AI-Native Storytelling - Not Assisted, Not Enhanced, Native | Cannes Next 2026

[00:00:00]Thank you very much for inviting us on stage here in Cannes.

[00:00:05]It's a great opportunity to get to discuss some of the topics that are top of mind for so many filmmakers and especially AI right now is what everybody is talking about. So, it is a great opportunity to have that uh at a series of panels and seeing it from different perspectives from different countries, from different parts of the world is really really intriguing.

[00:00:34]For this panel, we will look more about AI native storytelling across the value chain of filmmaking. So, we'll go from very much hands-on visual effects AI creation and to a more structural visual effects how it affects the business towards looking at how it we can work in a larger context with some of the most forward-thinking genre festivals and sorry and culturally rooted AI content ecosystems around the world. And to do this, we have a lovely panel today.

[00:01:22]Welcome to you.

[00:01:24]Some people you have met before, I'm sure. Redima Lula from Eros. Welcome back.

[00:01:31]>> [laughter] >> I will make the introduction very short.

[00:01:34]I see most of the audience were here just before.

[00:01:38]So, very warm welcome to you.

[00:01:41]Director Shin Chul from festival Busan International Fantastic Film Festival in South Korea.

[00:01:48]Welcome here.

[00:01:50]A festival that has been on the forefront of inviting AI filmmakers and also doing now a training program. It was quite intriguing to me to read about 10,000 filmmakers should be trained in AI. I don't think we we have 10,000 filmmakers in Europe. So, that is a very very interesting perspective to see in a part of the world where the population is so much bigger than around here in our local area.

[00:02:20]Um next in line here is Martin Madsen from my own home turf in Copenhagen, visual effects supervisor with very much hands-on experience in actually using AI in his daily work creating stunning stunning visual effects and also with a creative and also cultural perspective on why and how can we actually use it in the way we make films for the future.

[00:02:48]Last but not least, Christina Caspers, managing director of Trickster from the in the Cinecast group.

[00:02:57]Um one of the biggest or most influential visual effects companies around here, I would say, and with a lot of international experience in seeing how AI will influence the industry and employees and employ employment and how we work industry in visual effects.

[00:03:21]So, welcome all of you and welcome to you.

[00:03:27]So, I'm sorry.

[00:03:31]Um I think we will start a little bit with a general question to to all of you.

[00:03:38]So, to to see where does AI native cinematic storytelling stand today, creatively, technologically, and is it a hype, failure, future? Most likely future, I'm guessing.

[00:03:54]Uh give us kind of the state of play from your perspective and in your part of the industry and your part of the world if we start on the far side here.

[00:04:03]>> It's um definitely the present and the future. I would say it's completely transformed the way we do things. Um we've been always been a traditional media house, you know, one of the oldest um oldest media houses that brought Indian cinema to the globe.

[00:04:20]Uh I don't think there was any more thing thing more traditional that could have been in our pipeline to begin with from the way we green-lit films, from the way we supported writers, directors, and storytellers. Now um it's completely transformed to the way we ideate. It's completely transformed the way we bring stories to life.

[00:04:37]Um for example, world-building, concept art building, um from just helping our writers to be able to play with their stories, to be able to, you know, just prompt and give ideas from the way we create our stories itself. It's it's absolutely phenomenal what it's done. And I think that there's always been this fear that people talk about that AI is replacing us. AI is going to replace human creativity, but in fact, I think it's the most useful tool to empower human creativity. And I I I wish that more people would see that and I think it's it's our our job essentially in positions of that we have that we're very lucky to have to be able to educate people on how to be able to use these tools to empower them to be able to bring their stories to life. So, I'm very excited about it personally.

[00:05:22]>> Thank you very much. I would imagine that the most of the panel here would be on the very positive side of use of AI.

[00:05:27]I would be surprised otherwise. Director She, would you like to say what what what do you think of uh AI right now? What is it uh what where does it stand in in your perspective?

[00:05:41]>> Nobody knows, I think.

[00:05:43]>> [laughter] >> Nobody knows. Monsters or or God or I I think nobody knows. I I don't know.

[00:05:54]I think I Uh I'm very curious I'm very uh curious about it.

[00:06:05]>> Yeah, I think that is also a thing for for most of us that the curiosity to see where it can bring us is key to how we approach it and uh also I know Martin uh to a big disclaimer here Martin and I are actually working together on a couple of projects so just so you know it as well. So I know a little bit more of his daily work than I do know of our other panelists. So uh Martin could you give your start you giving your overall perspective of where AI is standing?

[00:06:34]>> Sure. Um I think AI is definitely in its infancy.

[00:06:39]I think it's it's like you say it's you we have to be curious. It's the birth of a new tool in filmmaking and I think where this can go is going to be extremely interesting and it's important that we don't shut it out. We need to explore it. I think it's sort of like when the the first cameras appeared, we didn't have a language for how to shoot film and we're starting to develop that language now. We're starting to figure out how can we use this whether if it's the full native AI productions or it's these merges with with real cinema, the cinema that we know, how can we how can we use AI to enhance the stories that we tell, make them cheaper, look better, or just make them realizable.

[00:07:27]Uh I think that's that's where I see AI at the moment.

[00:07:32]>> Christina, would you like to add anything here?

[00:07:35]>> Um I could I would like to add that uh very practically yes we're using it.

[00:07:40]Very practically also we're not allowed to talk about it according to our related to the NDAs that we all signed with the biggest studios which makes it of course difficult as you were talking about curiosity.

[00:07:51]I'm talking a lot also with fear and anger about AI and how to collaborate and get these people who have the or experience in these fear and anger, how to get them involved and this is something where I'd say I mean AI general AI machine learning is here to stay. We used it in visual effects for many decades and it's just getting bigger and it's more hyped so the visibility is different and the usage gets broader but it won't like and it won't go away.

[00:08:25]>> Most likely not. I'm very sure.

[00:08:27]Going back in line here Martin, you have recently moved from classical VFX and virtual production into AI powered visual effects and you have coined the term synthetic brain AI is an like an open creative space generated from latent space. Could you explain a little bit more of these concepts and and what they mean in practice?

[00:08:51]>> I'm going to do my best.

[00:08:53]I think latent space sounds very fancy.

[00:08:56]You would what I imagine this is is the realm of possibility. It's sort of like a cognitive brain that knows pixels. It doesn't know anything physical cuz it has no experience. So if we see this as a as a baby, it's just it doesn't have that sense of actual connection. So everybody knows that AI works off of probability. It's like it sees what happens and it it sort of figures out what's going to happen next but it doesn't really know consequence cuz if it needs to move something from A to B, it it doesn't know that it can't necessarily go through a table where if you're a baby with a physical connection you figure out that you can't walk through walls and that's sort of where AI is at and that's sort of that synthetic brain needs to be trained on actual relations in the world and I think that's really interesting and something I'm I'm working a lot with.

[00:10:00]You you spoke a little bit about people being scared of AI and where at Canne Film Festival the the festival of actors and and creators directors and what what niche I've sort of taken with AI is animals cuz we we it's difficult replacing actors. I don't want to replace actors. I've been doing CG actors when we have to for explosions or or when people could get hurt we've replaced them with digital doubles but we don't want to replace acting. But we do want to make sure that we don't use real animals on set when we don't have to. We want to be able to control them with much more feeling and emotions than we can normally and on budgets that we couldn't realize before. So I'm doing a lot of theatrical features with animals doing elephants, chimpanzees, foxes, birds, dogs, all sorts of things and it's not fantasy it's not made to be looking like something out of a Disney movie. It's behavioral analysis of real animals. So we sort of figure out how they work and how they act to to be able to recreate that. Bringing them to life sort of.

[00:11:16]>> I think adding to this what I find especially interesting here is that you can train your own models and that you overcome this uncanny valley where you normally had like 10 animators sitting in a room.

[00:11:28]They do create something but it Oh, might look odd to your eye just because there is something off. And with with exactly training a model, you overcome this and this is the natural behavior and the natural walk and this is something that you could not achieve with just the traditional VFX pipeline, for example.

[00:11:48]>> Right. And it's so interesting cuz you you actually create behavioral beings.

[00:11:53]So, for example, I've I've had a fox that I was doing Oh, sorry. A wolf that I was working with and the wolf was trained on wolf behavior, but it's playful. So, when I ask it to do something like I need it to rub against the sheet, it it sort of popped its head underneath which which I didn't ask for, but it sort of behaves like a real canine animal. And that's it's really interesting. Sometimes you get something that we also get on set, not a happy accident, but uh it's it's not a system that's doing what you're telling it. It's interpreting your creative vision and that's something brand new.

[00:12:35]>> That that is a really really interesting and would bring me me back to you really my because you Eros holds like a library of 100,000 characters, I think it is.

[00:12:45]And training on culturally rooted AI.

[00:12:48]So, so what is your take on on what they are saying here about grounding it in kind of a truth that is trained on on real behavior. Now you're talking more like human behavior and human characters.

[00:13:03]What what is What do you see kind of is the development in that direction?

[00:13:06]>> Well, I think we all when we watch stories, when we watch content, we want authenticity, right? And I think when we decided to build our large cultural models and train it on all of our data, that that was the idea that how do we bring cultural authenticity? I won't be able to make movies based on all of these LLMs out there because they're not giving me the true nuances of what our culture really brings to the table. And that's there lies, I think, the problem that we essentially want to solve. Um and like any industry going through a transformation, there's always this fear of the unknown, you know, what what will this bring?

[00:13:43]Uh I think that was always the case even when OTT platforms came. We were one of the first OTT platforms to have launched in India. And when I approached every single massive filmmaker in India, like I would say the top five filmmaker, they were terrified of it. They were just like, "We will never let cinema die."

[00:13:59]And then the notion has just been that over the past 15 years, every single large filmmaker is dying to make a series and dying to have that reach as many audiences as possible. So, I think the idea is to be able to create culturally nuanced stories that are authentic and have that human connection to them deeply, but also be able to have that story reach, you know, the massive amount of people and be able to have that base level of being able to create that story at a much easier surface, right? Then being able to have to struggle for 6 9 months to get that story out, to be able to get it accurate, to be able to get it authentic. And that's what these tools really give to us, I think.

[00:14:39]>> That is That is very interesting to see that perspective on on a on human behavior as well.

[00:14:45]Um Director Shin, uh what is a AI filmmaker to you, if I may ask? I mean, is it like someone who uses AI tools or is it a different uh creative grammar or that is naturally shaped by the technology? Or is that uh What does your curriculum actually teach when you are teaching AI to to filmmakers from now here?

[00:15:12]>> So, it is very difficult to talk about it specifically, but spiritually we have to talk about it spiritually.

[00:15:22]So, like a So, a very interesting way we have some very intensive and interesting way. So, Do you know BTS and Blackpink?

[00:15:35]In Korean music industry has its own training system. You know, we are going we are going to we are in the middle of setting training AI filmmakers training system now. So, >> Yeah, but not just for Korea, for for a wider audience or is it is it very local-based?

[00:15:58]>> Yeah, so that we have a lot of a lot of creators from Asia Asian territory. That's so that we We are having a good time and so fantastic time to learn more about AI's capability and the finding a very bad thing about AI.

[00:16:26]So, it is very frustrating, but I think that the at the end of the day everything will be resolved.

[00:16:37]That's AI can do anything. I think in the near future. So, the question when I started the started AI initiative in the big plan.

[00:16:50]Put on in the international fantastic film festival.

[00:16:54]Everybody told everybody talked to me that told me that the can AI even make a films? Decent films?

[00:17:06]It is their questions. But it is it became an ancient story, ancient question.

[00:17:13]After 2 years, now the the relative, very relevant question would be um uh would is is um something like how how can AI make technology art and business something like that. So, that um the all the all the defective uh point for AI had will be disappear soon. Yeah. So, that you can now find anything. So, that that is the most concerned point.

[00:17:58]uh concerning point about AI. So, that when AI can do anything so, the content with digital content we are making now is now will become a commodities.

[00:18:14]So, that and we we we will be living in the flood of content automatic automatically uh created by AI. It's also what shall we do? It's something like >> That that is definitely going to be a That's definitely going to be a challenge. Yes, we will be >> Yeah.

[00:18:33]>> swamped with content because it's available, it's doable for filmmakers even Yeah, what you can do in a basement now is amazing compared to what you have been able to do before. Um I know you have a short presentation planned. Would it be a good time to do that now? Are you okay with that?

[00:18:53]>> I should I think it is not a time to >> [laughter] >> the show you show show you a presentation. So, it is uh it is not um So, 4 years ago I was in the uh market mar market uh Canne Film Festival. So, that uh, at that time I I suggested, insisted that the we have to we have to redefine redefine cinema.

[00:19:28]Uh, because the it is the it is the start of it is the first era of the AI, so that there is a lot of a lot of signs, signals about the uh, the emerge emergence of the uh, streaming service. And the so that I suggested the uh, we have to have new definition about cinema.

[00:19:57]So, is it uh, some people were little bit um, um, they don't they didn't like it by my my suggestion. Because it is the uh, the Canne Film Festival is the most uh, traditional and very conventional uh, orthodox film festival in the world. So, that there were a lot of negative feel negative feeling about AI. So, uh, but now, um, so, it is the very old question about it.

[00:20:32]So, that now AI is not asking but demanding the re-invention of cinema. So, that maybe AI is reinventing cinema now. We we we are not sure about in the behind behind the scene they are reinventing the cinema or not. I'm not sure. Okay.

[00:21:01]Uh, so I I I'd like to introduce uh, I'm I I I I like to talk about the uh the past uh not the future. So, um as you know, the Korean film industry is have been fighting against the uh the competing against the Hollywood movies. So, that because 30 years ago, the market share of the market share of the Hollywood film was almost 60% in Korean market.

[00:21:38]So, that uh Korean market share Korean um market share of the Korean film is was just about 10% 10 or 15%. So, that uh always um our question my my question was uh how to win against the Hollywood movies.

[00:21:59]>> Which which is hard. I know. Also in Europe.

[00:22:03]>> But, the uh so, we had we had at the time a very uh interesting different tactics and strategy. So, that uh we we tried tried to find uh the the the story Hollywood could not find in any way. So, that we have to escape to the uh that that that territory.

[00:22:33]And I found out that I I I I produced a lot of successful movies including which were the epoch the the very critical successful point in Korean film industry. So, that uh I found that I really want to make a fantasy films and science fiction films, but we cannot afford it.

[00:22:58]We could not afford it. So, and the market uh cannot afford could not afford this. So, that uh so, I I I am the one who brought CGI technology to Korean film industry for the first time. So, it was the Um it was story of about the nine-tailed fox. So, so, so, it is very uh interesting journey, but the but the I I believe that the CGI will become cheaper and cheaper and cheaper, but it is it was getting more expensive more more more more and more expensive. So, that when you look at the the the Weta or the ILM or something like that, the huge amount of money you need. Yeah, so, then what what about you? So, your company is getting bigger, getting expensiver.

[00:23:58]>> It is difficult competing with like you say ILM budgets, $200 million American budgets.

[00:24:05]We're we're doing a feature film with chimpanzees, and we do not have $200 million for a European film. So, we have to if this film needs to if this film should be realized, it has to work in a different way.

[00:24:20]>> Yes, same problem. So, that we we have a small market. Small market, so that we cannot afford the $200 million $200 million budget. So, uh and after after the first film I I I produced the uh bring brought the CGI technology to Korean films. Uh after that, we set up uh VFX company, CGI VFX company. But, after second movie after after second very very successful movie I produced, after that, I gave it I gave my company to my staff, one of the very very talented staff. So that uh he made the Korean he made Korean CG VFX great. Okay, so that and after that I went to Hollywood to meet the Bruce Lee's family.

[00:25:15]So I I I at the time I really want to make a make a resurrection of Bruce Lee with CGI technology.

[00:25:28]So in present now you can you can call it an AI avatar or AI something.

[00:25:38]At the time it was >> [snorts] >> very very very difficult. So that I met every every VFX companies in Hollywood.

[00:25:47]ILM ILM and Digital Domain or many other countries. And even I work with Weta in New Zealand. Okay. So but it is too too expensive to it was too expensive. And the and Bruce Lee's wife and daughter I work with them and finally with the There there was a commitment of six 60 60 million dollars from Japanese big game companies. So it was it was lucky I was lucky. But finally it was failed. I failed.

[00:26:23]It was failure. It turned out to failure. Because uh Uh we cannot we the Oh ran out of budget. So that I can It is it was my first time experiencing Hollywood. But so that I can I I I could not control it well and the money with the budget budget financial problem. so I can I I quit.

[00:26:55]So, I stopped um uh processing the process and uh say goodbye to Bruce Lee's family. I come I I can come back I came back to Korea.

[00:27:07]So, that I I at the time, so I I I thought that oh my god, why why how to win the game the battle of money.

[00:27:23]So, that I I I asked my the the uh technicians uh we have VFX technicians to to make it cheaper, make it cheaper.

[00:27:35]But uh so, please uh develop the cheaper cheaper technology for the beginner.

[00:27:45]Uh so, uh uh creator without money can create something like um something like Harry Potter.

[00:27:57]Yeah.

[00:27:59]>> But but is is money the biggest obstacle of making movies? Is that the And and is that what AI is going to solve?

[00:28:08]Uh that we can just make cheaper movies or uh will we see a reemergence of genre movies because they can be made at a a lower price point than before uh or or what kind of change do you see Christina?

[00:28:25]>> I think it's so generally the industry has changed in itself. So, when we're now talking to also the big Hollywood studios the conversation is has shifted. So, we're now talking on eye level while it was like 5 years ago still like "Hi, I'm the studio, you're the vendor, so you produce what I want from you." Now we're talking like okay, we have this budget, we have this movie or this episodic, how do we get to from script to final pixel?

[00:28:50]And it's a joint adventure and a joint journey we're going on, which then evolves with whatever new technology we have on the market. So, I actually like what we're seeing at the moment because it helps us getting involved earlier.

[00:29:06]So, we start from previous or even before talking about what's possible, what's not possible, and then we reevaluate over the course of creating a movie. So, it can totally be that we start like, "Okay, we do this in virtual production." And over the course of the shoot, we say like, "Nah, probably we can go on set, so we do it differently."

[00:29:24]So, this conversation has shifted. Yes, it's the budget pressure gets more and more, but yes, also the creative freedom gets more and more the more tools we have in our toolbox.

[00:29:37]>> Do you think you'd see other types of movies coming out of of your of of Asia or it's you specific coming out of India or Malaysia where you are >> 100% and I think I think that's what's so cool because people who didn't have access before, people who didn't have relationships to be able to access before with studios, with production companies, with platforms, they can just be in their room making a whole film completely end-to-end and be able to then distribute that hopefully on our platform or or on any other social media platform out there and be able to get that film to to reach hundreds of millions of people eventually. So, I think I think that's that's what's fantastic, but I think the economics and the relationship between content green lighting and the business has to be looked at completely differently. Is it even lucrative anymore to spend $300 million on a movie? We've seen so many big studios having tons and tons of losses because they're green lighting super expensive films, but the audience it's just not reaching or you know, connecting with an audience. So, we'll see companies like that having to relook at how their business model and essentially say, "Okay, with the help of AI, with the help of technology, we can now reduce that budget, slash it into a like a quarter essentially, but still be able to make the same kind of film and still reach the same amount of people if not more."

[00:31:01]Um and there will be films which may not even demand AI at all, right? It's not like every single story demands that.

[00:31:06]Like we're working with a ton of filmmakers that What if there's a, you know, film which is just about one person and it's limited to one room? Perhaps you don't need AI for that at all and perhaps you can make the most compelling story just, you know, as an isolated completely human film. But with someone who doesn't have the access or the tools, that may not be possible. So, it it the dynamics have to change, the way that the way the business and the storytelling sort of relationship has to change as well. And I think it's it these are conversations that have to be had.

[00:31:36]Yeah.

[00:31:38]>> I think it goes along also with a glossary. So, what are we talking about?

[00:31:42]Also comes like, are we talking AI? Are we talking gen AI? Are we talking machine learning? Exactly starting with these differences because it's just like again coming back to the point, it should take away the fear of what we're talking about and enhance whatever we can do. So, what for example, we're currently working on a huge Indian production and having the chance to visualize rather than talk about a red car, we can talk about that red car just as a stupid example and this makes me from a production perspective the process way more budget-friendly.

[00:32:16]>> Yeah, I think it's a lot more explorative of filmmaking. You know, we we we visualize things in pre-production, we visualize things on set, and we definitely work a lot closer with editing because everything that we do changes the edit. And I've never changed an edit as much as we do now because we can really create new cuts. We can what happens on set we sort of re-envision in post. And that's that's something brand new for for phone.

[00:32:46]>> And we have it all in the clouds. We do not even need to upload and send discs ever. It's just like it's amazing.

[00:32:54]Love it.

[00:32:55]>> frictionless now in that sense and I think that that's what's incredible that you can really spend time doing what you want to be focusing on doing and just just being creative and being explorative and just getting stuff done. That's important, right? Rather than just worrying about these small, you know, inefficiencies in that sense. And And what excites me is that the next biggest filmmaker is not just going to be the same guy or girl that's been making films for the past 20 30 years and just winning all the awards. It could be someone sitting in a remote village in complete middle of nowhere, but just because they've used these tools, they've managed to capture someone's attention. And that's that's what's amazing. Empowering creativity in its in its finest manner.

[00:33:38]>> But then how if if that kind of movie is suddenly made and loads of them are actually being made, how do we not risk of like all of them drown in the big sea of AI created movies that are out there and are coming more and more out there?

[00:33:57]>> I think it's about story. I think vish- visual effects I it's going to be hard for me to say is is an eye-catcher. If the story isn't good, it's not going to be a good film. I think you can you can make you fantastic universes and you can play with our imagination, but we need strong stories. And at least for now, I don't think that's where we need to use AI. AI we can use as a visual tool or audio tool, but it shouldn't replace our stories.

[00:34:26]>> And how do we distribute these stories?

[00:34:28]So there also AI could have into place, right?

[00:34:34]>> I think, you know, we've all seen massive like really good-looking movies that have been like spent hundreds of millions of dollars on them just tank because they're not relatable, they're not authentic, the story just didn't resonate with people. So, you know, everyone it will definitely raise the benchmark in terms of making fantastically good-looking products, but if the story doesn't resonate, that's going to fall flat. And how as you said, how it will be distributed and being able to curate this content to reach its audience is going to be key. No one's going to want to go to a platform where there's just masses and masses of content, but not being able to curate curate it for you. And that that also as you said, AI plays a big part.

[00:35:16]Algorithms being able to to extend your choices and and sort of what kind of content you want to view. And not just feed you the same thing, right? We don't want to just go somewhere and just be fed constantly the same thing.

[00:35:28]I certainly don't.

[00:35:29]Um you want to be you want to be a taste curator, not just you know, feeding feeding the machine.

[00:35:38]>> Thank you very much. I'd like to have that any questions from the audience at this point? Anybody who would like There is one in the back here. Do we have a mic for this gentleman?

[00:35:48]>> I've got a loud voice. Um I I love uh I'm a film composer. I look at AI, I think it's great. It makes it speeds a lot of things up. You were talking about the fox.

[00:35:59]Um now my point of view is that that fox doesn't cost anything. With music and art, AI is profiting from our back catalogs.

[00:36:13]Um it seems to be something that something that doesn't get discussed much because everybody's very pro. It's a conversation about VFX all the time, which I which I think morally I connect to and I feel I think that's fine cuz it's learning from the programming. The problem with literature and music is that it's learning from past examples and then the people that have created that are not profiting in any way. It goes back to 40-50 years ago with sampling and hip-hop records were using things. It took a long fight for people to get it watermarked.

[00:36:49]This discussion doesn't seem to be getting discussed much with the concept of prior examples and profiting from it.

[00:36:57]>> I think that's a very, very good example. I think the the whole structure of copywriting is going to change a lot over the next 5 years.

[00:37:06]Right now, we're very focused on not training on copyrighted material, which I think it's going to change, cuz if let's let's look at AI like a film student. How do you want to work with a film student that doesn't know film? You need to make sure that you actually know what films are and I need to be able to reference the rain in in seven, but what I think will change is that any model cannot recreate the data set from which it's made. I think that's where it's going to change and it's been like that today as well. Like we we could we make Titanic one-to-one, but nobody's going to do that, cuz it's going to be a lawsuit and it's the same with AI. AI needs to know music, but it should never be able to recreate it.

[00:37:53]>> I think also adding to this the studios I'm working with are literally using it in a way that they first clear the content before we're able to use it and we're using it offline. So, not in any online tools available. So, we're creating our own tools.

[00:38:09]So, that and they are only created for the purpose of this specific show and probably this is something also to be seen from a composer from a music point of view. I know that this is difficult, but maybe looking into the future, not knowing your business as I'm from visual effects, but looking at it like pay-per-use model. So, if somebody use a certain part of your composed music, you should get a certain percentage or a certain fee for it. I think this is probably a model of the future that could work. It's It's I think there are a lot of negotiations with companies going on at least in the EU, not sure about the other markets, but this is where I'm from, so this is where I know about.

[00:38:51]Um but this is probably the model to go because otherwise it's impossible to track it, I would say.

[00:38:56]>> But Willemijn, you have specially specifically trained some of your models on licensed material on large number of >> Yeah.

[00:39:03]>> local films, right?

[00:39:04]>> Yeah. I think that that was actually what our previous conversation we were talking about that with sovereign AI is that most of these LLMs have trained on publicly scraped data, and that's why I think it's really important to be able to create these sovereign models where there's traceability, there's transparency on where the content is coming from. And as creators, you need to feel safe in that space that if your content has been used, there needs to be, you know, there needs to be clear accountability for you to, you know, be able to keep uploading your content up there and not feel like someone else is going to basically steal that authority or ownership of that content. So, there needs to be, you know, you need to be able to be compensated for that quite clearly. And I think it also depends on the legal frameworks in place with the companies that are creating these models and and where the data is really coming from.

[00:39:58]>> Thank you. And there's another question down here.

[00:40:04]Hard to hear, sorry.

[00:40:07]>> Hey, sorry. Um quick one in terms of how you're thinking about types of models you're using cuz there's obviously like the more ethically trained ones like Adobe, but maybe performance might be not as strong as the other ones. And as of the other side of the spectrum, companies like ByteDance with SeaDance and others that have kind of scraped everything and are more performative.

[00:40:27]But then I guess like, you know, depending on what you're doing, you know, might you might need something different, right? So I ideally use like an ethical one, but then like if you're just rendering motion, right? Is that you're not really creating any new copyright. You've already created the image, you've already created like the frame, and you're just animating it. So is that maybe something you consider something more usable in that sense like a like a SeaDance?

[00:40:48]>> I mean, for us it's literally our clients tell us what we can use. So it's really like, okay, is there something that you want us to use or is there something that or or strictly not to be used?

[00:40:59]We even went into discussions where a certain tool was approved and they and then approved approved approved, contract comes in, contract says, "Yeah, but you can only use every tool offline." And this tool was only available in the cloud. So how the heck should I combine this? Then you go again into the legal conversation with bigger studios, which you lose anyhow. So um for me it's really I'm following the clients at this at this point.

[00:41:23]Of course, if it's something where we can't ethically stand behind, I would raise my hand and say like, "Listen, it costs X amount more, but therefore you get something we can stand all behind."

[00:41:35]So we're also educating our client to a certain extent at this moment.

[00:41:39]>> And often the models aren't good enough, and we need to train our own models based on our own data. That needs to be something that we of course collect ourselves or buy.

[00:41:49]>> I mean, it's the famous 80% that you can achieve with a certain AI model, and then you invest like into the remaining 20% the 100% that you need to achieve a shot.

[00:41:59]>> Sort of like making an asset in 3D, and you just train your your separate small lower models.

[00:42:05]>> I I share your sadness about that.

[00:42:08]>> Thank you very much.

[00:42:09]Um we are about to close the discussion here. I'm sure we could continue a lot longer. We can definitely continue over a drink somewhere around here later on.

[00:42:20]Um, if we were to be sitting here again in 2027, uh, what will have changed in AI AI-powered and AI-native story? Could you just uh, from out here, can you give us an example each of you of what you think will be new next year?

[00:42:39]>> I think that every day something is changing, so it's absolutely very difficult at this point to even comprehend what the next year is going to bring, but I think that um, from from our side and as as um, a media and technology company, we're looking forward to being able to bring stories to the screen that are completely AI-empowered. In fact, we have one that is is coming in December this year, as soon as that, and it's an animation movie. Um, I'm looking forward to being able to use these tools in depth and just keep using them to empower empower human creativity. I think that's that's what I'm most looking forward to. And I hopefully we'll be here in a year discussing it.

[00:43:22]>> Absolutely.

[00:43:25]>> So, as I told you, we have so very very good uh, education system. So, with we we set up the education system with SBS, the one of the biggest broadcasting companies in Korea. So, that we we work together. And last year, one year, we trained 2,000 900 students uh, in one year, and uh, produced 485 85 films short films a year.

[00:44:03]It it was possible. So, it was amazing.

[00:44:06]So, that uh my goal is the train the creators uh 10,000 creators in 5 years. Okay, in the middle, I don't know what what's going to happen.

[00:44:23]So, because it is speed is so so fast.

[00:44:27]The development speed is so fast. So, that I I feel sometimes I feel that I'm playing soccer in the changing field to changing playground. Okay, goal is going this way, that way, bigger and smaller.

[00:44:45]So, every every rule is changing every second. It's something like this. So, that it is it is something like a dizzy.

[00:44:53]So, that's why I I I told you the uh I I so that I can assure uh so >> Thank you. Martin?

[00:45:05]>> I think it's the it's a fast-moving space and there's a lot of things that's going to happen. I think we still need a lot of control. I think there's too much randomness. There's too much instability.

[00:45:15]Um so, we definitely need to work on tools.

[00:45:19]Right now, there's a lot of people that are trying to control AI with text, which is it's such a small input that you can hardly get anything out of text.

[00:45:29]There's no emotion like prompt romance and you're going to get a white tablecloth and a candle light. That's not romance. That's That's That's something you do for your lover or your husband or your wife. That's self selfless.

[00:45:43]Um so, I I think there's a lot of development that needs to happen uh on the on the cognitive side of AI. Um but technically, it also needs to evolve into larger larger context windows, more resolution, longer shots, just technical things that are still being held back by the technical state we're at right now, which will change for sure. But that's something that's going to be better next year, for sure.

[00:46:13]Um But hopefully also, I think when we come back next year, we're not going to ask whether if this is an AI film. We're just going to ask if it's a good film.

[00:46:24]>> That's a very good point. Christina, do you have any >> Um I hope we're here again next year with the same curiosity and less fear about AI. That's it.

[00:46:35]>> Thank you very much. I think this will round up.

[00:46:38]Thank you to our panelists, Radina Lula, Martin Matzen, Victor Chen Tu, and Christina Caspers.

[00:46:45]Thank you.