RealmIQ: SESSIONS with Mark Turner

Welcome to Season 2 of RealmIQ: SESSIONS with Mark Turner who is President, Entertainment Technologists Inc, and Director of the Production Technology Program for Movielabs.  Mark was the lead author on the MovieLabs 2030 Vision for the Future of Content Creation and now actively works with the studios and MovieLabs to evangelize the Vision, build transformative technologies that support it and develop standards and systems to accelerate adoption.

Topics Discussed

1. AI in Hollywood

• Current Impact: AI's role in transforming the entertainment industry
• Initial Reactions: Fear and skepticism due to IP issues, deepfakes, and digital twins
• Shift in Perspective: Acceptance and realistic view of AI as a tool for enhancing creativity

2. The Role of Generative AI

• Generative AI in Content Creation: Benefits and limitations
• Practical Applications: AI as a series of tools to assist creatives
• Impact on Production: Streamlining processes and reducing mundane tasks

3. Industry Adoption

• Technology Adoption Curve: Hollywood's slow but eventual embrace of new technologies
• Recent Developments: Integration of AI tools in Adobe Premiere and Avid
• Workflow Changes: Evolving from individual tasks to integrated workflows

4. Efficiency vs. Efficacy

• Efficiency Gains: Reducing time and cost with AI tools
• Efficacy Improvements: Enhancing the quality and creativity of work
• Balancing Act: Maintaining quality while leveraging AI for better productivity

5. Proactive AI

• Future Vision: AI as a proactive assistant rather than reactive tool
• Use Cases: AI managing schedules, predicting outcomes, and improving decision-making
• Trust and Data: Importance of clean data and trusted AI systems

6. Industry Transformation

• Business Transformation: Embracing AI for long-term efficiency and innovation
• Human-Centered AI: Keeping creatives at the core of AI-enhanced workflows
• Training and Adoption: Educating the next generation and upskilling the current workforce

Selected Quotes from Speakers

1. Curt Doty:
   • "AI has taken over the technology landscape. It's trying to take over the entertainment landscape."
2. Mark Turner:
   • "Generative AI was never going to write a movie from scratch. It was only ever going to be a series of tools."
   • "We're a really inefficient industry. The way we make content hasn't really changed in 100 years."
   • "You can't just say, I'm going to drop AI into my business and not expect ripples in the pond."
3. Curt Doty:
   • "I call it creative-centered AI. You need that creative person, not just a human."
   • "It's really about the output and being smarter with how you're putting these things together."
4. Mark Turner:
   • "AI can help transform the movie business, which is in a crisis right now."
   • "We need to find those metrics

Transcript

Hi, I'm Curt Doty with Realm IQ. This is our podcast, Realm IQ Sessions, where we talk about everything AI with international AI leaders. And in today's case, I'm talking with an old friend and colleague. Please give us a follow. And today's guest is Mark Turner, who's president.

of Entertainment Technologists Incorporated and also director of the production technology program for MovieLabs. Mark was the leader on the MovieLabs 2030 vision for the future of content creation and now actively works with the studios and MovieLabs to evangelize this vision. Build transformative technology that support it and develop standards and systems to accelerate adoption.

Man, you are in the middle of Hollywood. So, Mark, great to have you on the show. How you doing? Good. Yeah, good. Thanks. Yeah, I have to say, you know, we have a little history here, actually some really good history. I've met you and worked with you when I was on the studio side in that. A  little race we called HDDVD versus Blu ray  and you were with Microsoft and I was helping with the studio coalition and that's where we met and then continued on some really cool projects on the agency side after that.

So I see a whole bunch of people don't even realize there was a format war about Blu ray, Blu ray one.  Yeah, it was spoiler for those that didn't know blue wave. It was the better format for sure, but whatever Long time ago, but we were doing some cool shit for sure And then we went on to do more cool shit for sure with you and Microsoft around Windows Phone 3 and Silverlight technology right and first interactive movies on smartphones That was some cool stuff.

Still unbeaten. Yeah. Anyway. So a  lot has changed. Certainly AI has taken over the technology landscape. It's trying to take over the entertainment landscape. I think it's one of the  last industries to really embrace it. Because of all the fears around IP and deep fakes and Digital twins and all those scary things, which I think could be worked out.

But what are your thoughts on ai from your lens, through your lens and expertise?  So let's just narrow that down a bit. So we're talking gen ai, right? I mean, ai, I don't think anyone's, that's really Yeah. AI around for 15 years we've been, we've been doing machine learning, learning and AI for years in Hollywood and it's fine.

Gen AI just set off a few, radar triggers for people in a deep irony in Hollywood, let's face it, right, because  This is Hollywood. We're the ones that make up stuff that don't really exist, right? That, that's the purview of Hollywood is to trick the audience to seeing something that isn't really there.

So when someone else comes along and says, I could do that too, then everyone goes, whoa, wait a second here. Hey, calm down. You don't want everybody being able to do this. This is magic. so There was a bit of a knee jerk reaction last year and it was like all you know, scary new technologies that come along.

Set off some overly protective senses. I think we're now, 2024, we're now at a point where people are actually a bit more reflective. Not all this, all the legal action around strikes and stuff has been settled, but a lot of the actors have been settled and the writers are back writing. So, so it's calmed down a bit.

But also there's this, I think, I'm hoping now there's a sense of realism that.  Gen AI was never going to write a movie from scratch. It was never going to replace a director or an editor. I don't think I want to watch a movie that was written by a robot and directed by a robot. It was only ever going to be a series of tools.

That would make people's lives more enjoyable and, and more able to get more creative stuff done, not less. So hopefully this year we now got people being a bit more reflective about this stuff and things are more grounded. Because I do think it's a very empowering technology. And we need it, you know, we're in a, we're in a very expensive,  risky business.

You know, making content is almost unlike any other product. That anyone makes, right? It's you know, if I went into forward. And I said, Hey, I want to make a new car. But I'm only going to make one of them but I want you to give me a hundred million dollars and I'm not going to do any market research.

I want you to give me a hundred million dollars and I wouldn't even guarantee you're going to market the thing. And so you spend another a hundred million dollars marketing it. But you're not going to get to see it until I finished. And when I finish. I'm just going to drop it on you and you're going to take it out the rest of the world and everybody's going to see it like that's crazy.

Yeah, and I'm going to build this team. We're not, we're going to build a special factory just for that one product. And I'm going to make a team that probably hasn't all of them work together before, but we're all going to come together and we're going to do it in 18 months and we're going to give it to you.

And then we're going to all disband and we're going to disappear and never do it again. You know, that's kind of nuts, but that's the way we do things. So it's kind of scary and it's risky and it's very expensive to get it wrong. So there's a very risk adverse culture to make sure that you work in this town again. 

 bUt with that, you know, we have adopted new technologies eventually.  Yeah, we did go from analog to digital and everything is now, you know, not cut with a pair of scissors and there's literally not bits of film on the floor on the cutting room. So, you know, we, we make, we get there eventually, but it is a slow process where people go through technology adoption curves.

So, you know, AI is early in that process where people are very distrusting of it. But they're going to come around and they're going to go, Oh, wow. That's actually really useful. And it's going to allow me to take some of the risk and stress out of this business. So I should embrace it.  Yeah. And you know, with the  new releases coming out and Adobe premiere, you can see it, you know, You know, replacing some expensive rotoscoping just with the push of a button and you know, how cool is that?

I mean, it was such a laborious thing to do the fix it in post things. But now if they, if one editor can do them  with a few magic clicks it's pretty amazing. It's ironic that, you know, Hollywood actually created the dystopian scenario. Of what the bad side of a I might be certainly with the T two  scenario.

And and and and because of that, you know, they they've had this minority voice and In terms of global industry,  uh, even though they're a big industry, but all the other industries have like embraced it. And, and Hollywood with this vocal minority has, has been kind of putting it off for a year. I mean, the strikes didn't help, but you know, I, I think the union things have been worked out.

To some extent  is ironic, right in the, you know, there was a, there's been a lot of experimentation going on behind the scenes, right?  So it's not like a year has been wasted, but.  Much like other industries, there was, you know, the initial  flurry of chat, GPT hype, right? Oh my God, this thing is like a human and it will talk back to me.

So there's all these people piling in to say, well, what does that mean? There's a lot of R and D going on, lots of little skunk works projects in every company of all sizes going, what will this mean for us? So that's all right. That's, that's been happening across media and entertainment, but the Adobe announcements you're talking about, which came out this last week at NAB Are critical because that's the bit where skunk works and people doing explorations behind the scenes coming, you know, move into this more mature phase where the AI is now embedded in a tool set. 

And at the point you're in a tool set, then it doesn't require an  IT department to go and break things. It doesn't require a whole security team to come in and check what you just did, and who's got data, and have you got access to it, and all this sort of stuff. And that's part of a product.  And that's what it really takes for any product, any sort of exploration to get to that sort of maturity level, is it needs to get into the hands of people who can use it.

So that Adobe release, and it wasn't just Adobe this week at NAB Avid were showing Some AI tools where you can just ask it to take. media of a certain description and it'll drop it in a bucket for you and it'll do some of your, you know, basic blocking and tackling. Especially when you've got really big shows that are shooting with multiple cameras and you've got all this content coming off the camera.

All that's just getting shoved in buckets and it's being tagged. These tools allow editors to just go, you know, give me  scene three, five, Take seven, or describe it as, you know, the one where the guy walks in the background, you know, give me, we all remember that one. Can you just give me that one? That level of sort of humanity to go.

All right, I'm going to go. I'll scrub through all that content and find it for you and drop it in a bucket. That's where we're at now, where the tools are now able to do that and take Gen AI and sort of put it in a context for each role in production and what it means for them. And then AI is going to be a lot less scary this year than it was last year.

Right? So editors last year might be like, Oh my God gen AI is going to cut a movie for me. Right now they're going to go, Oh, actually now I see what it does. You know, in the Premiere Pro examples you're talking about, the most compelling one, I thought, was that it will Firefly will now generate some frames of video,  um, at the end of a clip.

So if, if the clip doesn't look quite like the end of the scene, or even the beginning of the scene, which is even more intriguing it'll just add a few more frames. I'm predicting what was happening in that scene. It was about to happen in that scene or create you some video and in a lot of cases that just what you needed to just finish that, you know, blend effect or whatever it might be that you needed.

So that's when an editor goes. Okay, now I get it. All right, it's not cutting everything for me. It's doing the crap I didn't want to do, like setting up a production, or making notes, or finding stuff for me. All the mundane crap that isn't creative. Or things like that, which is generate me some more media so that I can actually do my job better.

 You know? Yes. Eventually we're gonna get to the point where you will, if you're a small sort of independent filmmaker, then yeah, you probably can edit just by using your voice or just by using text. But that's probably not gonna be a challenge to anyone in Hollywood or  Madison Avenue or in the sports industry or anything, because.

You know, in those cases, you're still going to want to create a person who's doing that work for you. So, you know, Gen AI is very democratizing and leveling, and it will give these tools to everyone. But hell, I've had the ability  on my computer to create digital art for years. It does not make me an artist.

You know, you've got to have the talent in the first place to know how to use these tools to create magic. And you know, I don't garage band, but I, that does not make me a musician and having an AI tool in an editing suite is not going to make everybody an editor. It may make some people slightly better enough that they can get a YouTube video out, but then never going to walk into Hollywood and take anyone's job.

So. Yeah, that's the good point is now we, we can all take a breath and go, okay, so no one's doing a job this year.  Someone's gotta be in the middle of it and they have to know what they're doing. They have to have to have a vision for an outcome.  And yeah, I talk a lot about human. I used to talk a lot about human centered design around UX and now human centered AI.

But it's really.  I call it creative center. AI, you need that a creative person, not just a human.  You need the editor that has 20 years experience of, of storytelling and working the hard way to understand  how these tools can actually be used. They're going to have the expertise to, to actually push the format to, to new levels.

It isn't going to be a newbie. Novice that, Hey, look, I'm editing. I made an AI movie. Look at me. It's like, yeah, but  what is it? You know, it's like two minutes.  Yeah. It's like, it's not a movie. There's no talent in it. There's robots maybe. And there's some explosions. And so  it's really not an AI film. So and, and really it's less about making an AI film.

It's just using AI tools within these. Accepted. Platforms that we're already using like Adobe  and Avid  and just making your film,  you know, and Helping you but it's not an AI film at the end of the day. It's just it's like, you know before Avid What was what we were doing to make films before? Computer graphics and CGI.

What we're, we're doing, you know, you're doing  before the typewriter, right? Shakespeare didn't need a typewriter. He wrote with a quill didn't make the words any less effective. So this is just another tool, but you better be creative or, you know, able to tell your story in the first place or the tool is what's in your head.

And Shakespeare would have loved to have a typewriter. Come on. Right. And actually, you know, last year, you remember last year, this time last year the world moved so fast. This time last year, everyone was talking about prompt engineers. Yeah. And I find it ironic actually that, you know, actually prompt engineering or the need for having people translate what you want into a prompt that an AI will then be able to understand and then go off and generate for you that already tells you that there's an AI.

There's a trick in describing  the creative process in such a way, right? Because you need to be able to get Gen AI to do text to video, for example. You need to be really good at describing what it is that's in your head that you need to get out. And you'll find that, you know, a non creative person.  We'll sit down, you can soar it, right?

If you gave Sora to most people who are not filmmakers and say, all right, then now you've got this amazing ability to make a film. Go on then. And most people are not filmmakers ago. Why? I don't know what, I don't know what I want to make. I don't know what I want to tell it. I, I'm not a filmmaker. I don't know.

How about, you know, two golden retrievers in the golden field. Yeah, I  mean, it'll make for a great demo, but it won't make, it's not going to win an Oscar and it's not going to be interesting because You've got to be at least the filmmaker mindset to be able to go make this stuff. So, you know, I'm not,  I'm not worried.

I wasn't worried last year. That isn't to say that I don't think it's going to have an impact. I think JAI is going to have an enormous impact. And I think it's going to be very positive.  Because  we're a really inefficient industry.  There you go. That's it. It's  the way we make content hasn't really changed in 100 years.

Yeah, we, we cut digital files. You know, but we still mail them around the place. You know, there is still a script and it's still typed up. And then print it out on bits of paper and hand it to people on set with different colored pages. I mean,  and then people are taking a pen and writing their notes in the corner of the script, right? 

We got a long way to go before this industry is as automated and efficient as every other industry you're talking about. And we kind of excuse people on the basis, well, they're creative. So they're allowed to, you know, not use software and, you know, not use collaboration tools and do things in a kind of insecure way.

 But we've almost run out of lead time for that sort of thing, right? That sort of behavior kind of has to end on the basis that,  you know, in 2019, 2020, 2021, the problem was there were not enough creative people around to make all the content that everybody was demanding. All the streaming platforms like, we need more content, more, more, more, more, more.

And we couldn't make enough. And there weren't enough filmmakers and visual effects artists and all the people in the world that it takes to get this stuff together. So everybody was working crazy hours to get it done.  Now we've kind of got the opposite problem, which is we can't make too much, frankly.

The bubble has burst on that. Now a lot of the slates are being pulled back and we're going, actually, wait a minute. There's only so many hours in the day that people can watch. Video and also going to work and play video games and listen to music. So they can't do all this. So we were making too much content.

So now we're shrinking back. So we still have a need to actually be more efficient right now. Budgets are going down. Time is going down. We'll get more down in the last time. So then you look at it and go, well, all the inefficient stuff that we've been doing. You know, moving big files around the place waiting for someone on set to turn up because no one can start until that person turns up because they're, you know, they're, Uber didn't arrive because someone forgot to order it for them.

You know, the basic blocking and tackling, time wasting crap, rotoscoping, you mentioned it, right? Post production workflows for this, recreating digital assets because, you know, we are, we can't find The 3D asset of the table, so let's just have someone go make another one again, because we can't find it because it's such a mess back there in the digital archive like.

All these bad bits of behavior we've done by just throwing time and money at it in the last 20 years has got to stop, right? And Gen AI is actually the ability for us to say, actually, here's our way of getting a shortcut to being more efficient, right? We don't need five people in there tagging metadata anymore because we can have an AI that will go and just blast through and do all that for us.

You know, I can just describe.  to the AI what I was looking for and have it go off and cycle through this thing, right? I can have it go off and, you know, text all the people on the production and tell them that we're going to start 25 minutes late because Jimmy's Uber is late. So nobody turn up because, you know, there's no point in rolling in right now.

So all this inefficiency,  we hopefully can get to either reduce the time, reduce the budget, or more likely spend the time and the budget  On making better content in exactly it's, you know, kind of elevate the creative output, right? So it's more iterations or more, you know, better pixels on screen, more dollars on screen as a way of describing it.

Right. But.  That's ultimately what we want to get to, right? We waste way too much time and money on bureaucracy and crap and admin and we can get rid of all of that or a lot of that just by being  more  accepting of new technologies and embracing some of these great opportunities that they open.  Yeah, I mean, what's happening in other businesses is business transformation, right?

And business transformation happened in Hollywood. Once you start looking at it through that lens with the many inefficiencies that you just described,  all the fears go away because you're really not. Dealing with the doomsday scenario or what's going to happen to acting as a profession.

It's like, no, you still need actors. Okay. You still need stars, but all this other stuff where AI can help should be able to transform the movie business, which is in a crisis right now. You talked about scaled back slates. Well, yes. The demand. Might be less, but the expectation is still there that I still Okay.

You know a studio doesn't output 12 movies a year. Now they do nine,  but I still want nine great movies. Yep. You know? Yeah.  Something's gonna win an Oscar, Oscar for Yeah. So they, they gotta be great. You can't, you know, just because your budgets are cut.  doesn't mean that you're the quality has to go down.

You're just focusing that money on fewer movies and with the new efficiencies, maybe make them look better or be done quicker. And through this curve, this learning curve over this next year,  let's just say, you know, Studios could increase their slate because they've gotten more efficient by using these tools,  right?

That's the goal  Because it's not a zero sum game. It's a it's an additive game. You know, how can AI help you produce more, not just save money, but actually create better movies and create more movies because of these tools. And no, with no necessarily job loss, it's just really about the output and being smarter with how you're putting these things together. 

Yeah. And yeah, look at so yeah, take Microsoft as an example, right? I used to work there  15 plus years ago. anD there's a finite amount of software that an engineer can develop.  But  software developers are the, one of the, the first group that have leaned into Gen AI, right? Gen AI is now writing code for software developers, and it's the same sort of problem as described in a movie, right?

You can't just sit in front of a blank canvas and, and have it, you know, You know, create Microsoft Excel. You still need to be able to describe what it is that you want. So it still takes a sort of programmers type brain to be able to tell AI to write code for it, because you've got to know what it is you want to try and achieve.

But once you have that sort of brain, you can create a lot more code, a lot faster with Gen AI. And that has been embraced. By other industries for huge transformation and we're only beginning to see the beginning of it, right?  That Workflow that enables the whole company to go. All right, we're all going to embrace it.

We're going to try it right enables Companies would be faster and innovate faster, right? There is, the amount of software that's coming out of Microsoft and the speed of which they're taking a model that comes from open AI and turning it into a product in co pilot is like unheard of. When I was in Microsoft, there's no way in hell we'd put out a product in two months flat.

I mean, it just never happened. But now spitting stuff out super, super fast because they've got a team that's embraced it. And I,  you know, ironically, the software developers are the ones who'd say. Who are most threatened by Gen AI.  Because it does have this ability to write code in multiple languages immediately. 

But they're not freaking out and striking and running down the streets freaking about it. They're going, this is going to be a great tool for me. I'm going to use it. I'm going to be able to do much more amazing things faster. And I want to get on the AI team to do these things, right? You know, it's docking.

It's that sort of attitude that all industries need to have, or all individuals in that industry better have, which is I'm going to learn how this thing works, because otherwise, the person behind me. Who's coming after my job, they're going to understand how it works. So I better lean in and I better understand it.

So, you know, I, I sort of find that empowering and I hope other people do as well. It's not intimidating. You know, if you, there's a bunch of videos that are online that will tell you everything about anyone you ever want to know about AI. And if you can't be bothered to watch a video, you can ask it to, to give me the summary of the video for you.

It's just a, it's a magical tool.  But, you know, that sort of embraces mentalities, I think what we, what we all need as an industry, and then we'll hopefully take some of these barriers down and start thinking about a workflow,  right? That's the other bit that people are thinking about.   aNd  we're so I told you, you know, there's an R and D phase.

Now we're at the tool phase, which is great, right? So individual tasks. You can see how that's being applied.  But we're still not at the point where you got into and work for this being sort of powered or enabled by. Gen AI. And by workflow, I mean, you know, I've got a series of tools and a series of tasks that need to be done by a series of people.

And they may all be using Gen AI in their tool, but what would happen if I put all them together and how is my workflow going to change?  That's like the next step down the process. And that's probably 2025, you know, big opportunity. But, you know, I've, I've had this conversation with a few media companies where they're like, Oh, great.

You know, if we've got a great tool that will do.  You know, automated QA, for example, right?  It'll check a video and it'll check that there's audio and video there. And if it says there's five audio languages, I'm going to check that they really are English, Spanish, French. I'm going to check they all sync up with the, with the lips on screen.

And if there's a subtitle, then the subtitle seems to match the language. Sort of QA job that most, you know, most post production companies got people that sit there and do that all day long. Right? So AI can do that sort of stuff. Great.  Bye.  You don't yet have a workflow that can accept dropping that AI into that process.

And that's the bit where I mean, it's going to change, right? So if you drop an AI QA tool in the middle of a human process,  then you need to think, well, is the video in the right format for the AI?  I do want to change it.  The AI is going to give me some sort of output. It's going to give me some probably you know, percentage of accuracy.

This was 98 percent confident this one was good. This one I'm only 85 percent confident on. You've got a right. A processor. And then what are you going to do if it comes in at 75%? You're going to go. All right. Now I need a human to go take a look at that. Clearly, that's a problem. So now I'm going to add a feedback loop where a human needs to take some of the stuff that came out of the AI that got rejected.

And look at that and say, Oh, what do I think? Oh, I think that was good. And that was bad. It was right. And then that's going to feed back into the AI. So it continues to get better. So what was a, I've got an AI tool now becomes actually I need to re look at my QA. Pipeline and I can make it more efficient,  but you can't just say, I'm going to, I'm just going to drop AI  into my business.

And not expect ripples in the pond, right? And that's the bit that people have not got their head around yet. So we're on individual tasks until you really get whole companies enabled and powered by these tools all the way through. We're not going to get to the level of, you know, efficiency that everybody wants.

And that's not specific to M& E. That's, you know, that's all industries. That's where everybody's at, I think. Yeah. At 24.  That's what I've been  Writing to and preaching about and talking about in the various lectures that I give is how to build an AI operation, you know, how do you operationalize a New process you you you need evangelists within your company to say, you know I'm gonna explore AI on a real project take a real project.

There's a tense. There's ten steps to it I'm going to take three of those steps Figure out how AI can just do that and eliminate the human out of that, those three steps and then figure out what platforms work best, right? Multimodal and create that workflow of that one project. Now with AI interjected, take all those learnings, share that with your colleagues and your boss and your peers and say, this is how I did it.

I was efficient in this way. I was more creative this way. And here's this documented process. Okay. Then you take that. And by the time a lot of these companies figure out who their evangelists are and start to say, okay, we've, we've, we've now adopted it into various tasks and processes. The software developers are kind of  just like Adobe did.

It's like, yeah, we're going to take all this stuff that's happening out there with open AI. We're just going to drop it into premiere. And make it part of our software that you're familiar with. And then all of a sudden it's a game changer and there's less experimentation. Now it's actually learning the tools on the platforms and software that you already know.

And when, when ChachiBT came out  November, you know, two years ago,  I knew it was going to become ubiquitous because it was such a simple interface.  And, and that's part of adoption. It's easy under Pete, you know, people never understood what web three was or did or what the fuck. Right.  And so here, no, I just type in something and it gives me results, immediate gratification.

Okay. Well, how can I use that for some of my work? And you just start asking and it starts doing stuff and you go, oh my God. So I knew it was going to be ubiquitous. I knew, Big tech was going to say, Oh, it's going to be in PowerPoint is going to be in word. It's going to be everywhere because big tech was going to be all over it.

And so we're at that point now it's ubiquitous. It is less of going to find the right platform. It's really looking for the little sparkles in a field that says, you know, can I help you with that? You know, can AI help you with that?  It's already there. So, you know, the learning curve  diminishes and the adoption increases  and, you know, that's where we're at.

And I think it's very exciting. And I think we're at the stage now where business transformation can take place versus just. bitching about the fear factor and you know, get on with it and figure out how it can really change your life and your work effort and your, you know, that, that AI operations, I think you got the you, I think you're right on it.

I think the  the bit, the other hard part, the bit that the people are missing is.  How am I going to measure  whether this is succeeding or not, right? So in you try investigating anything, right? You should have a business case for it. And everybody will say, well, what's your expectations going to, you know, drive profits going to lower costs.

What is it going to be? Why, why are you doing this? And what we need to find is those metrics that allow us to go. All right. I've tried it. And there's like three areas that you just talked about, right? And I've tried it here, here and here.  And I can actually prove that it did improve things, right?

Because. As we talked about, it may be disruptive, but it may not make it better. We don't know that. And if, even if it does make it better, how much better? Right? There's a, there's a cost to running AI. There's a cost to the planet. There's a cost in real numbers. And it, it better be delivering on the sort of exponential productivity increases we're expecting.

If this is just incrementing, you know, 1, 2, 5 percent you. This, then we missed the point, right? This is supposed to be delivering in leaps and bounds worth of more efficiency. And I think it will,  we just have to make sure that we actually bound this with some good metrics to go. All right, before I  pour fuel on this fire.

I'm going to prove it. I'm going to try it here and here and here. I'm going to measure it and I come back and say, all right, our assumption going in was it was going to do this. It actually did this great. We're going to move forward the next one,  but it's not all just go in and expect us to fix every problem that it's going to have.

And it doesn't.  A lot of  the times when AI  fails is also when it sort of makes stuff up is when it's got bad data in the first place. So, you know, in a lot of the use cases and demos and people show it's very clean data and very controlled and modeled environments. And I don't know anyone that's got clean data.

And I did a lot of work with a lot of companies, right? Most people have got. a mess. And they'll go, oh, it's all right. You don't need structure. AI will figure out the structure. Yeah, true. But it'll do a damn sight better and make stuff up a lot less if you can give it well structured clean data on the way in.

Because you're just begging for it to come up with something weird if you give it, you know, some spurious data parts. So, you know, Only as good as the knowledge base, right? Yeah, you know, that didn't change, you know. Garbage in, garbage out didn't change. It's just that this time you may get Garbage accessed too.

There's like three things. Yeah. Human input and the computer output, but then the garbage that's in the data set. Right. That's garbage. There's garbage all around. Right. You may be using a model that had garbage in its training set which is, you know, one that we haven't touched on is the training sets.

But,  and, you know, there are, there are now companies doing the right thing about licensing data, and I'm a big believer in this, and I've got an investment in some of the companies that are doing this. But, you know, it is very clear that we need to train data and understand Where certain models got their information from a so they're not ripping off people's copyright, but be in case there's a mistake and we can trace it back again, right?

There is you know, we had all these problems with facial recognition years ago, where it was, you know, miss identifying people because it didn't have the right data set, which have enough  diversity that matched. Human faces and you can see exactly the same thing happening again. You can say, all right, well, we trained on the whole internet.

Well, the whole internet biased in itself and the whole internet has got the worst and the best of humanity in it as well. So that is not necessarily a great data set, right? Especially when you're asking it something that is very Precise and nuanced and subtle. You know, like art. You can't just say, I'm gonna, I gave it all, all of art, and now I expect it to understand the subtle nuances of this.

I would rather give it a more refined data set and say, alright, just, just look at this. Now give me the answer. Don't look at the whole damn internet. So there's a lot of, we've got a lot of refinement to do as we go through this process.  You talk about, you talk about efficiency and I, I think that's been the bellwether for the past year.

And I think we need to move from that to efficacy. I also write a lot of that because it's like, great. You're more efficient, but is the work better? You're right. Is it more creative or just make shit faster? Pardon my French, but. You know, if you can change the mindset to, and this is where the human comes in, the creative, the true creative comes in is like, no, I, you know, this, the work that I'm creating is more effective than I'm more effective, not than I've ever been.

It's not that I can do it faster. Yeah. That's a, it's a mind shift that has to happen now. And so when you talk about KPIs and what are the metrics, there's also that metric of.  Was that a funny spot that it wrote? Because if, if no one laughs, you can't say, Hey, I did an AI comedy spot. It's like, but it wasn't funny. 

Ultimate efficacy test is like, did it make me laugh? You know, it's like no, you know, it was a bad dad joke that fell flat. So there's two other, there's two other bits that you made me think of.  Around the efficacy thing, right? So one is that  AIs don't complain.  about, you know, living standards and the quality of their pay and everything else.

Which is great because we can make them work  all the time, 24 seven, right? Humans don't want to do that. And they shouldn't. And the entertainment industry  is a classic example of a horribly stressful place to work.  Very long, you know, when you're on a production, very long hours. It's very stressful.

Lot of pressure, a lot of anxiety, a lot of shouting and screaming and you know, it's, it's hard. It's really hard. And therefore it burns people out and the quality of life is not great. You know, there's a lot of people do not get to see their kids when they're working because, you know, You know, they're too busy working and you're going to take it when you can get it. 

So having an AI that's offloading a whole bunch of that crap so you can do what you need to do as a human,  which is the bit where humans are good and here's the value I add. But I'm not gonna sit here all day, you know, reading all the crap in my email because I'll have someone else go do that for me.

Right. That's really valuable and we should embrace that. Especially in high stress industries like ours. So there's like, one good thing about AI is they don't complain. Which started me thinking the other problem with AI right now, which is that the great thing about the chat interfaces you're talking about, which, and I completely agree with you, that you know, the, the chat GPT, the revelation on it.

Was the UI, which was you just type shit and it replies to you like that was a bit of the blue. It was like that sort of, you know, it was like the invention of the mouse all over again, right? It was like, Oh, pointers and I can point and click and stuff instead of typing in dos prompts like that was like a huge thing.

So  but we're still at this point where AI  is. largely reactive,  right? It's the same thing with Siri, right?  Siri will wake up when I tell it, Hey, Siri,  but they all become much more all the people that are watching this, all their phones just lit up at the same time. I'd love to do that. Hey, Siri. So that is we, we got to get to the next step, which is Not just having to type it, it can read through the score.

Machine learning. I mean, that's what the letters, you know, written, or, you know, the paper is written in 2018. We're, we're saying we're, we're starting to see results where the computer is, is thinking. Not just saying, wake up at 7am. You know, Siri and Alexa, they've been around for 10, 12 years.  And that's, that is AI, but it wasn't machine learning, it wasn't machine learning.

They shouldn't have to you shouldn't have to come back and tell her, you know you know, will I make my flight next Wednesday? Right. A decent AI should be able to look at your calendar and the meetings you're booking and say, wait a second, you know, next Wednesday, you're not going to make a fight because you've got this meeting and you're going to be at Midtown Manhattan and you're never going to make it out of the airport instead, right?

A proactive AI, which is where we really need to get to. It will be the one that actually says, I've looked at all your data. And here's my advice, right? You need to do this this way. You need, it's your start of your day on a 9 a. m. Monday morning. I shouldn't have to type in, you know, what is the most urgent thing for me to do right now?

It should pop up and say. Hey, I've looked at everything you've got going on. You've got these emails came in overnight. What you need to start working on is this, and then this, and then that. In fact, I've already started a couple of them for you. Like that rea proactive, not reactive AI is, you know, that's when it starts becoming really useful.

Because sometimes I don't know what I need you to do.  Assistant that's, that's the conversational AI, right? Because if you're having that conversation and it's listening and responding in a way that is a conversation and it's learning from all that, not only in the chain prompting, but in your historical data, as you interact with these platforms as your personal assistant, like you speak of, then it is gaining that knowledge and will have that I wouldn't say sentient,  um, but it, it will be thinking for you, not just reacting and just say, yeah, wake up.

It's 7. AM.  . I go back to my GE alarm clock from 1972 for that. That'll, yeah, we needed to go back and go, actually you set the alarm at 7:00 AM you need a 6 55. 'cause just in the amount of stuff you need to get done tomorrow, you know, that. So that is a, you know, platform level thing. That's where AI starts getting useful, especially when it can digest huge amounts of data.

You know, and you look at something as complex as making a movie, there's a stupid, a stupid amount of data and conflicting data and overlapping data and it's high unwritten and it's different sources and stuff. That's where AI is really good at looking at all of it and saying, I can see a trend here.

You know, you're going to go over time and over budget and I can see it coming. So I'm going to give you a couple of heads up right now. You didn't ask for it. But I'm going to give you a couple of, this is going to go off the rails. You need to look at this part of the project and this part of the project.

And again, that's not just Hollywood. That's just in general, that's where we need to get to. And  I go back to what I was saying five minutes ago about trust and understanding the data models and stuff, because you only get that level of proactivity if you can give it enough data and trust that. The company that's running that service is not going to abuse that data.

They're not going to take your data and all your most important secrets, and let it get leaked online to nefarious actors, or sell you advertising against it, or do dot dot dot all these other nefarious things.  Because, you know, you can't get to that level of productivity unless you can get to that level of trust in that is my data and I'll allow you to go use this because I'm hoping you're going to give me value back in doing it.

 That's a whole nother problem. Yeah, no, that's. The lofty goal and let's hope we get there, but  you know, it takes bright minds and brilliant people to push the format and whether that's developers or solopreneurs or people in their various industries who see the opportunity and are actively trying to figure out how to make this stuff work and create a better, a better world for all of us and folks.

And you're, you're one of those dudes, man. So I appreciate the conversation. It's going to happen. Might as well embrace it and lead it rather than wait for it to run you over. Get on the train. Don't lay on the tracks. I have, I have been lecturing in various art schools and film schools. And I have to say. 

They're all embracing it at that age. So that's the next workforce. And so, you know, there's hope there's no risk. They're in complete fantasy land. They're in a school doing school projects. There's no, there's no movie executives telling them, no, you got to have Arnold Schwarzenegger. It's like, no, I'm just doing, I'm just learning here and using all these new tools.

How cool. So those people that, that generation is going to do some amazing. They'll have the tools, they don't have the grey hair and the experience of the real world that we've got, so we'll be fine. You need, you need, you need everyone. There's also data around, you know, employers picking their, their leaders and they know that gen Z and the alpha are going to be transitory where you have this mature workforce with generation X and the boomers who actually, if they focus on upscaling them, that becomes your AI brain trust.

Because the younger generations are going to leave your company, you know, after two or three years. So don't invest in them invest in your current staff. Don't don't lay off all those people. Yeah So, you know, it's a it's it's a very interesting time right now You know a year and a half into you know, gen ai and all of its impacts.

So I really appreciate the conversation mark you're always  And I just want to say  thank you, Mark. And please come back and talk to us again. And thanks to all of you for tuning in and catch more of our Rome IQ sessions on your favorite podcast platforms. And please follow us and smash that subscribe button.

Also, you can now follow us on X. Instagram and LinkedIn and catch Mark Turner and me and certain sneak peeks on all these social channels. And we have some interesting conversations. Mark, thanks again. You're a terrific guy and a guest and a great peer. I really appreciate the conversation and your thoughts. 

Nice to see you again.  Chat again soon.  Okay.