![[AI in Media and Entertainment 01.png]] Let me start by talking about a study that Gartner did. You are familiar with it, Gartner has something called a height cycle of the emerging technology. And they actually took generative A I and they said it is at the peak of the inflated expectation, right? So they said in their high cycle league just there's a lot of, and they're saying it's at the peak. And what follows that is a strong of disillusion where people are like, oh, that was all just a bunch of hype. It's really going to be a lot of hard work. I don't know, never going to pay off. So last August, they said that we were that the generative A I was at, there are over 150 sessions on A I. So I'm wondering, is the peak getting point here or are we really entering the trough of dissolution? Where are we with? A? It's like I, everybody looked at me as long as I was, I actually don't think, and I very rarely disagree with Gardner, but I don't actually, I think because there's something coming out almost every day. And some of these tools actually have leads, whether or not they have leads to stand alone or whether they can the larger pipeline and the larger together. There's a lot of, I think that we probably, if I had to predict, I'd say we've got about another 6 to 8 months before we get slow down. And let's take a look at and let's see. That is, I kind of agree and I wonder if A I is also large language models or language models in particular because you interface with into a language as opposed to Python or some other more esoteric way if it's going to reduce the the curve of disillusionment to value because really what we're talking about here is changing. Yeah, totally agree with you. I think uh it's, we're not let be, I believe the um just think about it in addition, in this generation, when you see it's something that half even as an expert in that. So once these capabilities are possible, um you know, you just, you can start to imagine so many new opportunities and possibilities. And I think it's so it's nothing, it's just the technology that make certain tools better, but it's really something that even changes all of the entire V should we went for a game production. Yeah, just to add to also the, we have obviously some things of them technology I think we all have seen here many amazing demos and capabilities that different research, the companies that are shown to us as a to show the potential of the capabilities of this technology. But I do agree that we are still defining how applications and we work on. Like I said, we are still at the very, very beginning of that, we are at the point where we can generate, you mentioned a video of text prom, but that doesn't end there. It's good to generate four seconds, 10 seconds or 20 seconds or even a million of video. But on top of that, we like we're all in class, we're all here senior position to tell stories, build advertising to what we have today is we have outputs coming from those models that look great. We have seen now outputs that the quality is. But if you want to tell a story that you have in mind and that game when you have and I think we are still not, we're sitting out there the early days. Go ahead. Yeah, I think that, that I think it's really like a wild. So the technology is developing so fast. You have so many people, all you know, different aspects of and even for research. So whatever you think that we want to work on somebody else or just release something new that, you know, we always like feeling that things are moving so fast that um you know, the work that we're doing is almost trying to imagine or predicting. Well, this is going to happen in the next couple of months, we need to be ready. So what you just said, the world is moving so fast, brings me back to something Bernard said, which is like there will come a point, we can debate when that point will be, but we're going to slow down. And I was like, really taking like that, is it, we're going to slow down or is this just one being a fly? We might be able to slow down to this? I think we have to slow down at some point because you have to assess what's actually there and what's real and what's not, you know, we see a lot of the parliamentarians, everyone wants to, but that's not what this technology, it's not about the tricks, it's about what you actually to make me or something or to make something better as far as the output or to, you know, help someone. And I think, I think that's one of if this could actually help me creative process. So that's why I, you off the, so that's why I'm not sure that yes, I think organizationally, organizations will try and test, they'll learn from their competitors and from standards making their bodies academic institution out speed now that you uh swing it, I think it's going to become more diverse than it's going to accelerate. But I agree that as technology adoption happens, organizations will have to decide this is working for us or we have the tech talent to take advantage of this technology. I think the regulatory bodies are going to take a while to catch up and that's going to be a team sport to make sure that regulation support that. But slowing down isn't going to be involved. Throwing technology s back in the box. So I don't think that's, I would never say put it back in. There's no is out there. I is I for when I talk real well, it may slow down. But there's also the uh the concept of start ups that may disrupt because I think for bigger companies more traditional that are rooted with a specific work. Uh I think the problem for them is they have to be slower for them to adapt. I no started with have the edge of the disrupting. I just seen whatever you did is just not arson. They don't have the technology to process debt that they have already invested billions of dollars in it can lead right to the next. OK. So as I'm hearing this unfold and thinking about um use the phrase Carla trick. And I think about a I, I think back back in the late eighties, winning the first, beating the first chess master, then you fast forward and then it's deep blue beating Kasparov in the nineties. And then all of a sudden you've got line in jeopardy, right? And so, and I say that just to say like, OK, this has been happening for decades. But idea w open A I and catching the team, I'm not part of it but it really, everybody's in terms of, of, of like, what's possible and there's an old saying that, you know, the smoke gets the attention with the fire is where the action is and I feel like it today, you know, is, you might be the smoker but narrow air, you know, might be the po using A I to solve very specific problems. Um simplify complex terrors, tedious things and to post production. 100 breathing that course. Correct. Yeah, I 100% agree. And that's how we part they were we apply research for. What does it mean that we have a research company is that we invest on every layer of the staff, the models we work on a, we work on the infrastructure to and at the end, we build applications and we build products and we build tools for everyone and we designed the company in a way that we wanted to understand how, as I said before, the full control of the al in order to create that you need to understand every single layer of the, of the start to provide that kind of life again for the the advocate for Hollywood students or for students, whatever place to be, that's what they were looking for. And that's how we build hope that this is how we approach and inform the products to identify possible directions for these cases that this technology can help us solve. And we build tools for that and we build the research that is necessary to support those tools give me make that a tangent a specific example the problem you're solving is, yeah, of course, we have 130 tools that are different, how they are different. The United States. I believe that work in different kind of moments. One of the first ones that we did so a photocopy in the browser and instead of you having to process of like training tracing frame by frame or something that you just point the subject and click, it creates automatically a mass mass for you as a propagator of the sea and in a matter of sails, that's one against the other the first, the less than a year ago. And we are seeing a lot of you may the, the support for historic boarding for narrative line and making sure that your idea gets well executed. We're using that for the record and for a lot be wrong for a, are you gonna tell him? I want to go down and how you spoke at a broadcast engineering and it conference yesterday and you one of the talk about any of them, but I'd like you to talk about some of the stuff you're doing today. But specifically, I was going to ask about the fallout and that's a really cool show. And I know that your company worked on that and did some stuff. What are some of the specific things that you were doing obviously today and production? Yeah, we can talk about every aspect but we've done significant art the week before. That's the part you very, there's a lot of possibilities with object to A ID technologies to do things much better than graphics, it has to do with that. The algorithms are mostly really, we things look as well as possible and we started with one of these but and doing the show, there were certain aspects into the show. So one example is eos in terms of storytelling, the show was there some flashbacks of the original play and we see that he's acting is put on some weights and they wanted to have the technology that could actually back into use. So like like and then uh yeah, so some of the challenges for that kind of thing off the shelf technology because it in some four kh er to deal with that very it's very challenging, of course. So we have to develop augmented other applications that are happening today. I'll give a couple that are a little more. Uh close connection is an example of tools that we have. Whereas using open the technologies are automating a lot of that automating going from transliteration to better translations using post fashioning. One lu one um Another example is that A I so so or so if you think about creating, you can actually have a, I find the, I wanted to, it will either automate clips for that or automate putting together a nine. Ok. But you can the, my most, well, first of all, that lot of they don't right now, um, all the a nickel they like you video. I mean, that was, um, debate but, ok, never something patient you said. Yeah. Um, talked about almost a I know is artificial intelligence but is assisted something called human in the loop. So it's not that a high just takes it and runs it and does it all for you, but it gives you a first draft, it goes out and pulls together all the funny scenes and gives you 30 60 90 seconds for you to then say, OK, now I can more easily hone in and focus on my creative intent. Is that how, how is that right? Is that how you see it play out? So Microsoft, if you track our A I technology, we call it co pilot for reason, it's not autopilot. And if you look at the co pilot logo, it's less obvious, but essentially it's a round logo. But what it started at was two hands. So the idea is it always has a human hand. The intention has not been to get humans out of a room, but rather to make more impact for every human interaction that happens. Personalization and scale. I mean, from a human computer space standpoint, there are I think three concepts. One is you fully replace the task with an A I. The second one is you assist the person. So and ideally, it would make the person become better at its skill. Uh I mean, one example is auto complete. But if it figures out that the person is uh you know, becomes better, it'll stop doing it. That would be better. The third one would be really supercharging a person so that the person can things much more. So what I take away from all of that is that A I is a tool and that it's going to potentially free up people to be more creative or to focus on different areas, which means they, any time you have a new tool, you have to be trained on how to use it and, and, and so page when we're preparing for some of this. Uh You used uh skill, skill is a word I think of as a noun, right? You used it as a verb and I think what you said was this is going to be, we're going to be skilling, skilling, scratch my head. What is she talking about? But explain what you meant by what is skilling and what are we going to have to do with this new workforce? This is, this is going to be an air of investment for all organizations around the world. If you want to get impact out of A I, you're going to have to do sort of both play offense on scaling. How do you do better prompt design and engineering? How do you open up the understanding of what's acceptable design and then defense uh understanding how to empower your employees so that they are using the tools that you find acceptable. One of the concerns that I have in the industry right now, I also support the education industry and so I spend an inordinate amount of time on teacher chat groups looking at what they're doing, what they have and they're recommending to each other. All of these, I'm going to call them parlor tricks. I like that these parlor tricks that are llcs and when you go on to them, maybe they do something really simple and elegantly like create a quiz from a PDF great, except for you don't know who they are. They don't, they're an LLC, they don't have any license sort of background or agreements. And I worry that they are a front for taking student and teacher data and without policies and training that have school systems, say these are the acceptable tools that we can use and training that explains to teachers. And I don't want you to go on your private email and use this either, uh you know, that sort of policy and practice at every organization level is going to be part of scaling for your A I workforce to 100 because we are in a time and place right now where there are not a lot of there, there's not a lot of guts to your executive rate. There are those, as they say, they're bad actors. Everybody. So that is authentication is going to be certain. So this is an acceptable place. She her. Yes, this is what? Yeah, I get and it's based on the previous conversation about, you know how fast things are changing too. This is not a oh good. You're skilled bound. This is pondering, you're never done. And Alejandro guys both run start up companies that have to deliver today. Like I'm guessing there's probably some people in this audience who would like to get into a job in A I like what skills do people need today? What are you looking for in terms of the types of folks or is it like you don't need any skill except the willingness to learn? I mean, that's not the case. Yeah. So it used to be, I mean, I'm running a service since 2015 and it used to be that you wanted to hire experienced people who have significant uh you know, programming software development R and D but, and even the students, right? So sometimes you wanted to hire the people who have more uh more research expertise. But nowadays it's a bit different. Uh nowadays, uh when we hire people, it's usually people, fresh hosts. Um because because it's developing so fast, the skills that we need, I have to learn people fresh out of school. Like how, um, so I think that's a significant change and so you really want to hire kids in our start up switches. And I think for, you know, anyone else, like, uh, even, you know, three years after school, I think it's really important that they are working or they are familiar with, uh, anything that's really new and you don't really need to go to school, learn these things. And so a lot of things you can, there's a lot of online resources, like tutorial communities. People are very good neck to exchange ideas, exchange skills, but it really depends on what aspect. So for the more technical aspects, I mean, you want the, yeah, pretty much. We just would, you would have, I, uh, and I know all the machine learning frameworks and who have produced uh portfolio publish of computer. I know, but for the creatives, I think it's really interesting because you don't necessarily meet the person who has 1020 years experience in my eye. You? Yeah, I wonder though if that's, I mean, look, there is gonna be a role for deep expertise in an era of a, I, so I don't think it's all going to be 27 year olds who are with two years of experience. I think it's over time. What you're going to get to is something that looks very much like the bell curve of the workforce today that there are going to be roles where you go. Yeah, there's something to somebody who's been in this industry, whatever that industry is for a long period of time, you can look at things and say we've tried that and now in an RV, maybe this is going to work instead and then there's something to be said for earlier in career, maybe less expensive or maybe new to that career. It's maybe less of how old they are but where they are in terms of their expertise and cost. Yeah, I'm speaking more in terms of like the uh but of course, it depends on the role, if it's a more leadership role, who knows the direction, who has the expertise of. But really talking about, you know, the, I mean, at least in the art in my space, I think the, I mean, when, when we look at all the applicants, it's always the fresh out of school that you have the will the skills, which is something that we really have to. Yeah, I think at this moment in time, at least on the research side of things, I think deep, deep expertise is really crucially important. Again, as we were saying, we are in day one of these very early and we're still defining the foundations of how these systems would work. So for those are some, the level of expertise is needed in order to build those pillars in order to build those things that are required for us to eventually on board new people with less experience and less expertise on the creative side. I think on a listen, we have, we have detecting or we haven't been seeing that adaptability to switching between different skills is becoming crucial. And with this, I mean, we are seeing product designers turning animators and we are seeing animators, studying filmmakers, we are seeing filmmakers studying coders, top coders becoming game development. And I think we are entering this phase where expertise and skill starting to get a little bit more blurry because there is more accessibility. There is more ways to enter a new space than perhaps before was a little bit harder because it requires some kind of there was a learning curve to learn some of these, I guess for some projects bringing some deep expertise will still be needed to go more in deep on specific things, depending on the project, depending on the government on the goals. But what we're seeing today is that it is open to, I like to access some new stuff and combine them together for maybe an overall vision, an overall kind of like project direction is becoming really good, observe as well, artists becoming technologists and researchers becoming like hardcore researchers wanting to learn how the creative process works. So it's really interesting. Yeah, I always say that we were all going creative. Every one of us would want me in overtime either it was a bad art teacher, it out of us. But when you start to think about this, what's going on, you see, everybody is going to be Leonardo Di Vinci. You know, they're going to be a mechanical engineer, but they're also going to be an artist. They're going to be a technical person. I often look at this and say, what are the new opportunities that are available and the new jobs that are going to be coming along. You know, it's sort of like the revenge of the English major right then. Prompt engineering is all about knowing how to use language. You know, when I say English major, I say that because my wife is her revenge is coming. But um but, but I I, you know, she's been in the house for years, my Children live towards. But I also look at this and I say, hey, you know what this could provide us the revenge or the return of the render those of us who were editors at one point, we used to love when we can hit the button and say, oh, it's got to, we can go and take a walk and come back. Some of the leads that we're doing right now, they might bring that back. Um But I do think that there are new state of science has just shot through roof right now because we have to think about all of this information that we get different. But the scientists are now creative because they have to be creative in how they're to get us to. And Bernard, I think that's actually right. You know, Microsoft owns a small company. You may have heard of it. It's called linkedin. And we, we've done some research on A I skills and job postings and like what let's read if you're talking about, this is not a, something like this is the fastest growing on boarding of job postings with these types of skills. Uh and a focus on hiring in the history of linkedin that we've ever seen. So every and this is not a, oh, we have to train the 26 year olds or retrain the 46 year olds is that everyone in the room uh should be at ned to hone their skills and learn, which obviously in this audience is probably true. You would be in this room. But I do, you know, for me, that's a cultural issue for my organization, for our customers, for our ecosystem. Well, the funny thing to me is that A I is probably reading all of those resumes before they make it through. Well, it's also resumes so that might make you feel better. So I want to come back to something that we were talking about a little bit earlier, which is how A I is being used today and I want to talk about or I want you to talk about places where it's coming up short. What are some of the things that you can't do that we want it to do and that, you know, and maybe talk about a time horizon if you've got one on, you know, what it would be to, you know, hey, it's just not able to do this. I mean, it took the A I model, I think it was like 16 years. I could be wrong on that before it beat the chess master, right? The chess master won for well over a decade and then A I won and then six years later beat the world ran. So, but like, so it's OK that it's going up short. But like what are some of the areas of media entertainment where it's not yet quite, you know, hitting it but hey, hold on within another whatever, 2, 10, 5 years do you have any? Well, it's interesting because I used to think that A I can't be creative and now it can solve a lot of creative tasks so it can write stories for you. It can, but we have to rethink about what does creative be mean. And um then we have to think about like, how does E I even work or what do we call me? Yeah, it is just deep e networks that specific architectures that have, that can process a lot of the uh tons of GP US. So what it means in the end is basically the A I that we understand, right? Now it's not necessarily what we see movies, but it's basically just a very powerful data processing machine. And it's, the analogy is a little bit like what if we didn't have internet and we couldn't search, if we didn't have access to Wikipedia, we can know all the information about anyone else. And now what the I does is that it's just processing all the data in a way that it's easier for us to understand what it is or even synthesize new content in a more efficient way. But I think the main, the main problem was a I, I mean, it generates very impressive things, but one thing that we didn't touch on is how do we really control it exactly. And it reflects on every aspect is the tragic b doing exactly what I want is it, does it have that personality, the perceived personality that I'm expecting is my video generator generating exactly that specific content that I want. That's why we still have the traditional PX because you still can't do all that. You can't say, well, you know, I want the person to have that specific height and it doesn't have a clear understanding of physics uh and everything but doesn't even understand, it doesn't even have the data representation for that yet. Um So I still think there's a lot of research that do there. Um But a prediction of where that would go. Uh I think uh probably an E I system that has a better uh understanding of how the world works. So that can model the world better in a way that it makes sense for humans. You know, it's interesting to see that because in the last week, there's been a lot of progress on spatial reasoning in A I that is super interesting and I think will start to move us towards that. I think what I'm seeing and I'm not enough of a deep technology expert, the folks on this panel. But I think part of what we see and what I hear you saying now is that people are using A I to address opportunities in certain workflows. So it's very good at that. When you, when you design it for that, it can take care of a specific workflow where I think the future is going is around systems, multiple A I systems that interact well together that are orchestrated large language models, small language models on prem models of the cloud models and optimized for scale. So I think we're also very early in the LM space on the technology itself, which means that it's expensive that we we're throwing very expensive processing power at every single problem where not every single problem might require that level of processing power. But that too is just part of a cloud adoption curve. So you know the cloud, the cloud and the language models will get more efficient, more effective and therefore cheaper to buy but right now, since we're on the early part of that adoption, it's also expensive. Yeah, I totally. So, I think the simple thing that is not quite there. Right. See, um, because you can ask the same question and continue to get various iterations. It's just, it's, you know, it's built on a neural network which is supposed to benefit the human brain. Everyone in this room has an opinion, the system acts that way as well. So there are a few people who are trying to, you know, for me in entertainment to say, let's build something that can be more consistent as far as your output so that you iterate within a single image. Let's use that as an example. Um If you can get there, you know, right now hashtag C and you know, as you put it in and it says, OK, now this is my, however, if you look at the pictures side by side, they still look different. So I think consistency you can solve that that will help fits forward in a way that for the Yeah, I agree, I think it's consistency. I think it's controllability. But I also think it, I also think that A I systems are yet not as smart as we think they are. And I think reliability is also crucial similar to what you were saying, consistency. I think it may systems yet or interacting with many models, you can get to a point of getting good results until like 95 98%. But that 5% remaining, that remaining 5% that remaining 2% might be crucial but might be some might be the key for what you need, what you need for a specific product or for a specific product. And you cannot afford having that margin of 2% 5% that is not reliable. And that's why I also believe innovation research is not going to slow down. There is so much interest and process of just trying to optimize for that last 5 to 2% that is remaining there, that will allow things like consistency, that will allow things like. So I want to shift gears a little bit and I'm going to try and weave a couple of things together here but feel free to pull them apart if you want to. But we've talked a lot about how rapid uh things are changing, how fast things are happening and how tough it is to keep up with the technology and in that sort of environment, you know, is the technology outpacing our thoughtfulness about the ethics of how it should be applied. And so I want to think about what or ask about what are the ethical implications of A I. And the thing I want to weave into that is the notion of A I is a global phenomenon and ethics are not universal. Um uh Microsoft's a global company, um uh how you're now in Abu Dhabi teaching and working there, you're Alejandro from Chile. And, you know, I mean, just, it's, he's a global, I mean, like just on the stage there's, you know, there's, there's a whole world of considerations that are out there in different ways and different parts of the world may be thinking about it. But how do we think of the ethical implications? And is there an international component where, you know, ethics should perhaps be differently? Yeah, I mean, you know, yes, there, there, there is and, and I think you, you hit on something um right. Uh and wrong can be subjective based off of where you are. Um However common sense seems to be universal. And I think when you put the common sense factor into what we're doing, there is a desire in most human beings not to do something that is going to harm other human. And I think that we have to approach this technology with that as a baseline, it is moving fast. There's a lot to learn, there's a lot for us to do. Um But just because you can, does not mean that you should. So there has to be a way for us either as an international organization or something to take a real hard look at what's being developed. Um You, you never wanna, you never want to like silence or push down uh research. You, you wanna at least allow the research that how to use the outcome of that research is what it's supposed to be. And I think that's where the ethics really come in. My feeling is that all companies should be ethical and responsible. That may only not be the case. Um But if we are to survive as a human organism, um, whether we're going to be successful at saving or destroying the planet or whichever side of that argument you land, we have to about what are the, the by products of what we're doing? How is this going to be effective for the goal that I'm creating? But what collateral damage can be done that I should be, we, we shouldn't be trying to instill that at the very base development. And Rena, I really think that what I'm seeing and I'm interested in your opinion from the part of the elephant that you're hanging on to this really is where responsible A I principle is that there is within that the technology industry, some broad level agreements at the 10,000 ft level of responsible A I means that it should be safe and reliable, that it should be private and secure, that it should be fair and inclusive. And that every part of those uh six components needs transparency to the end user, how A I is being used. And so, you know, at, at the highest level, I think we are designing for that as hyper scalars that we have some census with the larger consumers of A I and some of the developers of the models or all of the developers of the models. Uh And we have started to gain consensus with policymakers. Uh at least in the G7 around that framework. The challenge is going from principles which you have consensus on the daily practice in engineering or design decisions or monetary decisions or all of the things that it takes to take a principle into standards and standards in the practice. And so I think that a lot of the work has to be done with policymakers to make sure they understand the responsibilities of A I, making sure that they are aiming for the goal that you're saying the common sense without dictating too much about the how you get there because technology is always going to move faster, the more they inculcate into policy, very specific how the farther behind we're going to get from their policy implementation and making sure that standards bodies like the one that you have are helping with that. And what does that mean? And how do we do it? Yeah, I really love the aspect of the common sense. That's what everyone should have in common. I was asked the other day like about like what do I think about regulations, you know, us versus China Europe, Middle East and everyone has a different opinion on how it should work. And yeah, I was thinking that maybe it's not a bad idea that everyone has a different approach because as this, here's the thing, I think nobody really understands what is the recognition it, because it's developing so fast. Policymakers don't know what the capabilities are or how the protect actually works. And technologists, you know, or tech companies might have other incentives of like what they want to build, you know. So I think we need to give it a bit of time. Of course, we have to make sure that we prevent any catastrophic events or anything illegal being done, but we're harming people. But I think, you know, I'm kind of like still optimistic and I think that, you know, we should give it some time and then see how things evolve because ultimately, I mean, we are developing everything for humans like for us. So it's like human centric design. So um yeah, I think, you know, just like, I think about like a couple of years ago when everyone was scared about deep fakes. So, I mean, it hasn't caused the type of damage that people were talking about. Of course, people have used it in some instances for disinformation for scamming. There were a couple of high profile cases, but it was, it's not the end of the world. And also big, big tech companies are I think doing a really good job in terms of like, uh I mean, they can't stop every disinformation but in terms of like what they're trying to do, like incorporating uh you know, fact checking mechanisms, etcetera. I think, I mean, that's, but obviously, yeah, and that's why I think like CP two A and some of these other things that, that, that are really focused on content protection, name, image and likeness protection, um authentication and coming up with tools that can help us to identify. And I think those are, those are really important areas, a lot more focus right now. Content problem making sure that there's a chain of evidence what is created through and make it easier to identify. Yeah, I think that's key we have, we have systems that are today that allow us to protect some specific things like content moderation, maybe watermarks to A I generated content and things like that, that can help us solve some of these problems in different parts before we commit to regulate things that we don't know. You both mentioned, we need to understand, we need to work with policymakers and policymakers need to understand how this system, but the system is still not stable. There is no stability how we are doing this yet. So it's very early and understand all implications and possibilities and capabilities to maybe things too early. But I think there is a lot of things that are coming from different people, from the industry and different people working on the idea of like what are the practical things that we can take today in order to avoid misuse of? OK, we got about five minutes left. I want to kind of go back to where we started in this. Um you know, uh are we at the peak of the hype cycle? Are we going into a trough of disillusionment when we come out on the other side? You know, according to that, uh that, that process is typically about 5 to 10 years where you reach, you know, um actual sort of realization of the technology. And so I want to ask each of you, we're going to start with you Alejandro and then we'll go down the line, but just ask each of you to say, OK, here we are at N A show 2029 or 2034. We're in that 5 to 10 year window. How has a, I transformed the media and entertainment business with firm? Yeah, how 10 years? Yes. So now that we're closing, I'd like to think of that question as an invitation of today. Start thinking from that upon today. Like today, I feel like we're over obsessing with the capabilities that are happening today. We're seeing a lot of malls that promise a lot of things, a lot of demos that are showing a lot of the potential. But as we all have said, this is changing really fast and there is a lot of progress coming. So we should be thinking of what is coming next from now on. We should be thinking of 12 months, 12 months from now, 18 months from now and how we see this technology affecting the work that we do or the industry with which we work, we work to me and at runway, we always like to look back at previous technological breakthroughs. And to not go too far ago, we like to look at how the camera transformed many things. And specifically the film camera when the film camera came, we had, we have plays, we have theater the first time. The thing that we did was to point the camera to play, right? We record plays and then we thought about like we're going to distribute this place. We don't need to do theater again, we can share it with the world and, and that was a natural reaction because that's what we were doing at that moment in time, right? But years later we cinema was born and it was, it created a new form of art, a new form of expression, a new way of telling the stories. I think we're pointing to place today with A I, we haven't yet crack. What if this is going to be created in the future as new ways of telling a story? Maybe it's a combination of other areas that we're seeing that are getting developed like a IV are more immersive type of content that will merge at some point with what is happening with A I I would love to have a crystal poll today to say this is exactly what is going to happen. But I'm 100% convinced as it happened in the past that we might be doing more of the things that we do today. We might be creating movies, we might be creating stories in the way that we create them today. But this will unlock new possibilities. Meaning maybe we're going to start seeing something different than movies, something different than maybe touches the immersive or maybe touches of the game development. And for that also, we're going to start seeing more different distribution channels of where this is going to happen. So to me, it's really exciting of what's going to happen. And I say, with a lot of optimism, going back to your question, what has it changed in media and entertainment? I think not enough, I think we, you know, as, as we all, I guess agreed on, it's like we're at the very, very beginning of it, people are just starting to explore what, what is possible. And I think we're right in front of something that's called the colliding exponential, which is when you have new technological capabilities or multiple technological capabilities that sort of like, you know, have this like an increase, an exponential increase of what is possible. Uh I think that's what's happening right now. And I mean, right now, I mean, we're just starting to see, I mean, in our space, the use of native A I in film, I think is still, um there's only a few instances but the technological capability is showing that look, this is what's coming ahead. This is what you guys need. You know, we have to think about incorporating into your, uh into your business. And, but of course, it's not so it's not so it's not so easy. So it's quite challenging because you can't just say, well, we have a new technology, you know, especially when we entered the film business. And so the uh uh I mean, there's a lot of standards, there's a lot of um you know, processes that we need to adapt to, uh which is something that from a research and technology standpoint, it's not something that we're always used to. But um no, I think, yeah, we're really at the beginning and um there's going to be a lot of changes. I'm excited about the democratization of data that these models bring. So A I has been around for a while but you mostly found innovation at large enterprise companies that could afford data scientists and large enterprise it investments that could take advantage of that. And I think that what we're going to see is an explosion of more creatives, more personalization, more ability for smaller, faster moving companies or organizations or individuals to take advantage of the technology. And I can't think of actually a better industry, the media to show that explosion of personalization and creativity. Yeah, I'll go really fast. I believe that sort of hyper vocalization of the content people are going to be able to experience in a way that they never had before. I am looking forward to being able to tell stories to my Children in a way that none of us can do today, whether it's through a hologram or something like that or some sort of augmented reality. I think those are going to be the very good. Well, thank you all very much. Please join me and give a round of applause excited and.