are so many layers that come along with it but I've been really impressed to see that everything is is so protected and there are there are companies utilizing our software with Legacy IP that goes back I mean well over 50 60 70 years that are seeing incredible results but generating new ideas and have no fear that their IP is going to slip out into the ether and and be uh their data will be breached so I think it's about just creating the

right silos to make sure that your your consumers feel safe um and at least from the point of of view of being new at a company I can say that it is a positive sentiment that I wasn't sure going into this side of um of gaming I was I was sure I was going to see so it is refreshing and and yeah just to uh just to like um piggyback off that like we're we're seeing exactly the same thing working with game studios who you know in some cases have got licensed IP so there might be a third party license holder involved so there even more concern um you know but also even if it's just just their own generated IP first party IP they still don't want that leaking out you know as you said Into The Ether um frankly we've seen more concern around protecting their IP you know our customers IP then we have about the sort of broader ethical questions around training models on IP in general um what we do um for these companies is we we help to really train and find tune um dedicated models that

really learn and understand to use how they use their IP correctly how to you know we we feed that with a lot of like including like you know just sort of 3D models and other um assets that they have that maybe not even featured in the game itself but that they've used during the production in order to really fine-tune the model to make it really capable of using and adapting their IP very well and and there's also an approval process around that and of course yeah to to the point there about um you know protecting and siloing that IP of course we make sure that all of the licensing that we're doing with the models that we're using if we're using any commercial models um that none of that data gets used to train other people's models that it's all very much kind of um yeah siloed within within for for our for the work that we're doing for our customer and uh yeah we've been able to to sign deals and and deliver great results on the back of that um so it seems like people are at least some companies are able to get comfortable with it yeah yeah I agree it's I think it's kind of at this Point become standing operating procedure for anyone trying to

sell to to provide those guarantees I think the interesting bit is the like at least the the one that's not quite um kind of making uh isn't quite widely used yet as the copyright safe based models there's some coming out I think I would separate the core technology which is kind of like based on diffusion and Transformers some of the kind of core AI Tech from a particular implementation of that aite Tech around things like say stabil diffusion or flux uh so depending on the implementation some are copyright safe some are not there's kind of different uh balancing acts that these base models do to kind of hit certain requirements like if if a company wants perfect copyright you know safety meaning there's no like there's no way for you to write say Taylor Swift into a prompt and get an image of t t because that is also IP infringement then there's also base models available say on on layer for providing that guarantees as well and we've also seen some companies like basically this is a decision made by legal teams and compliance teams they weigh the risks okay this one produces better results

but it's you know we need to really ensure our team is trained to use them properly to avoid risks or they say you know what we absolutely want perfect guarantees from the base models and those are also available even though with slightly different kind of performance characteristics but I think beyond the IP ethical questions and other really interesting ethical questions that we are you know hearing is okay in the future if you know if everything's a driven where is creativity and all of this is it going all going to be like this you know Art Factory with zero creativity and that's I I I don't think so I think there's creativity happening in in new ways but that's also a concern we're hearing like will every game look like you know each other now will everything be you know like AI generated and looked like it came out of the same kind of uh you know Art Factory uh and I don't think that's going to be the case but that is a real concern I mean there's obviously so much more we can get into here I'm sure we could go all day but I would love to take a few minutes to open it up to the audience uh I imagine a few of you might

have questions so we have about I think about 6 or seven minutes if uh anyone has a question for the panel oh come on guys not one question yeah that's a surprise what we're talking about uh ethics things another practical thing about scale oh we got here thank you very much a question about scale um uh AI use isn't free right so you know if using a a Model A Foundation model to uh in your game at runtime or whatever um are what are the challenges of scale because there's there's cost associated with that right and and you're in you're in HW to open AI or whatever I I think the there's scale and cost like I I guess they're they're related uh cost is coming down so today like imagine you build a game where you actually incorporate like a video model into it imagine like at real time you're generating videos for people to see today it costs like a dollar to generate like a 5-second video uh or maybe if

it's like copyright compliant it could be even higher because you're paying like the license uh kind of fees to the to the data providers in the future that's going to come down significantly because of several Trends one these models have become commoditized meaning better and better model are going to be coming out from all these AI Labs you know doing the research so it's just going to become cheaper too gpus are becoming cheaper and right now they're kind of like the the gold uh of this new era but they're going to be cheaper and cheaper and better and better performing and and therefore because they're better performing it's going to cost less time to produce these assets meaning again less GPU time means it's cheaper so I think that Trend will be like More's law right like in two years it's going to be 10 times cheaper to produce these assets another two years it's going to be another 10 times cheaper so I think that's the future that we all have to prepare for or it won't so it's how there go or it won't and the the so I generally I agree with you however open a and the big guys have got a problem which they've been giving

away their product so at some point they need to flip the dial and make some money and what you've seen over the last 12 to 18 months is open AI start to crank up the the the the charges of it so you know just be careful on relying on open Ai and anthropic and the big tech companies because at some point they're going to you know open AI especially is so focused on shareholder value they have to start making a profit and that profit will come from us that's where it's going to go so my my sense is there is a short-term bounce that the cost will go up and that we need but I agree on commoditization there will be other solutions that come into that space as well I would just add like I think meta is doing kind of like the the age-old kind of commoditization plate to undercut their competitors like and Google is also trying to compete so I think that kind of massive competition among massive tank companies is going to help the application significantly so if you're relying on open Ai and they jack up their prices you can always go open source with llama you can maybe look at Gemini from Google and these are based on open source Tech like Transformers

from 2017 and they do have some IP that is unique to say open AI but not that much like you can see how fast others have caught up to open AI in like say six months so it's not like the the openi tech is super differentiated it's like six months ahead maybe but the the rest of the players are catching up super fast question there's kind of two parts to my question so the first part is you touched on it there when you mentioned meta um I feel like open source is going to win out over time as costs come down and it's going to be really tough for open Ai and the rest to continue to make money for their shareholders so do you think China is actually ahead of us on this the attitude the mindset towards IP being more open and the sort of copy iterate fast I feel like that could take the lead in the long run and we might get stuck with our IP laws and not be able to innovate I mean such a such a good question deep question my my thoughts on this is I think the China has really good models around 3D Nvidia right now

um not on llms yet but you know llms could come out as well I think the real interesting piece there will be maybe regulation and how fast can tech companies innovate versus deal with regulation that's probably going to be the main determinant of who's leading in the future I think this maybe raises oh sorry um well just to kind of quickly follow on from that I mean how much are we actually relying on kind of Mo's law uh to deliver results and kind of help the roll out of AI on games or do you think you we already have enough even if we don't see massive improvements in the power and efficiency of the foundation models I think there's so much more we could do yes uh yeah I'd say uh in in marketing um we're not relying on moris lore I think it's it's already we already seen great efficiencies that can be brought at the cost that things are today if it gets cheaper that's even better um in in-game development more more generally I I think probably that's it's there's still still an expectation

that costs need to come down in order to make it like economical over the long term okay and I think we had one more question one more question hi thank you um yeah uh we we've been playing around with uh local AI on actual mobile devices um and we've got some interesting results I was wondering if you had customers that were following the same Trend and what kind of experiences you're getting for that anyone have any experience here only that I know Unity has been working on this for a long many years like seven eight nine 10 years uh on local on local language models language models specifically on mobile so I think you're right I think that the the the way in which those models can be optimized to run on those sorts of handsets is going to offset that scale and cost problem we were talking about earlier agree with you okay um okay maybe we squeeze in one more quick

question uh yeah just coming off the back of the last question um do you think that would potentially open the door for um large language models be run on in single player games on in the console space cuz the switch to from my understanding will have Nvidia tensor on the GPU so I think that might be something we'll see in Hardware in the future yes and on that note thank you very much to the panel and thank you for joining us this morning [Applause]