Latest / Broadcast2Post by Key Code Media / A.I. For Creative Teams: Work Faster, Create More

Transcript

A.I. For Creative Teams: Work Faster, Create More · Broadcast2Post by Key Code Media

• 0:05
  Speaker A: Three, two, red one.
• 0:11
  Speaker B: You've got hours of deliverables, multiple versions to deliver, and someone asking for a shot from a project that you finished five years ago. They remember the shot, they don't remember the project name. And somewhere in the middle of all of this, you're supposed to figure out your AI strategy. There's plenty of attention on generated video and synthetic actors. But if you work in post, juggling interviews, promotional content, multiple deliverables, and finding footage you already own might be a more immediate problem than generating footage you don't. The workload keeps growing. Your time, staff, and budget, those tend to be a little bit less cooperative. I'm Michael Kammez with Key Code Media, and today we're looking at where AI can help in post production today. That not only helps creative teams find footage, but also accelerates editing and automates repetitive tasks. Plus what you need to know before connecting them to your facility. Because a slick online demo video is one thing, but actually getting it to work with your media, your editors, and your infrastructure takes a few more questions. Starting off generative AI versus Analytical AI. When someone says a product uses AI, that doesn't tell you much about what it actually does. It's kind of like when the client tells you that their footage is QuickTime. For this discussion, we can divide the work into two broad AI categories, generative and analytical. Generative AI creates content that might be an image, a video clip, a synthetic voice, or a visual effect. Adobe Firefly and Runway are examples of these tools that bring those capabilities into production workflows. There are still questions about copyright, consistency, and control. Can you get the result you want? And when a client asks for a change, a very specific change, can you make that change? Analytical AI examines content you already have. It can transcribe dialogue, identify speakers, write, recognize people and objects, and create metadata describing what's in your footage. Generative AI might help create a missing shot. Analytical AI might help locate a shot that's been sitting on your media server for five years now. For a team with a large archive, I'd give that second use case a close look. You've already paid to produce that material. Being able to find it makes it a lot more useful. Next up, AI, media intelligence, search, logging, and asset management. For years, finding media has depended upon file names, folders, and someone manually entering useful metadata. Or it's depended on that one person who remembers where everything is, hopefully if they're not on vacation or retiring. AI search makes this more like searching the photos on your phone. You you can look for a dog or a mountain without having tagged every picture yourself. In a media management environment, that could mean fighting footage of a particular athlete or locating interviews when someone discusses a specific subject or topic. For example, Iconic by Backlight can use AI services to analyze media, transcribe dialogue and create searchable metadata across your entire library. But there's a difference between finding a clip and and finding the moment you need inside of it. A tag that says football might get you to the correct recording. If that recording is three hours long, well, you still have some work to do. Time based metadata ties information to specific points within that media. Dialogue, speakers, objects and scenes can be associated with timecode, so the search can point you to exactly where the relevant moment is. Adding semantic search, which searches by meaning in context, you're less dependent on guessing the exact keyword someone used when they logged the footage. For a documentary editor, that could mean finding an interview passage about a subject when the speaker uses different words to describe it. AI Suite by SNS, Cara1 by ScaleLogic FlowAI from EditShare and still iconic by Backlight are among the companies bringing AI capabilities into into media management environments. They handle what I like to call top down AI that is the same AI results are for everyone on the team, not bespoke AI results on a per user basis. Now when you evaluate these systems, look at how precisely they get you to the material. Returning a long list of clips leaves more work for the editor than returning useful moments within those clips. Next up, AI assisted editing from raw footage to first cut how much of the first cut can AI handle? Once you've found the footage, you still have to make something that people actually want to watch. Take an interview based documentary before an editor can shape the story, there's transcript review, sound bite selection and the work of getting an assembly into the timeline. Adobe Premiere Pro's text based editing lets you select dialogue in a transcript and use those selections to to build a sequence. If you're working through a lot of interviews, that gives you another way to tackle the material without reviewing every single word through traditional timeline editing. Quickshire takes that idea further, using AI to identify potentially useful story moments. Create selects and prepare initial edits for professional editorial workflows. And now Quikshure understands video as well. Blackmagic's DaVinci Resolve has AI and telescript for for timeline assembly and multicam Smart Switch for camera selection. And of course Avid Media Composer has Phrase Find AI and dialogue searches and Script Sync AI for script based editing. Of course, AI is also being used for other things like masking dialogue, cleanup captions, translations, and reframing content for different platforms. But you know it and I know it. Number one, these tasks can consume a good part of your day. And number two, it's probably not at the top of your career enjoyment list. Am I right? Right now, the creative judgment still needs attention. A system can identify what someone said, whether that line belongs in your story, whether that pause should stay, or whether the reaction shot changes the meaning requires you, the editor, to evaluate that result. Now I judge these tools by how much useful work they hand back to you. The editor Faster logging and better selects can leave more time for pacing, context, emotion, or simply catching your breath. An assembly that needs extensive repair may leave you right back where you started. AI Infrastructure on prem Cloud or Hybrid? Here's a question for you more engineering centric folks. Where does the media go when someone clicks that AI button? Depending on the application, processing might happen on the editor's workstation, on a server at your facility, or through a cloud service. That affects the hardware you need, its cost, and, of course, how you handle confidential or unreleased material. Running AI on premises gives you more control over the content and, of course, your infrastructure. It can also require significant GPU resources, plus storage and network performance appropriate for the workload. Now, the processing has to happen somewhere. If it's in your facility, well, you're responsible for providing those resources. So cloud services let you use computing capacity without maintaining dedicated hardware for it. You still need to account for processing, though charges, bandwidth, and, of course, security. With a large media library, those considerations can become substantial. A hybrid approach may make sense. You could keep high resolution media on existing shared storage, handle certain tasks locally, and use cloud processing when you need additional capacity. Now the more useful question is, what fits your workload? How much media are you processing? How often can it leave your facility? What does it cost to run the work where you're proposing to run it? Those answers should drive the architecture next AI agents and MCPs connecting production workflows so far, we've talked about individual tools performing individual tasks. An AI agent could coordinate several of those tasks across applications. Now imagine asking your assistant to find yesterday's interviews, generate transcripts, group the recordings by speaker, identify useful sound bites, and prepare the material for you. That request touches several parts of a production workflow. An agent could potentially coordinate the steps through connections to your media management system, transcription and and editing systems. One technology supporting these connections is Model Context Protocol, or mcp. An MCP is an open standard that lets AI applications connect to external tools and data sources. An agent can use these connections to work with the systems. It's been given access to the connections, and the agents have different jobs. The MCP provides a way to access tools and data. The agent uses those connections while carrying out a task. Blackmagic Resolve is exploring agentic AI capabilities via mcp, as is Adobe, namely via their CEP and CXP extension and panel architecture. And similar ideas are emerging in media management and workflow automation. In fact, one of my first questions to all new technology companies is do you have an mcp? That's just how important it is. And of course there's still work to do around reliability, security, and permissions. For example, clean up the project is a request I'd want clarified before an assistant starts reorganizing a media server. We've all seen what humans can do with that instruction. Now for you engineering types, this is worth following because it could change how work moves between your systems. The value depends on whether those connected steps produce something that the next person can actually use. Next up, choosing AI tools, Integration, Security, and ROI let's first start with a problem that you can describe, such as our editors spend too much time Finding footage gives you something to investigate. We need more. AI just gives you yet another meeting. If search is the problem, evaluate media intelligence. If the team is buried in multiple deliverables, well then look at editing assistance or automation that addresses that work. Then check the integration. Does the tool work with your editing applications and shared storage? How does it connect to your media management system if it creates metadata, can other applications and your team members use that information, or does it stay inside the product that generated it? That matters when you're trying to make the whole facility just a bit more efficient. Now for cloud services, I understand where the media is processed, what data is retained, and whether your content could be used for model training. Those answers need to fit the material that you're working with and measure the result using real production media. A rough cut generated in 30 seconds looks impressive if the editor spends three hours correcting it. Well, those three hours need to belong in your calculation. Include the work required to get the media into the tool and the results back into your workflow. Measure whether the team gets to a usable result faster. I'd start with one bottleneck, test a tool against it, and then expand from there if the results justify it. AI gives post teams several ways to reduce repetitive work from finding an interview answer to preparing, selects or handling routine finishing tasks, not replacing the creative process. Whether those capabilities help your team depends on how well they fit the work and connect to the systems that you already use. At kiecode Media, we help organizations design, build, integrate and support the av, broadcast and creative systems behind those workflows. So if you're evaluating AI or upgrading shared storage, or modernizing your post production environment, visit keycodemedia.com to connect with our team. And I'd love to hear what's working for you. Which AI tools have earned a place in your workflow? And what's one task you'd be happy to hand over? Tell us in the comments if this was helpful. Give the video a like and subscribe to the KeyCode Media YouTube channel. I'm Michael Kammes. Thanks for watching. AI is quickly becoming part of just about every creative workflow. But figuring out where it actually adds value. Value is a much bigger question. So today we're going beyond the latest AI tools to talk about how creative teams should actually be thinking about artificial intelligence. Where is AI making a meaningful impact today? How should different types of creative organizations actually approach it? And as these tools become more capable, what happens to human creativity, ethics, and the skills creative professionals actually need to succeed? Joining me today is eve Bergquist, an AI researcher who spent more than a decade working on AI research and development at USC's Entertainment Technology center, or ETC for short. He's also the co founder and CEO of Quarto AI, where his work focuses on using AI to understand audiences, culture and how media connects with people. Yves, thanks for joining us. So Eve, you've been working at the intersection of AI and media for more than a decade. How did you first get involved in AI for media and entertainment? And how has your work at USC's Entertainment Technology center evolved over that time?
• 14:15
  Speaker C: So, story of my life. I've been an AI researcher for almost 20 years, but I think 11 years ago now, I walked into the office of my boss, my boss today, Ken Williams at the Entertainment Technology center, and I said, hey, you don't know me, um, but I think you should, you should really have a, a research project about AI. And at, the time it was, it was called data analytics, but it was already AI and you don't have to pay me. Which, you know, was the, the magic words for him. And you know, I'll get it started, you know, I'll get it funded. And then very quickly the thing got traction with ETC members and Ken was, you know, I'm very lucky that Ken trusted me, this crazy person coming in from the streets and it has. So I looked actually the other day, I Looked into my files for my etc sort of archive folder and the first event we had that was explicitly an AI event etc was 2017. It was like a workshop that we did with the studios. Of course they didn't wait for us to know about AI, but it really became a very singular focus of our work from the very, very beginning. We were talking about knowledge graphs and people were looking at us like are you insane? We were talking about deep learning, we're talking about all this stuff. And it was really kind of interesting how it evolved. It really started with the low hanging fruit at the time was like, you know, data in marketing. Right. It's like how do we make more data driven bets on marketing? So you spend a lot of money on marketing, like how do we develop a better understanding of our audiences? Then it sort of briefly, you know, we sort of did some work around AI in the development process to try to understand that sort of cognitive relationship between audiences and stories. We did a lot of work there that was really, really interesting. Worked on like Mr. Robot for NBC. Did a lot of really, really cool things. And you know, since I would say it actually started earlier than than when ChatGPT came came forward, it started about a couple years before when conversations became very, very intense around, you know, how to prepare. At the time it was still called a little bit like big data about how to prepare the studio for the age of AI. And obviously since November 2022 and November 2022 when ChatGPT was released, became an extremely intense conversation to the point where I used to have the AI portfolio at etc and now everyone has the AI portfolio. It's like electricity. It's like everyone. And I work more in sort of AI enterprise systems. But my colleague Eric Weaver has done work using machine learning and AI for years as well.
• 17:28
  Speaker B: Now you alluded to it earlier, but I'd like to talk about this a little bit is that you're also the co founder and CEO of Quarto AI. So I'd love you to explain what Quarto is doing today and how does that work connect back to your broader research around AI and audiences and of course media.
• 17:46
  Speaker C: Yeah, so my core area of expertise is really to try to understand, to use AI to understand storytelling and stories. Right. And I studied Cordo five years ago to essentially build products around a lot of the research that I had done personally around it. So Quarto is a company that mines the world's largest set of audience and market intelligence sources to understand, you know, in the war for attention where, where Is that attention? Right. Everyone's fighting the war for attention. Whether you're a politician, whether you're a brand, whether you're Disney, whether you're a creator, whether you're an app company. Right. Whether you're Delta Airlines, you need to get people's attention to your products and services. And marketing that involves understanding where culture is and predicting where it's going. And so we have these models that are our own models. We have about 200 models that we're using on this enormous amount of data to predict where culture is going in every single area. Beauty, pharmaceuticals, obviously entertainment and media, food, you know, airlines, credit card companies. We have several dozen customers at this point. We're growing really, really fast. We've built an agent that is now something you can actually talk to that just is going to crunch the data in real time for you. You can, you can type, you know, you can chat with it, you can talk to it. And now we've gotten to something that I think is really interesting and super passionate about, which is market simulations. So once we have all of this cultural data about all these segments, right, so we're no longer in the four quadrant world, we're in the million quadrant world. So we have data about all these segments, what they like, what they want, you know, their behavior, their emotions, and very, very deep psychographics. We can actually present to them different iterations of a product, including entertainment products, and see how the product resonates with different segments. So we can build these like, you know, 100,000 people simulations with this again, real life data that we're collecting in real time to understand in every single industry, possibly that is consumer industry, what kind of products are going to work with what kind of audiences. And I'm really excited about this because that is AI applied to complex systems, which we are one of the only companies to do. And I believe 100% is the next big thing in AI. So the current LLM or the current models that people are using, cloud, et cetera, are not very good at managing complexity and predicting complexities. I think the next big thing in AI is what we call agent based modeling, which is what Quarto does, which is very, very effective at representing a pretty complex systems. So super excited about that.
• 20:51
  Speaker B: So I would assume that then you've got to be pretty jazzed about the concept of harnesses coming out where you're specifically being able to craft different agents to address specifically what you're talking about, which is not just a generalist, but very specific agents that are role based.
• 21:09
  Speaker C: Right. So that to me is so number one, one of the most exciting things out there right now. Number two is really going to be the thing that my team and I work on at etc. Right? It is how do we create a media in IT and specific harness for agents to operate in, in a way that's, you know, safe, transparent, respects all of the internal, you know, access controls, all the, all the legal agreements that, that exist within the studio and things like that, while enabling a lot of autonomy. And we're starting with localization and versioning because that's the area that we think A is, is a huge pain point. B is sort of pure manufacturing type stuff. It has some creative. Right. But it is, it is a lot of it is really manufacturing and C, a lot of the models, a lot of the end models for it are actually really, really mature. So that notion of building an industry specific harness, which by the way, cimte has a big role to play and so ETC and cimti are working together is super exciting to us because it is, it sort of reminds us of the work that ETC was created to do around digital cinema. The interoperable master format which we got an Emmy from last year. This notion that ETC is the place where we discuss and create these interoperable systems and standards for the industry to really deploy to accelerate the deployment of technology. In this case, sort of autonomous agents. Super excited about it.
• 22:50
  Speaker B: Let's shift a little bit to creative teams in particular. So where are we with AI for creative teams today and what development should editors, producers, designers and other creative professionals be paying the most attention to?
• 23:07
  Speaker C: So there are so many tools out there, so many models and
• 23:14
  Speaker B: so many grifters
• 23:15
  Speaker C: and so many different. Yeah, so many grifters and so many types of creatives. What we're seeing is sort of so taking it, you know, sort of boiling it down to the creative process. We're seeing a lot of sort of personal behind the scenes usage of AI in creative aid. Right. Creative brainstorming. Right. A lot of creatives will use LLMs to bounce ideas off of. So it's a really, really good tool to get you from zero, from knowing zero to something. Right. It is horrible and horrendous at actually creating something that is new and interesting and, and reflects your own voice. But if you're in those really initial stages, it's actually a very, very useful tool to kind of frame your mind around something. We're seeing a ton, a ton of experimentation around previs and creators coming to producers with actual scenes and the look and feel of the piece, which is really awesome if you think about it. I mean, I don't know if you've ever seen a pitch a PowerPoint presentation on a Creat A project from like 10 years ago, but it was pretty bad. Like it
• 24:37
  Speaker B: looks,
• 24:38
  Speaker C: it
• 24:38
  Speaker B: looks archaic now.
• 24:40
  Speaker C: Yeah. And not only archaic, but, but it didn't really like good. I mean it was like kudos to these, to these folks for selling a project based on so little information. Now you can actually come to the producer with your film almost done or a mock up of your film almost done. And I think that really makes, makes a huge difference. Again, serves the creative process. There is increasingly like straight up, you know, virtual sets. You know, my friend Ryan Kavanaugh is company Acme vfx, you know, now has a couple of film projects that they've completed using this mix of, you know, real actors and virtual sets using AI that honestly, that looks incredible. You know, we'll see if audiences resonate with it. And there's the part that I'm personally really, really interested in is really post production, right? Because there's so much, there's so much like repetitive, straight up dumb, like mind numbing work in post production that people in post production VFX will say, Hey, I 100% want to automate this. I hate this stuff. Right? So that's the part that I'm really, really interested in because a lot of so the creative stuff needs to remain the creative stuff, but there's just a lot of opportunities to automate tasks. Not just jobs or work, but there's tasks. And like no one likes rotoscoping. Like name me one person that's like, I really enjoy rotoscoping. Rotoscoping needs to be AI driven. That's it. And there's how many of these tasks are out there? Thousands of them, right? So I am super jazzed about AI overall in all of these steps. As long as it doesn't step into the creative process because that's just the stuff that it just can't do and it's just awful at. But in the manufacturing aspects of content production that are going to make the industry move at the speed of culture, look, there's no reason for a movie to take three years to make. That sucks. That's just horrible, right? So if we can boil it down to like a year, man, I would be so happy because then there's so much more we can do. If we can take, you know, forget about cost. I just want to talk about like timeline. If it Takes one year versus three years to make a movie. I would be over the moon happy about it. I think that's really where we're going.
• 27:31
  Speaker B: So you mentioned rotoscoping, which I agree with you. It's mind numbing work and I think it's something that we would love to see more of. I mean despite the fact we've been using some form of AI for rotoscoping for quite a while now, I think we need, we need even more of it. But what other technologies or applications aside from just tasks, right. Are having the biggest impact right now? Is it more on the analytical side? Is it more on the gen AI side? But where are you kind of seeing the most improvements right now and the best use cases of AI?
• 28:06
  Speaker C: So we're seeing a lot of really, really interesting use cases in semantic search of content. I'm getting. It's not like I, and I totally get it. Like I was, I. I'm shooting video for my son's football team. I had to tag over the weekend a thousand clips with the name of the player. It was like, it was unbelievable, right? So AI now can really help with that in a big way. So a lot of the archives now in the studio system are searchable semantically, which I think is really awesome, which accelerates a lot of the work. A lot of upscaling, de aging, re aging, a lot of stuff in the VFX part of the pipeline, which I think is very, very exciting because it just again makes people move faster. I would say from my standpoint, I'm very personally interested in. So it's cool to automate certain parts of the production, post production process. But for me the biggest question still remains. Why do people liked the Nathan Fielder movie so much? Right? Versus another, right. What movies should we make? What stories should we be telling? How should we be telling them? How do we make sure that Hollywood understands culture again in all of its complexity in a way that it can drive it again? That is my number one priority is to use AI to help the industry really, really recreate this sort of very, very deep understanding of culture so that it can lead it again. Right? And that's really. Culture has become again we went from the four quadrant world into the 10 million quadrant world, right? And it's, it's extremely fragmented culture is very fragmented culture is very high velocity. I mean I've written about this a little bit. It is virtually impossible not just for Hollywood, but like Coca Cola, Nestle, Unilever, which Lego, which are all customers of ours to really surf the cultural wave anymore because that wave is divided in so many different little waves and it moves so fast that making sure that the people who are making media content are making content that is relevant and inspiring and break through and new and gets us back to the Hollywood or the advertising of the 70s and the 80s. That's what I want. And to do that, unfortunately right now you can't do that without data because there's just so much going on everywhere. There's no way that a single person can really truly understand what culture is.
• 31:10
  Speaker B: You bring up a good point. Now I don't want to ask for any of your ip, but I did want to not challenge, but you had mentioned, and I think a lot of us feels this way, that AI isn't creative. Right. It's, it's regurgitating to some extent. So I'm curious how you're able to balance the. Let us harvest the different data center, pardon the expression data centers, the different data repositories online to find out what that cultural wave is. But also not just rehash the things that are being put in these same data repositories that could kind of cloud things into making it the same thing over and over again.
• 31:51
  Speaker C: Absolutely. So which is why we're now telling studios and advertisers what story to tell. We're saying, hey, these segments are the fastest growing segments in Gen Z today. And do you know what the fastest growing segment in Gen Z today is? It's art.
• 32:07
  Speaker B: No, it's what?
• 32:08
  Speaker C: Art. It's art.
• 32:10
  Speaker B: Excellent.
• 32:12
  Speaker C: Museums, Art history. Fastest growing area of Gen Z Co. Doesn't it blow your mind? That blew my mind when I knew this. It's not anime, it's not even entertainment. Entertainment's number two. It's still pretty close. But something's happening in culture right now. And so what we're doing is we're not telling people what to do, what stories to tell. We're saying hey, these segments are growing really really fast. Like horror. Mangas is one of the fastest growing areas of entertainment right now. Make content for it. Like people love this stuff. Multi genre content. We're not saying write this movie. We're saying hey, if your movie has is at the intersection of four or five genres, it has 120% chance. Like the odds of it of being a success are 120% higher than if you're three genres or under. So oddball stuff. Dunnie Darkos make more Donnie Darkos. We're not saying hey you should do it like this and like that and you should have a dog in it. Whatever. No, no, we don't want to do any of that. First of all, those models are garbage. But we want to say, hey, we want to push creators to be more creative. Because what our data shows is the formulaic stuff doesn't work. What works. Nathan Fielder. That's what works. Backrooms, that's what works. Obsessions, that's what works. People want to be surprised. In fact, if you look at the data, there is a very high correlation between the volume of mentions of the word surprise in audience conversations about a title and the financial performance of that title. Why? Because people want to be surprised. They want culture to be dangerous again. They want Hollywood to be dangerous again. They want stories to be dangerous again. So that's the part that really excites me about what we do. And the moment that we're in is because in this moment we are probably standing on the cusp of a complete creative revolution back to the 1970s. And so, and I'm here for it and I'm really excited about it, and AI is going to play a role in it.
• 34:39
  Speaker B: Let's talk about another topic that we run into every time we bring up AI, and that's ethics. And I know obviously you've been writing about ethics and AI for, for quite a while, even before the current AI boom. So as AI becomes kind of embedded into more and more creative workflows, what are some of the biggest ethical considerations that creative organizations need to be thinking about today when they're using AI?
• 35:03
  Speaker C: So there's a ton of them, you know, I think so there's something that everybody's talking about and something that no one's talking about. Right? So I think if you're, if you're dealing with AI models, your number one responsibility, ethical responsibility, to really understand them. So what goes into. And it's like food labels, right? When you're eating something, you want to know what you're eating. So we are actually working with studios and technology companies to establish what's called model cards. Model cards are essentially a food label that a model making company will share with a customer, any customer, in this case, a studio, for them to really understand how the model was trained. What are the identified biases tested against a certain, a bunch of bias standards and stuff like that. So I think understanding AI as a whole, understanding the models that you're using are really, really important so that you can make really deliberate decisions. I think making the models transparent on the technology side, that's for the tech industry to make them really transparent and auditable. Super, super important. Again, we need to know what we're consuming, how it was trained, how it affects our decisions and stuff like that.
• 36:24
  Speaker B: Just to follow up on that, what is the realistic probability of that? I mean, if we look at some of the players right now, the clods, the OpenAI's, et cetera, and what are the chances of them saying this is what we've indexed, this is what we've looked at, not only from a potential litigation standpoint, depending on how you look at it, but also them actually keeping an exact record of what everything was trained on. What do you think the realistic probability is of something like that?
• 36:53
  Speaker C: Well, that's a very, very good question and I'll tell you why it is going to be very high if the industry speaks with one voice and says we need that and we're not going to use you if you don't have it. Oh, and by the way, we have all these open source models over here that are really good, that we can
• 37:12
  Speaker B: tell us
• 37:13
  Speaker C: that are transparent and auditable and we know what's in them. So it's like if you want us to use your stuff, we need to know what's in the stuff and we need to have an agreement about how it was trained and everything needs to be kosher and everybody that needs to be compensated needs to be compensated. But I think we're in a really interesting moment right now in AI where all these really, really big companies just have debt. Like there's just absolutely insane amounts of debt. There's very credible competition for open source models and smaller companies that are more nimble, that have very high performing models that are here to play, they're not here to tell the media industry what to do. And they are very aggressive, very hungry and they're willing to do anything to get their models in, including in especially open source models. And so I am actually very optimistic that we are now the power dynamic between technology and entertainment is shifting back to what it should always be, which is customers first and creators first. And thankfully the open source revolution is getting us there.
• 38:28
  Speaker B: This may be more of an economics discussion at this point, but as AI is making content easier and cheaper to produce, do you think that the value of human creativity becomes more valuable and less valuable? And just as an example, we look at what someone may have been making hourly 30 years ago as a film editor and then we had NLEs and there was this false equivalency of well, the NLE costs less, so you should be paid less. And then we fast forward 10 years to the point where we had NLEs that were $1,000, right. Which is vastly different than the hundred thousand dollars Avid Systems or your Moviolas. So we saw kind of the price drop as it became more democratized. So do you find that there's going to be less value placed on human creativity as this technology becomes more democratized?
• 39:21
  Speaker C: Two things. One, human creativity is not replaceable. Uh, you have to think about like I think about this a lot when, when people go to the movies and they buy a ticket, which for a lot of people is very expensive, what do they buy? What are they buying? They're buying the work. Uh huh, yeah. Trying to get someone to pay like 25 bucks plus popcorn or whatever. So like $100 for like a push of a button, get an AI AI film, like good luck, right? Because they'll pay. Sure, they'll pay like three bucks, right? So it's like that's why AI microgenre works because like the value exchange is so low. But when people pay for entertainment, they pay for the work, they pay for the training, they pay for the, the excruciating years and years and years and years of hard work, right? Of not going out with your friends because you're working on your craft. Like all this stuff, that's what they're paying for. They're paying for the pain, they're paying for the sacrifice that leads to genius. That's what they're paying for. I am 120 million percent certain that human creativity will play a bigger role not just in entertainment, everywhere. And I think the jobs of the future are those who help all the humans out there get deeper into their own selves and tap into their own creativity more. That to me is going to be a huge job moving forward. The second thing is there is no question that this technology revolution will displace jobs, will displace value with displaced skills and stuff like that. And I want to be really. We have to listen to these people, we have to listen to their pain, we have to listen to their fears. I mean look in Hollywood they're our friends, they're our family, right? Their fears are completely legitimate. And you know what? We need to listen to them. We need to take them very seriously and we need to help them. We need to help them understand AI upscale themselves and give them all the tools that they need to, to be able to exist and thrive in the next world if they choose to do so. Right? And that's really the stuff that, etc we feel very, very strongly about, we're actually about to release a really big, ambitious education program for the entire media industry about AI that's going to be free to help them, to help everybody that wants to learn, understand and learn and upskill themselves so that they can, you know, thrive in the next world that we're in.
• 42:15
  Speaker B: I don't want you to give away all your secrets and everything that's going to be in that program. But for folks who may be an editor or a producer, a designer, a filmmaker or other creative professional listening to this conversation, what skills would you be developing right now to remain valuable as AI becomes a bigger part of the creative process?
• 42:35
  Speaker C: So it's a, it's a, it's a tough question because it really depends upon where, where you are right in the, in the pipeline. But I would say the thing to do above anything else is to experiment. Use it. Experiment. Use it as much as possible. Learn about it. Do whatever you can to learn about it and understand it at a deeper level, to be able to understand how it affects your work and which parts of your skills are getting taken away and which part of your skills you need to double down on. Right. Again, it is a very complex picture. Very few actual jobs will get automated out of existence. There will be a bunch of them, unfortunately. But most of the positions that are in the media industry will have a lot of AI in them and will also demand more from skills that are human skills, creative skills that right now are diluted in a lot of dumb tasks that we all have to do, right? But a lot of these dumb tasks being automated, then you can double down on the stuff where you are making a contribution as a human, as something, you know, with your personal creative voice. And so I think that, to me is, is exciting. I would say what I, what I tell people is obviously leadership skills, salesmanship skills, relationship skills, all of the humans, the human stuff, I would say, compute what I call computational thinking. Computational thinking is really, really interesting. Computational thinking is the ability to think computationally about things that are not computational problems. Like, how do you think computationally about racism? Take a problem that is a linear problem and try to port it over and try to look at it from a computational standpoint. I think that ability to apply not just AI, but computation in general to problems that are not seen as competition, like, for example, narrative and storytelling. Think competitionally about narrative. Think competitionally about storytelling. Today, that's the most important skill, and not just in entertainment, but everywhere.
• 45:14
  Speaker B: Eve, we could talk about this all day and we've got questions. Piling up. So I think we're going to wrap here and if you could hang on for a few more minutes, we'll get to some live questions. How does that sound?
• 45:24
  Speaker C: Absolutely. Let's do it.
• 45:26
  Speaker B: Excellent. Hold tight.
• 45:28
  Speaker C: Bye.
• 45:29
  Speaker B: Thanks for watching. Broadcast to post. Don't forget to follow keycodemedia on social and contact us about your projects at keycodemedia. Com.