Panel: Supercharging Development – How AI Tools Can Augment Everyone.

[Music] I'd like to welcome to the stage uh you saw him already um the moderator for the next panel is Philip dorak from futa Ai and he'll be accompanied by Tommy from cosmic Lounge Maria from flying wild hog Ronnie from Clarity Q Christopher from model Ai and Nathan from modulate so folks come to the stage please and we'll we'll kick things off a round of applause for our our final panel of the morning welcome back thank you everyone for coming I think we have absolutely amazing uh line of topics and we don't have have that much time uh so uh let's

uh let's start uh discussing with an opening question that I hope uh gives uh everyone opportunity to uh direct their thoughts on the ideas that they find the most interesting uh so if you would uh look ahead for next 5 years we are conservative only five uh which three areas of game development do you believe will benefit the the most from Ai and here by AI I I would say we have broad definition of AI it's just advanced technology anything can can be called AI these days uh but uh we see that the advanced technology and especially generative AI has changed the market a little bit so uh in next 5 years three most significant AI Technologies and potentially if you see One Direction in software engineering that has not yet been adopted by the game developers what that what that one would be

Christopher there we go that's probably better uh well five years is a very long time so so so I don't I don't think I necessarily know um but I think for me um obviously uh Behavioral Systems in AI is going to have a huge impact uh on what we're going to uh on what we're going to see in terms of production and the same we'll see from from procedural content generation I think a lot of that is already coming through in terms of how games are being made and I think that will be super impactful on the production process um and we're just seeing the beginnings of that uh but the the third thing that I think a lot about and that's uh that's less how AI fits into paradigms of how we're building games at the moment I think that's starting to become well understood that it can have an impact there um I'm really interested in seeing how game design changes around the use of runtime generative Ai and other AI methods I I think on that side we're still waiting for someone to take a really left f fi

approach to what you can do with AI in the game design context and you know I I come from the Indie space myself originally so I'd like to think that somebody there will come up with something mind-blowing and creative that we've never seen done with with AI before more pragmatically software engineering practic practices well modeli we do continuous testing um that's what we deliver to a lot of our customers with uh with artificial intelligence and that's something that I think that the game industry still has some ways to to uh to adopt and chiefly du of effect but you can't really test a full game without having a certain level of intelligence in there found another mic by the way so you can that yep so I I think um NPC behavior is probably one of the most exciting areas that we're going to see a lot of these changes um I think at first it's going to start on the multiplayer side but I think the ghost Recons of the world and a lot of these single player titles uh where you manage squads or towns or or whatever it is you're going to see a lot

of a lot of innovation um there specifically um excited to see what happens in the in the QA space um with turnaround time and and things like that um and I think one of the other places uh this is slightly adjacent to game development but I I would see a lot of the Departments that are historically um underserved or under resourced within departments um whether it's QA whether it's Community whether it's your moderation team um the tools that are coming out to um expand the reach and functionality of these teams is is pretty incredible right now cool so uh Tommy huta from cosmic Lou and what we do is uh CLE puzzle games using our own technology which has very different game architecture and our designers can use noode tools to come up with a lot of different game mechanics and and use AI for that uh so the question is like in which areas there would be the biggest benefit uh I would say say that uh Aon will be having a big

impact on every single area of game development uh but of course in some areas faster than than others but I also think that uh there's no game area of game development where you still wouldn't need expert game creators so a great artist can make great much better art with AI than a not so good artist same thing with game design same thing with development uh the to me the most obvious uh areas for game development that you know benefiting from AI one is um analytics like AI will be able to find correlations that would be much harder for humans to find out than how we for example approach that is that our uh we sort of expose the rule book of the game to AI so that it can really make you know correlations that would be that would be hard to otherwise detect uh then second area is uh content creation so I think um game developers sort of have been you know have needed

to think about like can we do this kind of game because it would be so content heavy we can't do it it wouldn't makes economical sense and I think that's also changing with AI you can think of new kind of game Concepts because content creation if you use AI right the right way you know it would be would not be such a bottleneck it has been and then I think a third most obvious area would be personalization so when you can create a lot of cont content and you you can analyze you can do uh I think in one of the earlier speeches there was um like you can do hyper personalization everybody could have their sort of own game because there there's so many areas to personalize so that's one one thing and then the last question was that what area company should be using but are not doing if if you see software engineering uh practice that is used in large software houses that is not used in gaming industry yeah well I I think um anything quality like code quality related things I think games industry would benefit a lot from uh using Ai and

other Technologies for having better code because usually games are sort of one of things so you don't really need to care um um how you know quality about so much but I think uh that matters yeah hi guys so I I wouldn't be original I would say h Behavior system of the NPCs of course I would love to see some uh Improvement in the game Design Systems and also the coding so Poss possibly like faster and better detecting the bugs in the code and um moreover since I am a narrative designer and I'm a lecturer uh that actually teaches narrative design I would love to incorporate the topic of the AI systems uh to teach people to teach the designers how to wisely use this AI for our purposes and not to be afraid to use it a specifically in the narrative

design hi I'm Ronnie I'm the co-founder and CBO of clarity q and we bring gen to product analytics and that goes with what Tommy just said that we see a lot of advancements coming in the near future that everybody will be able with AI and geni to really be able to speak to to their uh data and get insights and not having so much heavy dependency on systems and scatter data and things of that sort and in addition uh with the rise of aent Ki and it goes back to the things you said uh games adapting dynamically to what's happening to uh the storytelling to uh the different difficulty of the user and what's exactly what exactly fits the user that's playing the game thank you so um I think we had some strong uh drive towards uh Chang changes in how uh non-player characters will

behave in the games uh I'm looking forward to it myself we probably will uh be uh again evaluating the question uh when it's fun and when having an agent that is behaving optimal in some way is actually not fun and rather boring and there is very interesting space in front of us I would uh follow up uh with a question when we look at procedural generation um it is it has been around for decades we can think about old games that were already using procedure generation at certain level and uh this level has been moving um more towards um empowering what we can do with procedural generation recently and uh there has always been one one of the largest challenges of uh procedural generation has been to setting up the right boundary between what we want to have

creative control over to be able to guarantee that the game is fun to play and what do we leave to the procedural generation this is a open I would say uh very very open question and uh I would like to ask where where do you see how far we can go with the automated generation for example with uh narrative generation or can we even generate game mechanics uh we can go in same order okay um well I think from my perspective I think generative Ai and some of the new content generation methods that are coming through they enable to generate more content to deeper depth at larger scale perhaps like you generate more text and you get like longer examples it'll be easier to get variety things like that we've seen talk like the same thing goes for 3D content and other kinds of content as well right but the thing that doesn't go away at least from a game design perspective to my mind is that any generative system that you're

working with is still an instrument or a tool or it has certain properties that you need to understand and if you're putting that at the core of your game design and especially like if it's runtime at the core of your game design so you're not using this necessarily to produce a lot of stuff and then configure it and then ship the game it's actually part of your core mechanic I don't think you get away from kind of like the fundamental challenge which is understanding this generation system and making sure that you put it to good use for a good experience um it's just that now you probably have a wider array of places you can take the output of that system so in many ways you just you get a harder problem from you know like all the way back to Elite I think where you had prly generated worlds right you have the same challenge but now you have it in so many more Dimensions um so so that' be my sort of like immediate reflection it makes it harder from a design perspective in some ways but also opens up new opportunities I think to to build off that uh holistically if you look at AI and its capabilities right now um what

it really does is give you the ability um to to add a scale to your games at a speed that you probably weren't able to previously Do It um so when you when you talk about the the problems that you want to solve we you know a few years ago we were we were graced with games like balers Gate 3 incredible branching narratives and and things like that um we also saw a tremendous number of Indie Studios uh pop onto the scene with with really with really Innovative games I think what we're going to see is these Indie Studios coming uh coming to play with um the ability to fill out quests in a way that they weren't previously able to in a way to build in with the abilities to build worlds that they weren't previously able to do um while still maintaining these small like one to 10 person Studios so I think um that's that's something I'm I'm really excited about and um uh yeah I I think that's

all right I continue from here I think this uh gets to a fundamental question of uh can AI create or design games and and to me it's about uh does AI know what what is fun does is a AI able to tell if some new idea is a good idea or not and and generally AI is pretty bad of doing that so it can look at data and you can it can sort of like if somebody tells good this is how you measure good AI would be doing fantastic work so there's some uh sort of narrow or narrow or selected use cases like for example we use AI our level designers use AI as a one tool to generate levels it really uh sort of um increases the efficiency and they can try so many things but because they are able to get the use the player data to set those metrics and and sort of requirements of like you know it has to be in these lines so in this scope and and um so that's use case where AI can actually create stuff but then technically we could put have an AI

model creating new puzzle game mechanics but that's not the problem to solve the games industry is full of uh people with a lot of ideas so the problem we try to solve is like how do we how do we get those ideas out so that humans can play them and we know if they are they are fun so um that's AI is not there yet yeah actually I I totally agree and this actually is a topic that uh that brings uh the matter if AI is going to replace for example narrative designers or writers I would say I would dare to say that writers definitely not because AI is not good yet at detecting and uh navigating through the emotions while writing um the storylines and everything however I would love to see uh the advanced tool that would help me generate the narratives and detect problems logical problems when it comes to storyflow and narrative flow and that I would

believe it will happen soon uh even sooner than later and I would definitely uh use it and again what I said before I would love to teach about how use it wisely and offer it and make sure that the designers and the aspiring designers know how to use it wisely and support themselves um I agree with everything you said I think we should look at it in the short term and the long term right now we see a lot of advancements and games are more personalized but it's going to be there's going to be a block some time that because AI as you said can't know emotions and what's fun and what's uh uh immersive and we'll have it's going to be interesting if we have the abilities and if technology will bring abilities to understand the sentiment of the user and exactly what's what the user is feeling through voice detection and things like that to see

maybe then we can take it to the next level but for now we'll always need the man in the in in the picture and to have humans in the loop because it's not going to do it by itself but in the future who knows thank you so it seems to the general opinion is that uh we are not yet at a stage that we tell some AI make me a game and it's going to be there it's uh is going to have more steps and uh for example The Narrative design has been uh one good use case for U neuros symbolic reasoning symbolic reasoning describes the world in logic and is able to create the logical narrative that uh after that is filled with a language model that just produces the glue that populat uh The Logical the logical sequence of actions that are happening uh this has been showing uh pretty strong values I think that um maybe maybe we can uh bring a few gallons of wine and talk about it later in the

evening definitely Count Me In uh let's move to next question so when we think about procedural generation it is um it is the capability to take some constraints and populate the rest of the world so those constraints are satisfied those constraints uh can be uh actually describing uh who is playing the game it can be the personalized Vector of the person that is engaging with the Vault with the game itself and uh as such uh it uh helps to enable much stronger personalization uh when the game is built around the person this is a different different angle on procedural generation I um have uh for example one one recent title that quot my attention has been uh bth no connection not we are not investing in them but uh it was just

really interesting game that has built on top of uh blocks that are represented by decisions of the player done during the game and then the whole storyline develops as a comic book and this comic book follows your characters that you play with uh it follows up how they are aging it uh you you start working and as take as characters after your original characters died they children and it builds up again and again in a one long story this was an amazing experience for me and I would like to have this as an open question if you if something recently impressed you with regard to level of personalization and if you think if you have seen the Black Mirror episode John is awful where uh the whole world is generated around you and you are that's that's the level of generation where we would be able to completely generate your experience it's a metric level of generation uh so the open question is do

you think that will happen soon and what impressed you the most with regard to personalization recently Christopher all right um well I I mean will it happen anytime soon I I think we're still ways away uh for for that kind of like an experience where everything is in is entirely personalized I think another question for me is whether you want to have a 100% personalized game or if you would like to experience kind of like the int ENT of the game designer through um through the experience that you're having so if you make the system entirely autonomously and fully personalized if you have that kind of like narrative there uh or or sort like that kind of like experience who is we who is we offer is is we offer yourself are you being is your experience being determined by some sort of like average of the collective unconscious and will you be able to relate that to anyone else after that experience right so at least and and maybe you know maybe I'm showing my age but I've kind of like been raised with the notion that playing a game is also a conversation with

somebody who wanted me to have a certain kind of experience and so I I wonder what that fully U personalized experience looks like and maybe it would be great but uh but but I I still like to see the human behind the game um but then maybe that would be represented in the system that generates experience like we talked about before right um and in terms of where that would would come from I think what we've seen recently with uh with you know immediate frame generation from from um from from large models definitely has the PO like that's something that I found really impressive and that I think points in that direction from a technological standpoint uh but I think it's just still like very early days before we see that do something that's at a grade where you would be able to uh to ship it as an experience that was a great answer um I think I I agree with with all of that and I think um there are probably some technological uh limitations that are are going to result

in this this being years out not not any time in the next six months to a year um I think the other thing that we're going to see is um is probably twofold right uh one is going to be privacy discussions um how much how much data um is this company allowed to collect about the individuals that are playing while it's generating the story around you and um and then the the headline grabbing AI hallucination part right so uh are there going to be instances where an MPC tells you to commit some Act of self harm tell you you're um I don't know insignificant or or things like that is it going to teach you how to make explosives uh these things are happening in uh with open Ai and and some of these other companies um I think when and if it happens in gaming it's going to be absolutely devastating to to those companies so uh these are a lot of the considerations that that we need to to think about before we go and try to

build games that that can sort of just create a narrative around um a single person all right so uh one example of U using AI to create or personalize the world is there's a a company called bit magic uh that I've um happen to know them so in the game it's like a third person Adventure game where you can just type in like okay do a castle and give me some some related Mission or or something and so it's it's pretty magical but then I think in this personalization there's the other other end where I think that most of the personalization we see in games you the player doesn't even know that it's personalized and of course there's been um there's use cases already which has been there for long time long time like you go to a freet to-play in-game shop you see different products different products promoted based on your behavior and all that so there's the monetization aspect uh but then I think that uh where personalization will be used a lot it's

like how to make the it's still the same game but it just made more fun for you because the AI or the personalization knows that this player should try this feature right now that would make it make the game less boring or more exciting so I think um that's that's I would say probably most of the personalizer will be something you don't even know it's happening yeah that's true that can be true uh I would say that uh that may be a dream come true from The Narrative perspectives um when it comes to creating uh immersive engaging uh experience however I would also um think over the emotions that it's going to generate as you said like some NPCs talking to you or creating some storylines inside it should be believable and uh so far I think when it comes to emotions AI is not yet ready so there there's always should be like a

you know a person behind this to um to to be emotionally involved in those story lines and uh be familiar with the structures and uh generating evoking those emotions so um that would be awesome if AI would understand those emotions and creating you know the relationships inside and um actually um meeting the requirements of our personal you know spaces and our person um identities right to when it comes to uh feeling the emotions um as you mentioned at the beginning you asked about uh black mirror so I think we're and maybe we'll look at it in a year from now when we'll laugh at me and everything will be Black Mirror and everything is f fully personalized but I don't believe it will be I think uh it's still we still have a long way to go as we all look at it and the emotions and everything again it's going to be it's going to improve over

time I don't think it will all be revolving around us but just specific things specific uh items AC according to the data we'll have about the users uh but not everything will be uh that way and I hope uh no one will follow me like a Black Mirror um but it's going to be interesting and regulation is going to be super important then but that's we can worry about that then as well oh we had some great answers I'm myself uh concerned about the moment when we start processing emotions in Ai and what we'll get out of it uh since um for example uh that was one of the largest worries in autonomous systems um that were presented in movies like Terminators and scii where we have projected the emotions that people have into the AI and um at that moment it

started behaving like us which wasn't uh good for anyone in the process uh these days AI is still very constrained it doesn't have the uh limic system it doesn't it doesn't manifest the emotions that in our cases are in in the case of human race is built on a hormonal system that sends triggers into the brain and causes uh random uh subsets of the brain to out of the sudden focus on certain thoughts and this hormonal machine has been built historically through Evolution with completely different um design Focus if we uh start uh creating those models of emotions with AI I'm very interested in the results and uh I on the level of personalization we have U some of the organizations in the world the large software hous is like Google and Microsoft they have the largest most

precise vectors that are defining you as a person uh often better than you know yourself because they use it to sell you vendor stuff they are good at predicting what you are going to want they have invested invested enormous amount of energy and resources to build really good personal vectors who you are I wonder what happens if those vectors become available for personalization in games because at that moment you can tailor the experience to levels that we haven't seen before I'm also wondering when this happens and how it's going to look like let me follow up to next question so uh talking about the large software houses and uh when we look at the advancement of software engineering and I I shouldn't uh identify one company or let's say that among them uh they have claimed that um software

testing will disappear that uh AI in general software engineering is going to uh completely cover the testing uh Necessities that are at the market at this moment uh I wonder if it is uh the same in the game industry and to be to be transparent we are a company that is making uh agents that are doing automated testing of games and so is Christopher so I think it's a good good question to pass to Christopher uh all right uh well well I can definitely start just from from our perspective I mean I think we talked about that a little bit before that uh in terms of like one software engineering practice that I think can still be integrated more into the way that we work in games it's kind like this continuous full loop testing right and I think that has just been really difficult for games because games are meant to be uh operated at kind of like Peak human intelligence in many cases

even you know either in terms of your reaction times or in terms of your reasoning and things like that and they're very heterogeneous I think games is probably or I would I would venture it's one of the kinds of software that has the greatest variety not just in terms of like what you do inside of it but also How It's Made because every game team will eventually kind of like put together their own thing right so so so in that way I think you know testing games end to end hasn't really been possible until we started having more generic AI methods that we could more easily you know integrate up against those games to have them operated so you know initially we obviously you can do things like unit testing you can do testing at the code level you can do there's a lot of things that you can do automatically already but kind of like really fundamentally answering the question does the game work requires it to be played uh and observed in many different ways it's a super complex question um and so so I think yes we'll we'll go in in that direction eventually right but if you go and ask people in the industry and you ask people like do you do automatic testing then I think

the latest numbers we had and S this is from from memory but 70 something per would say yes we have automatic testing in our studio and then you're like okay do you have like a CI Pipeline and like all those different things and then that number drops to like 30% and then you like once you have those pipelines you ask people okay like what are your strategies and the number of of of game Productions that will come up with more than three or four when you're down to like 15% and so that just means that there's like a lot of you know uh there's a lot of opportunity for the industry to make their own lives Easier by going in that direction uh so that's definitely something that that I would hope to see and soorry I'll I'll pass it on just a second but I would say if we can figure out those parts then we can start automating the answer to the question of like does my game work and I think that would set the human part of the production process but human testers is free to spend more time answering is it good I suppose so I think it's that last part that's that's really important right

um uh the industry you know over the over the past years things have been rough especially um actually not especially but I don't we we haven't we haven't learned a lot I think so if you look at and and then uh sorry to call out aaa's but if you look out AAA Studios like we're we're still we're still dealing with the the crunch we're still dealing with uh having to get games out um at clips that that are near impossible um and if you look at the advancements with automated QA um this helps speed up right because they're they're not going to get rid of these deadlines um because they're publicly traded companies um if you look at the the automated QA this allows them to to get these products out at a higher quality um on on Time right and um I I think if if we're being positive that that saves a lot of jobs uh which which

is a is a great thing um to to your point I still don't think um the emotion I still don't think a lot of the exploratory testing or the edge cases um are are things that AI at any point in the near future are going to be able to do um so I I think what we'll see is a transition from um you know the standard test case execution or things like that to uh looking for exploits for emotional testing as as you mentioned a lot of the exploratory testing and and things like that over time so I I don't think the um the human goes away I don't think the industry changes that that fundamentally I I just think it evolves a little bit and and shifts where it's testing yeah I don't have too much to say on the testing side but I think it's an interesting area especially like QA and you know quality assurance is that uh every I think every game company would be happy the less time they need to spend on that side so I think uh I I think are will be very welcome for

solving any problems on that area uh and I will be boring because I would get back to emotions right now however it's a very valid important and interesting question and uh I would say that when it comes to story and narrative in general perspective it's hard to check it um like 01 so there's no yes and no direct qu direct answers you you cannot say that story works and story doesn't work because for some people it will be working and for some people it won't because there will be lacking some emotions or some you know relationships or some other human factors so I would say that uh it's very helpful when it comes to detecting bugs and problems like logical problems in the storyline however uh I think checking the immersion and engagement in the story it

would be very hard for an AI to do it because it's like a human factor I guess uh I agree with everything you said and with emotions and everything like that I think we're going to see additional type of QA because now we need to QA our AI agents and we need to see what the Q what our AI what our AI agent does and uh so in addition to the regular QA we're going to have for games we'll have the QA for emotions or whatever it is from humans that will look at it but we'll also have real new QA that we'll need to do here as well new types of bugs can I add something about that tool that actually reads emotions reads is like a uh I know it's too too farfetched maybe but um I use a lot those generated text to speeches when it comes to generating voiceovers just to test you know the dialogue lines uh throughout the level before we get to the prop you know recording sessions with the talents and actually um these

times those generated lines those text to speeches tools are um incredible and they actually can um generate and evoke different emotions when it comes to tone uh when it comes to you know um shouting not shouting uh speaking in different you know uh with different emotions actually it's it's only a matter of how you write it and how many you know uh exclamation marks let's say you at at the end for example however it actually it's good pretty good and you can read it well it's not like Ivona in you know 20 years ago and this is something I would say that it reads emotions somehow when it comes to you know good coding uh coding yeah uh we had some great answers I agree with Ron that we have um completely new environment in front of us when the AI models start generating more and more content we will have to

come up with fundamentally different methods for testing it because the traditional methods will start breaking up the agents will need to be more autonomous more independent and Maria I also agreed that the emotional testing is probably something that we will not give away to AI anytime soon that the player experience and uh the emotional feedback is uh is I I don't know how to test it at this moment myself and uh Christopher I agree agree with your point of view we are we are uh game gaming is more specific field than General software engineering um we are in some transition and something will be automated something will take us more time I have uh one last question we are on the Edge of Time uh so uh this is a I would say quick quick question when we look at uh what all the agencies in the world are

saying about contribution of AI it will save on average dozens of percent of uh the operations costs across all Industries in the world which will free up lots of free time to people who will have that free time to use it in some way and uh since gaming is the largest industry in entertainment I would expect it that people will start playing more games in more creative ways and that the industry is going to grow significantly uh I would like to ask if you agree with this point of view or if you see some competition into entertainment and you think there is something else that people will do instead of playing games that will become the major consumer of the free time that has been provided by the cost savings Christopher sure I'll be brief uh well I think historically there's always been a notion that automation would give us more free time but we're still waiting for that to come through unfortunately um but but if it does it's definitely going to be

games I'm a bit of of a contrarian here I guess but I think it'll start with games and it will end with people actually going outside and picking up Sports and and trades and and things like that um I I think we we see that um I know uh in the US right now there's a a sport called pickle ball um that's become very popular um so I I think we're constantly being force-fed um media and entertainment and things like like that and and I think we're at an inflection point we're actually starting to see uh humans push against that and and try to be human a little more um but I I do think gaming of the the the things within the digital space will be the biggest beneficiary yeah I I'm sort of turning a little bit the question I don't think people will have more time and why I'm saying saying that is that in game teams now that they are using more efficient tools with AI I think instead of them spending less time creating games they will be using the same time of creating games but they will be just doing better games and or more games and using the

time you know with with better tools and probably same thing will happen for all the other people so uh we'll just be able to do better stuff but I don't think the time will be there will be more free time available unfortunately I want to keep my job so I would say games definitely I agree um I'd love to think like Nathan that everybody's going to go hanging out and going doing sports I would love that but I'm uh guessing that uh we have a few options uh gaming is great for us and for everyone here uh also um I'd say social networks if we have more time social networks and uh VR AR and streaming uh adaptive streaming we might see uh but like Tommy I wish I'd see some more time that's the only thing we're missing right now thank you everyone um it seems that we are in the best industry we can be in

and u i of course also love Nathan's Vision that we will spend more time doing sports and social interactions it's very ideal and uh thank you thank you for coming I think this was a great discussion and U be around talk to us thank you thank you