Brainstorm AI Singapore 2024: How AI will impact human creativity
Brian COX, Director of Gameplay Engineering, Inworld AI Dian GUAN, Co-founder and General Manager, Asia Pacific, Patsnap Donn KOH, Co-founder and Design Partner, Stuck Design Hanyi LEE, Founder & Creative Partner, Anak Moderator: Ellie AUSTIN, FORTUNE
Category
🤖
TechTranscript
00:00Hello everyone. Welcome to the stage. Thank you so much for joining us. So this is the only session at the conference that we're dedicating to the creative industry. So that's a lot to discuss. And Brian I want to start with you. It's a really simple question. Can you explain the most significant way in which I is revolutionizing gaming.
00:20Sure. So at Inworld AI we provide AI models and microservices for game developers to add all kinds of NPCs gameplay features and content generation.
00:33OK.
00:35We'll come back to that in a second because you showed me a very cool video before we came on stage of you kind of explaining a world that you
00:43wanted to be created and I simultaneously creating that world in a lot of detail. So there's a lot to discuss that.
00:50The answer is the global leader in AI powered IP and R&D innovation intelligence. You work with companies and organizations like
00:59NASA Tesla Disney huge organizations. How are you working with these brands and AI on a day to day basis. How is AI impacting the
01:09way that you're working with these organizations. Past specialized in providing smart data and intelligence for R&D research and
01:18development professionals throughout the R&D lifecycle all the way from ideation to discovery development to commercialization.
01:27So if you think about R&D it's much much more than the Eureka moment of a genius guy coming out with a brilliant idea. It's actually a lot
01:35of tedious and very long process for drug discovery example. It takes more than 10 years to develop a new drugs a lot of data
01:42collection searching trying to synthesize the data fighting patterns and eventually come to the product that works. So past snap
01:50build our own large language model and also on small models trained on the specialized data set patents academic journals and
01:58papers. So our model is like a very specialized chat GPT. We understand R&D professional much better. OK. Don let's come on to you.
02:07So stock design is a design and innovation consultancy. And I believe that stock has developed its own AI tool called
02:16HyperSketch. Why was it important to you to develop your own tool rather than use one that's already on the market.
02:24For us a hyper schedule is a reaction to the AI tools out there which I think are often very much focused on trying to get the
02:32best quality of output using the best data. But but for us we needed something that would amplify creativity and sometimes in
02:43that space bad data actually is the best data because we're trying to get the unlikely things that surprises us and things that
02:52haven't been good answers of yesterday. So for that main reason we do have a sketch as a way to amplify our work. And can you very
02:58briefly explain to us what HyperSketch does how how you leverage it to improve your work with your clients. Oh sure. If you take a
03:07simple doodle half a doodle maybe because you have half an idea without a prompt you could just get it to say make some products for
03:14me and you will see in high fidelity pretty soon. It's a matter of seconds. So that gives you something to bounce back with almost
03:21like a brainstorming companion. Fascinating. OK honey. Let's come on to you. So you're the founding creative partner as I mentioned
03:28of an app which is a global design company that works with the likes of Netflix Apple Banyan Tree. And what I'm interested in is at
03:36what stage are you using AI in the creative process when you're doing work for these massive organizations. I mean we tend to break
03:44our process into three stages. The conceptualization stage the exploratory mock stage and the execution stage for us the middle
03:52portion of the exploratory I guess mocking stage. That's when it has been amazing. It has variants as depth. It gives you I guess a lot
04:00of opportunities and possibilities but definitely not the first nor the last. And OK so why not the first or the last. I guess I mean
04:08our bread and butter is ideas. So I guess for us if you replicate AI and you use AI for the first or the last you're just going to get
04:17the instant artificial sameness and blandness which is I guess the worry that most people have in general. And yes if you use it in step
04:24two what does it enable you to do and how does that accelerate your work potentially. I mean if you're like a two man design shop you
04:30can pretend that you're 100 men strong or you can pretend that you know you know knowledge about I don't know about house or you know I
04:37mean something that I have no clue about. So it allows you to pretend to be other people. It allows you to be transported
04:43transported. It allows you to many places at once. I think that's really the beauty of it. Brian I want to come back to you on the
04:49subject of guardrails. And so you know if you are trying to create as we were discussing earlier a whole new world in gaming how do
04:57you make sure that that AI doesn't go rogue and maybe the characters don't start talking about things you don't want them to talk
05:03about or doing things that are dangerous and not something you'd want the players to replicate. So we allow our clients to first
05:10choose which I model they want to use either the publicly available ones or our own. And then we also have safety filters and
05:18customers can choose which filters they want to enable or disable. So for example if you want to have a character that can talk
05:27about alcohol or politics then you can enable that or you can also disable that so that they stay on track of the topics that you
05:37allow them to talk about. Got it. OK. And I think at the moment your AI models are that all focused on English language. But I
05:44believe that that's about to change. You want to tell us a bit more about that. Yes that's right. As of today we support multiple
05:51languages. So we support English simplified Chinese Japanese and Korean. And we got some feedback from our partners LG you plus and
06:03then sue to make sure this is culturally relevant to give a better experience to players in those regions. And why did you. My
06:10final question on this is why did you choose those languages. First of all there are a lot of video game players based in China
06:18Japan and Korea. And that's why we wanted to focus on those markets first. And that's also the regions that we received the most
06:27requests from like hey can you please support our language. And yeah very excited about it. You see yourself broadening out to
06:34cover European languages more in the future. Is this something that's going to keep rolling. I can't make that promise yet but I
06:41believe we will. OK great. I do want to open up to questions at some point. So please bear that in mind. And if there's anything
06:48you'd like to ask our panelists get thinking about it. I want to go to you Dan. And I know you don't think of yourself necessarily
06:57as a creative but you're someone who enables creatives to do to do the work that they want to do. What's your. We've spoken about
07:05some of the massive possibilities of AI. But what's your fear as we start to integrate these models into the creative process. What's
07:11keeping you up at night. And how do you think your fellow your creative here should be solving for it. Well for our case you're
07:20right that we enable creatives to R&D is a very classic form of changing. I mean converting human creativity into new inventions and
07:28new product that benefit of us. So from our perspective I don't see us as as a problem is more on the positive side when we observe
07:39how our customers embrace AI. I think first of all our customer who are the research and developers by nature they're curious people.
07:47So they are one among the first in the organization to embrace AI and R&D space. We already see so many success cases already. Just take
07:56one example. Maybe some audience already know in terms of drug discovery. As example it takes average 10 years more than a billion
08:02dollars even a new drug. And it's a very very tedious process. And one very important step is to identify the 3D structure of a
08:10molecule. Without that you cannot develop a new drug that can cure the disease. And this used to take a very well trained
08:18professional a few months at least or even years to decode one particular molecule. But deep deep mind's alpha fold managed to
08:28solve this very difficult problem by using AI to analyze the patterns and recognize the patterns. So that significantly transformed
08:38the entire industry. So we see more and more use cases where clearly aren't the people see the benefit rather than the problems of
08:47using AI. But of course one risk is once this repetitive tasks being done by AI because the R&D process even as creative there's a
08:56lot of non-creative tasks for example running the experiments repetitively collecting data analyzing the data. So about one third to
09:04half of the employees in these R&D centric companies are doing those tasks. Once this is taken away by AI what do these people do. It
09:13takes years of training to be a professional in any particular domain. So that's one potential risk that we have to address. And I want
09:20to continue with that subject and ask John and honey Don if you're you know a young person who wants to get into the creative
09:28industries in this day and age whether that's as a musician a writer an artist. Is it mandatory for now. Now for you to also
09:34understand generative AI coding because in a way those quite technical mathematical skills are at odds with the traditional
09:41creative skills. So how do you balance those two sides. I don't think you have to technically operate it but you need to have some
09:47literacy with which these systems behave because you want to use it to get better results than what the system is giving you.
09:57Everyone has the same tool. So you've got to kind of understand how to use it well to get a result that is firstly more
10:04competitive than what the system gives you directly and also better than what others are doing with it. So so that that is the core
10:12thing. I think I don't think we need all to become a scientist or you know. OK. You need to have literacy in it but you don't
10:17need to be executing it yourself. Yes. Yeah. OK honey. So last year a group of authors including George Martin filed lawsuits
10:27against open AI alleging that chat GPT was changed trained on that original work without that consent. And so my question to you is
10:36what should creative companies be doing to ensure that they don't become the subject of legal action for wrongfully using original
10:45source material. I know that's a big question.
10:51I think that I mean most of the times we've used it it's mainly for commercial use for clients. So I think that's why I like I
10:57mentioned before it should be used mostly for exploring. But in the execution stage which is usually when I guess the usual
11:03copyright laws and I guess infringements come about. I think that's when it gets a bit tricky because I mean if it's not created by
11:09some kind of human behind it then there really is. I mean I guess no no rights towards that. Yeah. Don if you've got anything you
11:16want to add to that. I mean there are a lot of similarities even though I think the legal debate is still out there with me seeing
11:24something on the street and then accidentally using it in my design. In the end we still have to kind of check that the final
11:30output like what honey has mentioned is not infringing on anything else that has really been done. So it's not very different from
11:37how we've operated innovation so far. OK. Are there any questions for any of the panelists. Yes. Oh fantastic. Let's go with this
11:46question here first. And if you could just stand up when the microphone comes to you I believe it's coming from here and say your
11:53name and the organization that you're from and then ask your question.
11:56Hi my name is Lois. I'm from Gov Tech also known as the government technology agency. So my question is actually kind of thinking
12:10about this whole lawsuit thing in a different perspective. So can I interpret that the presence of such legal issues facing this
12:21A.I. based companies are actually an indication of valid profitability or a valid market such that it's not. It's not a bad news.
12:31Right. In fact if you look at it as an indication it's a proof of viability. So I mean recent news about perplexity which is a company
12:44trying to get around this as well. What are your thoughts on such seemingly negative news. Yeah. Thank you.
12:54Who wants to take that. I can jump. Yeah. So many of our partners and clients they provide their own data. So that solves this issue.
13:09So you can either use the existing air models and most of them will tell you what they're trained on or if you want to provide your
13:17own data set then you can and then you can ensure that all the data that the A.I. is using is is your own. And then you can avoid
13:27lawsuits that way. OK. Anything. Yes. You can ask a question and then there's another one that we're going to come to. If you
13:35just wait for the microphone please so that we can hear you. It's just coming. Thank you. Hi my name is Steve from Spa. So
13:49yesterday the grab I was saying that they are creating a lot of creatives through a right. They're personalizing it. So if they are
13:57say one million customers in Indonesia and each has a unique need and they read their pattern over a period of time how they are
14:05interacting with grab. So they're able to personalize it. So now ads are getting created through A.I.
14:12Secondly Accenture has an ad arm where they said OK let's make a creative which should appeal to people. So they picked up things
14:21which people like something like pets or kids and they incorporate all those things and give it to and they created ad. And then they
14:30told a creative person to create an ad and then they tested both. And the one created by A.I. has got better. So how do you interpret
14:39that. How the industry the creative industry the TV ad industry is going to shape up. Yeah. Coming into picture.
14:45Who wants to take that. Don I could take that. Yeah. I'll just say that. I don't know if everyone here feels the same way but when I do
14:56work if I feel lazy I don't use A.I. Yeah. It's contrary to what a lot of people think. It's because when I when I feel lazy or when I when I
15:08do work and I involve A.I. I have now a thousand options to compete with and decide between. So when when I'm kind of like feeling too
15:17tired I'll just do something off my own and then submit it. Now I don't think A.I. allows us to kind of get better work. In fact the
15:25demand is there for us to do this exploration of A.I. and then either pick really really well or I'll do it because you know that if you
15:32don't if you just use that your competitors are going to do the same. So the market forces are going to force you to innovate.
15:39Actually there was another question towards the back. OK. This one here please. If we could just get a microphone.
15:52Thank you. Hi my name is David from PayPal. So I'm just interested in how this gaming industry is evolving.
16:02Given you know generative A.I. and I've read news about how certain companies have business models that are quite brutal right. Really
16:10laying off half of the artists especially those who are strong into the art right. You have to train for years only to be let go. And now
16:18using the company just uses Jenny I just keep the best artists there just to monitor the results and you know make them come up with
16:26graphics for add ons and stuff. So I'm going to seek your views on how these you know Jenny is impacting this workforce. Right. And what are
16:35the solutions. And do you think it is ethical for you know some of the managers and the top honchos in these companies to lay off all
16:42these two to the artists. Yeah. Great question. So the way I see it is using in world. We don't replace people. We use in world as
16:56a tool to help creatives reach their goal. So for example if you're a narrative designer and you're creating a character before you
17:05had to write every line the character would say. Now what you have to do using our tool is you set up the character's personality
17:15backstory emotion all these parameters. Basically you customize the character and then that character will speak using the
17:24personality that you've assigned to that character. And because they can give dynamic responses now you don't have to make a linear
17:31game. You can make story events happen and all the characters will react naturally to it. So our tool doesn't risk the jobs of
17:41narrative designers. Instead it gives them more possibility to be creative rather than change rather than writing the individual
17:49lines. They now have to create the entire personality of the characters and then let the characters speak naturally. Does that answer
17:59your question. OK. OK. That. OK. We can do one more very quick question please. And then if we can get the mic very quickly and then a
18:12very brief answer. I'm April Sandoz who I'm from Sandoz education. My question is simple is how do you feel about the future with AI
18:25to create less individualized atomized individuals. Will it actually increase greater communication amongst people.
18:35And. Truly reflect greater creativity and culture of people as people as as community as men as humankind. So how will that impact
18:46human interaction especially in these days. Honey do you want to take that very quickly. Maybe a slightly cynical view but I mean if you
18:54understand that the basis of AI is that it pulls from existing material that's in direct opposition to the pursuit of originality which is
19:02what most creators are here to do. We don't want to copy. I mean even if the even if we end up you know I mean referencing the
19:08intent of originality is the aim. So I guess if you have that in mind as a basis or kind of a guardrail then yeah I think it will
19:15definitely help with communication in general and just you know speeding things up. But it should never replace it.
19:22Brian D. and Don and honey thank you so much for joining us. There's so much more to unpack but we really appreciate you being