#276 - Bob Bonniol
Bob Bonniol is an Emmy Award winning Director, Production Designer, andProducer. He is known for his implementation of extensive media andinteractive features in his productions and installations. His practice spans the worlds of theater, broadcast, opera, installations, and architecture.
In 2023, he has widely deployed new interactive technology and AI to elevate live concert production, with tour designs for Billy Joel, Blake Shelton, Megan Thee Stallion, and the One Night Houston Festival.
In the course of his career he has worked from Broadway to Beijing, helpingclients like Disney, NBCU, ABC, Live Nation, AEG, Marvel Studios, and Lucas' film gathers and dazzles audiences of millions.
We had the pleasure of Bob being on the show in February of 2023, you can check out episode #183 on our website www.geezersofgear.com
Please welcome Bob back to the Podcast.
This Episode is brought to you by Elation and Main Light
0:00 What can you say it's Bob Bonnie all really cool talk about AI Finally someone who really is into it even more than I am and and super smart guy. Obviously Bob has written his own code. He's a very smart computer guy. So fun talk. We got into it pretty heavy here on on this episode, which is episode #276. So I really hope you enjoy it. It was fun and I look forward to doing it again with Bob because we only got we only got so far into it. So enjoy. Thank you for joining me today for Gears of Gear episode number 276. Today's podcast is brought to you by Main Light, a national dry hire rental provider specializing in supplying the latest lighting and stage equipment technology. Their extensive rental inventory includes moving lights, trust solutions, control consoles, LE DS and an array of weather resistant IP65 rated fixtures perfect for outdoor venues. Whether for theater productions, outdoor stadium events or TV film broadcast, Main Light is the partner you trust with the gear you want.
1:14 Committed to quality and reliability, Main Light also provides a robust selection of used sales options to their client with each item meticulously maintained and owned by Main Light. Coming with a 60 day depot warranty. Strategically located across the US, their four key locations include Teterboro, New Jersey, Wilmington, DE, Nashville, TN, and Las Vegas, NV Discover how Main Light can equip your next event by visiting Main light.com today. And today's podcast is also brought to you by Elation Professional. Please take a moment and watch. This wonderful video.
2:11 Inspire your audience's imagination with Elation's Pulse Strobe series. Designed to bolster creativity and deliver breathtaking kinetic intensity, the Pulse series empowers designers to create dynamic lighting experiences that captivate any audience. The Pulse series synthesizes mesmerizing lighting experiences with power and precision, allowing designers to create rhythmic, heart pounding lighting effects with effortless control elations. Pulse Series. You've probably already seen it and therefore I'm going to drive you crazy by telling you again that I went to see this week David Gilmour from Pink Floyd fame, of course, and designed by Mark Brickman, of course, who
3:41 we've had on the podcast recently. I believe it was episode 271 if I remember correctly. And so Bob or I'm sorry, Mark invited me to see the show. And sorry, I was reading about today's guest, Bob Barniel. Mark invited me to see the show along with John Wiseman from PRG of course, and hung out with them, spent some great time with them. Also Jerry Harris from, you know, the founder of PRG, of course he invited or he, he was there as well. And I got to spend some time with him, some quality time with Jerry, which is always fun. Sorry, I'm just trying to get this light out of my eyes. There's a LED friggin video light thing in my eyes and it's driving me crazy. So anyways, got to spend some time with Jerry, got to see the show, love the show.
4:34 I'm not going to talk too much about it because I've posted about it already a whole bunch. And if you want to see some pictures or videos, you can go to I think either my Instagram or the geez of gear Instagram. I can't remember which it is, but there's some reels and you know, it's just a fabulous show. It's it's so well designed, it's so unique and different. And I just really enjoyed myself and enjoyed spending time with some some great friends as well. And it's always fun to pop into New York for 24 hours as well. So I would say if you get a chance to see it, go see it. But unless you're in New York, you're you're not going to see it 'cause the tour is very short. It was only four cities. It's in the fourth city now, New York.
5:16 And I believe there are two more shows, if I remember correctly, maybe tonight and tomorrow night or tonight and Sunday, something like that. Maybe even just one show this weekend. But yeah, what is today? Today is the 8th. So I think there's a show tonight and a show Sunday and that's it. And then it's over. And so I hate to say it and I don't want to be the one to say it, but this could be the last chance you get to see David Gilmour or Pink Floyd or David Gilmour doing Pink Floyd or any of that. And also, it's really worth seeing this design. So if you have, if you're in New York, if you have any ability to get tickets, if, if you actually by the time you hear this, it's over. So I apologize, forget all that stuff. So anyways, it was great.
6:01 Watch it on YouTube. I guess So what else do I want to talk about? I'm not going to do a big intro today because we are actually going to talk about AI today. And I've been looking forward to this one because there are a few people in the industry I think who are as geeked out as I am on AI and today's guest is one of them. Bob Bonnie. All of course, is an Emmy award-winning director, production designer and producer. He's known for his implementation of extensive media and interactive features in his productions and installations. His practice spans the world of theatre, broadcast, opera installations and architecture. In 2023, he widely deployed new interactive technology and AI to elevate live concert production with designs for Billy Joel, Blake Shelton, Megan Thee Stallion, and the One Night Houston Festival.
6:59 In the course of his career, he's worked from Broadway to Beijing, helping clients from Disney, NBCUABC, Live Nation, AEG, Marvel Studios, and Lucasfilms. Gather, gathers and dazzles. Dazzles audiences of millions. Speaking today is challenging for me. We had the pleasure of having Bob on the podcast in February of 2023. You can check that episode out. It's #183 and that is available on Spotify, on YouTube, on lots of places, or on our website, geezerofgear.com. So please join me and welcome Bob Bonnie all again here to episode #276. Bob Marcel, how are you?
7:52 What's? Happening. I am lovely, man I'm I'm good. It's a beautiful fall day here in Seattle. Yeah. And I've been living in Los Angeles for the last two plus months. So I just got back home after my residency in Los Angeles and. What was that? Can you talk about it? Yeah, there, there. You know, there's a variety of projects that I have going on in LA, but primarily I was there because my daughter was doing two months of training and ACT lighting on becoming an MA3 support. Oh, very cool. So she's a programmer. She's a budding programmer, let's put it that way. Yeah.
8:37 And had had become quite proficient on the ETC consoles through the course of her education. Yeah. Just not surprising. ETC kind of has that marketplace. They do, yeah. Sewn up, but she was quite eager to learn about MA. She's been on site with me a few times and she's always been particularly blown away with, with Felix Peralta and his proficiency with MA. Yeah. So I, you know, I reached out to Ben Saltzman at ACT Lighting and said, hey, do you guys have any opportunities for training or for that?
9:18 And he's like, it's funny you mention it. We are actively looking for people to put through this program, you know, to come, you know, sort of serve an internship at ACTS. And of course that my daughter Izzy, she got to learn about Zac tracks. She got to go through all of the training on MA 3, but then go yet deeper and learn how to debug the new software and learn oh wow. You know, so she learned it at a level that's not just a programming level, but at a deep functionality level. So we've just, I was primarily there to make sure that my, my daughter was successful. You know, let, let's put it this way, any of any of our listeners who are parents know that, you know, sometimes your kids have a, a, a variably successful relationship with like living in the world, you know, yes. How old is she?
10:16 She's 20. Oh, OK, yeah, my son's 20, so I trust me. I get it, you know, just keeping him from killing himself by doing something dumb. Is is, you know, most of my goal in life. I, I worried less about that with Izzy than than than, you know, just her, you know, a, you know, call me conservative about this, but setting my daughter free in Los Angeles, where she's ever been before. Yeah. Maybe a little harrowing, yeah, but but I also wanted her to be free to really focus on what she was up to. And of course, I have infinite reasons to be in Los Angeles. Yeah. You know, both projects going on, but also developing developing new stuff.
10:57 Too, I did something similar and I, I, because again, I guess one of the differences between a young man and a young woman is you're a little less nervous about sending away a young man. And my son went to London by himself for six months and, and I didn't worry about him at all. I just, I got, I dropped him off with his bags and stuff at an Airbnb and said, see you later, you know, and that was that, right? And then went over to visit when he was racing or whatever. So yeah, I mean, I get it, I get it. But it's really cool that you did that with her and, and helped to make sure she was successful. And obviously, it's nice when you have connections in the industry that you can sort of help some of those things along or or at least, yeah, so good for you.
11:48 Good fathering. Just doing my best. You know, back when I was a production designer for Nickelback, I remember talking to to Chad's mom, the lead singer Chad. I was talking to his mom and we were talking about being parents and, and, and you know, I have a girl, obviously she had a boy and I and she said, well, I'll tell you Bob, you know, when you're raising a boy, you only have to worry about one dude. When you're raising a girl, you have to worry about every dude in town. Yeah, I heard, I heard, I heard a slightly more colorful version of that same story. There is that there I can tell you that that is not a quote it. Was. Yeah, it was. A slightly different verb going. On. But I think that gets the point across. Exactly.
12:31 Exactly. No. And I get it, trust me. But I mean, you know, again, with boys, I think really like, you know, again, my son likes to go fast in everything that he does, right? So like. Course he does if. He's pushing a shopping cart through the store. He's riding it like a skateboard or a scooter or something as fast as he possibly can and probably racing his body or something. And you know, just bad things tend to happen. You know, it's like funny story, when I was designing my, the house that I used to live in pre divorce, when I was designing the pool on the back of it, I had a guy with CAD and you know, he's designing all here's the waterfall and this is going to be the hot tub over here.
13:14 And here's where we're going to, you know, whatever. And I said that distance right there between the house and the edge of the pool. How far is that? And he said it's 8 feet. That's pretty standard. And I said, we've got the room. Can you move that double like 12 to 16 feet? And he goes, I mean, we can, but won't it look a little strange that far from the house? And then you got to build a bigger screen if you build a screen over it. And I said, yeah, but that upstairs window, that's his bedroom. And I guarantee you we're. Measuring the trajectory of the. Yeah, yeah. If it's too far, he's not even going to try it, right? But 8 feet, he's going to be jumping, you know, he's going to jump from there to the pool.
13:54 It's going to happen. So I'd rather not inspire that to happen right the minute I leave. But anyway, I appreciate you coming back on again. It's been a year and a half and lots of things have happened in the last year and a half. God forbid. I mean, our industry went that shit crazy and now it's crazy in a different way. And you know, I was talking to an account exec from one of the big companies the other day and he said, you know, I went from the crazy busiest year in my career and, and biggest, most successful year in my career to like this weird year where it was incredibly busy, but they were all smaller deals. And, and so I was still just as busy, but I didn't make any money.
14:36 And you know, it's just like weird, it's weird. Our industry's weird, but you're staying busy. It's a, it's an up and downs thing. I'm lucky in that I've got professional ADD, you know, and plenty of our colleagues are very focused folks like they're, you know, they, they either focus on their their concert career or they focus on doing more brand event type stuff from maybe a focus in architecture. But because I am like stay focused on much for very long. The net benefit of that is that you know what mode we've we built a a business it's about playing in all three sandboxes. Yeah.
15:17 So as businesses seem to have their cyclical ups and downs, usually there's an offset so. There's overlaps. Having a good concert year, we can focus on the corporate work. If the corporates are pulling back on spending and there's less work there, we can look at architecture and music and you know, between the three verticals, we we seem to find enough to buy the groceries. Was that on purpose, or was that just entirely by accident? You know, if you look back at the history of the company and the like across my career and Colleen, my my wife's career, you might infer that we were very smart about how we did that. But. But in fact, you know, I'm, I'm here to say it was a little bit of a happy accident, you know?
16:03 And to that end, I tell people just starting out, say yes to everything. Yeah, try everything. Go work in fashion. Yeah. Do fashion shows. Go work in business. Go do corporate events, go do theater, right? Go go see what architecture stuff is like. Because I, I think, you know, so many of the, the best practitioners at what we do have had that diversity of experience. And I think in the beginning of your career you are well served to not worry about as much about what pays more or maybe what you think you're interested in and just to get that diversity of experience.
16:43 Now, I think as a career proceeds, what you really ought to be doing is saying no a lot more because. Then of course you. Determine what your focus should be and and you know where where you are best able to contribute to things. Well, and so many people don't understand that it's generally harder to say no, especially in business. Like what business people have the hardest time learning to do is to say no what to say no to, not what to say yes to. But it's funny, you know, when you talk about the, the multiple different disciplines and, and that kind of thing. So I went and saw David Gilmour in New York this week. And isn't, isn't. That show, I, I get to see it at the Hollywood Bowl. Yeah. Oh, cool.
17:24 So, but I was going to say that, you know, Mark had taken basically a couple of years, I think 8 or 10 years off of lighting and became an A very good artist. Like he's he's become an artist. Yeah. And and so seeing that show, you could see very much art blending with a, you know, classic rock show and so well, like, it was such a beautiful show. It was just so unique and and so cool. And I mean, it had some aspects of some old Pink Floyd shows with the big lasers and stuff. So much fog, more fog than I think I could ever remember anyone using but just an incredible show.
18:11 I loved it. I this morning again I went and looked back at the pictures from the show 2 days ago and and I was just like wow you know what a beautiful show. I think that Mark is in a really fortunate position there. I mean, it's, it's, it's evident and, and, and it's true. I think that the David Gilmour regards Mark as a, as a, a trusted and and equal collaborator, yeah. In artistic freedom, so. Few examples of that. I mean, OK, you know, Steve Cohen with Billy Joel, Billy Williams with you too. You know, there there's some some very rare examples where the artist has has not just given permission, but has fully embraced like a A and what is usually a long collaboration with with a person in particular. So you know, Mark's got that with David and and Mark having had his art sabbatical, as you say, you know, he was he came to the store.
19:13 And what I love about Mark is he's so performative, right, Like like he's he in as much as he is. You know, he's like jamming with the band, right? There's not a lick of time code going on there. Well, and I don't know if you heard his podcast, but he talked about that. He said if I'm going to do this and if this is going to be my last David Gilmour show because I don't know, because you know, David's up there. I think he's turning 80 this year or something. And he said, so I don't know if this is the last one. If it is, I want to, I want to control the lights. I want to actually operate the lights and I want to make, you know, create art with the lights.
19:59 Yeah. And, and he did. I mean, he really did. It's, it's such a very, you know, it's not like it's outrageous technology or it's just an outrageous lighting package or anything. It's just really cool use of light. It is, is. And so I like it. I think it's really cool. But you know, it's a. It's just interesting that you talk about. The different disciplines and saying yes, because I think when you like, if you grow up a rock'n'roll designer and you stay a rock'n'roll designer your entire life, you're kind of boxed in a little bit, right? But when you start pulling in theater, or you start pulling in art, or you start pulling in different things is when it gets interesting, I think.
20:42 Well, and I think that that was a that was a differentiated for me in the beginning of my career. Yeah, I grew up in Rhode Island. Yeah, I was coming up in the scene in Providence and I was, I was mainly, you know, I was kind of, I was very wishy washy about what I was studying either in the high school or college. I was kind of half of my time with Tech. Like I was, I was working to learn computer programming, which has since informed so much of everything about what I've done, which I think we're going to get to. But then simultaneously, I got a job at Trinity Repertory Company in Providence, which is very well regarded regional theater as a spotlight operator.
21:24 And then and then that's where I learned to become an electrician. That's where I met my wife. Who? It took me 12 years. To go on a date with me. But you know, when I met Colleen at Trinity, she had just come off the road with the original Joshua Tree tour. She had been on the lighting crew for you too. She had been out with Springsteen and had worked with Jeff Rabbits for a while, and I thought she was amazing, you know, but I would say, but that was a theater and we were doing incredible theater. We were. And it was a very theatrical. I also met Anne Militello there who came into design shows. I met a whole bunch of other great L DS and I got steeped in this theatrical tradition of what lighting look like.
22:05 Now at the same time I'm also working, I found a job at a club in Providence called The Living Room, which was the one of the big concert clubs in Providence where basically for the compensation of as much beer as I could drink and maybe if I were lucky I'd get a 20 or you know, you know, 20 bucks a night, 40 bucks a night. I was running, you know, an old Leprechaun LP 1002 scene preset board and 24 park cans. And I was doing bands and I was learning how to busk, you know, when I was, I was learning how to be performative with lighting, with music. So I was doing music from the beginning. I was also doing theater from the beginning. And I had then I also stumbled into the local corporate market. All of this is an electrician, you know, but also then very quickly I was bringing the tech stuff that, you know, like I was, I was, I'd worked out all kinds of, you know, I was fascinated by graphics and computers. So one of the first thing that I did at the living room, for instance, is I went out and I like, I found an old ruined GE light valve, which is an improbably gigantic, incredibly stupid projector to work with, but it, it had been abandoned and unusable by an AV company in Massachusetts. I went and got this thing and I would bring it to the living room and I was doing, I was, I was literally using like a video toaster or an Amiga computer to
23:38 like, do I I'm not going to insult Joshua White by saying I was doing liquid light shows. Yeah, but I was doing like, I was doing like psychedelic projected texture on these bands at, you know, by like fiddling around with the computer at the same time I was doing the the lighting so so you could see all the pieces that now I've become known for. Yeah, at that point. But I'm, I can't tell you that there was a conscious decision. Yeah. Oh, I know what I want to be and what I want to do. I want to combine all these things and I, I was just a like a 21 year old kid, like saying, you know, man, my, my beer money jobs sure are cool. Yeah, well, I mean, it's, you know, those lucky accidents and and things that happen like that, you just go, hey, this actually turned out really well.
24:29 You know, like a lot of those things like verilite, you know, I remember the the story that I heard on my podcast where, you know, the the light wasn't supposed to be on while it moved. That was just and. Then they to. Reposition the fixture. The shutter was supposed to be closed in the middle and and there was a glitch where it stayed. Open moved and everyone lost their mind, right? Yeah, and went well, do that again, you know, and suddenly, you know, the look that became so famous happened. So I like that kind of stuff when that. Happens. I adore that anecdote. Yeah, sometimes mistakes are are, you know, the best part of the show. But you know, the cool thing, one of the cool things, not to go on and on about this, this Brickman show, but one of the cool things about it is I've seen the same song multiple times on video and stuff from this show, and every time it looks completely different because it is because he's performing it, you know?
25:24 Among the things he's doing, he's got, he's got that Corel paint thing, which is feeding the VPU on the MA, which is letting him literally be rhythmic. Yeah, and and, and and motion based and emotional based in the way he's running the lighting and pieces of the show. And I just think that that's it's just phenomenal. You know, I've heard busting. On. Acid talking about that show and how different it is from anything else. And that's Mark. Mark thinks without constraints. He doesn't worry if something doesn't exist or if a workflow hasn't been determined. He's going to, he's going to, he's going to get it figured out and he's going to be, he's going to put it to work and he's going to put his, his fingerprints on it.
26:09 Yeah. You know, you see that show and you can't. Well, the, the very first time I ever met Mark, it's a story I've never shared, but we, I was with Martin and we were creating our largest, most expensive booth and trade show at LDI ever, which was in Miami Beach. And we hired Mark Brickman to design the show on our booth. And so Brickman like took him forever to come up with the design. And I'm not mad at him for this. This is just how he operates. We put walls around him and you can't do that with Mark. And so he came to us with a design that had like a half $1,000,000 in video on it.
26:55 And our whole budget, including the party was like a half $1,000,000, right? And so we were like, Mark, this ain't going to work. And he went, wow, I can't do it then. You people are too small thinking and, and so, you know, in a panic, you know, we, we, I called Peter Morse and said, Peter, you were really our first choice. Can you come do this with us? You know, and, and Peter did an incredible job and I don't know if you were there in 1994 or whatever with the. Bicycles. It was there. It was one of my, it wasn't the first. It was, I think it was my second LDI and everything. OK. Yeah, yeah. I mean, it, it was, it was like legendary the the show that Peter did and how hard it was for us to pull it off with pals that were like overheating and just things failing left and right because it was a hot box of a room there, you know, with no ventilation and stuff. And but, but it worked out really, really cool. But that was my first experience with Mark. And what I learned was Mark needs to be with bands the size of Pink Floyd.
28:01 Like you can't. But then the next thing I saw him do was matchbox 20 with those incredible with the motors and the and the rig moving so quickly and stuff. It was just incredible. I was like, wow, you know, and he was the first guy that I saw do that. So it was just so cool. But anyways, moving on. So one of the reasons I wanted you to do this, and I'm glad you finally agreed to it, is because when I think of who the two probably geekiest AI talking guys are in the industry, it's me and you. And I'm not much of A geeky guy 'cause I'm not that smart. But I do love AI, and I do love talking about it. And I've certainly tried a lot of AI stuff and have experimented an awful lot.
28:44 And I actually want to share something with you real quick before I forget. Hang on one SEC. Let me figure out how to do this. This is only going to take me one second. Here we go to share audio Share a tab instead. Well, how do I do that? It's not going to let me share audio you. Suck. Hang on a second, let me figure this out. Oh, Oh yeah, I don't think it's going to let me share audio. This ain't fun. Well, I'm gonna do it without sharing audio. That's what I'm gonna do.
29:26 All right, well, I'll just, I'll use my mind's ear, as it were. Well, it's your basic. I've seen some wild stuff. It's no good without the audio. It's. Like, you know, a magician I can kind of have. The audio A new AI tech too. He's impressing that in 823 using all the crazy. Interactive tech and AI to make live concerts. Absolutely mind blowing. We're talking Billy Joel, Blake Shelton, Megan Lee Stallion. Bob's been behind the scenes making their shows pop and get this, he even rocked the one night Houston festival. It's insane how he takes all these high tech gadgets and turns concerts into these immersive experiences.
30:19 Dude kind of looks like check out what Bob. 'S been up to and learn a ton about AI as well on Geezers of Gear episode number 276. So that that was created on a on an AI video app called NVIDIA in about 5 minutes. And all I did was I took your bio and copied and pasted it into the the script for the video. Yeah, and it so do you use NVIDIA at all? No, no, I don't use NVIDIA at all. I use a little bit of a different tool set for the video that I do, but. Yeah, well, NVIDIA, if you're looking to do a quick little marketing video or something like that, it's just really super easy to use.
31:00 And again, you can just paste a document into it and it'll look through the document and create a video based on the. But what it used to do is it used to just use stock video and stock images and stuff. So it was OK, you could do a kind of a cool video, but it was just like guys shaking hands and like just boring stuff. Now you can say OK, like that one I said I wanted animated. And, you know, it's about this Geezers of Gear episode number 276. And then I put your bio in there and it came up with that in like, you know, 3 or 4 minutes probably. So if you can do generative video now and you can do it animated, you can do it lifelike, you can do it in all kinds of different ways. It can be fun, it can be serious, it can be, you know, so it's actually a tool that I use quite often just for creating little market blurbs and things like that. But yeah.
32:02 I like it, it was good. Yeah, yeah. I mean, it was just something funny that I thought I'd, I thought I'd show you just using one of the apps. So, you know, I want to talk about some of the things I use on a daily basis, which will be totally different than some of the things you use, of course, because you're using them for to generate probably content and to generate. Yeah. So give me give me some examples of of ways that you've managed to incorporate AAI into your workflow. Well, there's a few different ways and there's a few different. Of course, when we when we say AI, that's a that's a broad topic, right? Yeah.
32:43 I mean, you know, for the record, I started, I, I attended a computer camp when I was a teenager at MIT. Yeah. And I remember that when we went there, you know, it was, it was very cool. But they said, OK, what we're going to do is by the end of the Cam, you're all going to have written a computer program. All right? And out of the, you know, out of the 100 kids there, 99 were like, I want to write a computer game. Oh, OK. Which is predictable and cool right now from the time this is like 1980 3 maybe? Something like that.
33:27 I was in high. School at the time I did this, I did this and but what I wanted to do was I wanted to write a program that would teach the computer to talk to me, you know, and not necessarily verbalize, but I wanted to be able to type, hi, how are you? And have the computer go, Oh, I'm I'm having a good day or whatever. Like, I wanted to be able to have a free form conversation with a computer, which the instructors were both, you know, they were like, oh, wow, that's cool. And they were also like, you're really screwing up our Lesson plan here because you're supposed to want to make a computer game, right? Yeah, but. You know, but but what I did was at that time in the course at camp, which was a month long thing that I did and you know, I think it was a freshman in high school or something was I wrote, you know, a piece of code that was effectively a chatbot.
34:22 OK. Yeah, it was a very limited chat bot because at the time all you could do was maybe guess at what people would say to this thing. And then you could come up with a range of things it could say in response. So you could only have a very rudimentary discussion with the singing. And it only had so many options as as to how it could reply. But so now we look at modern day large language models. So ChatGPT anthropics clawed Google's Gemini. These are the the, the, the, you know, the offspring of that chat bot thinking that goes all the way back to that. So I've been noodling with this early on. And, and even if we go back to 2010 when mode, when we put our interactive screens in Microsoft headquarters in Studios A through D, we were using machine learning to allow the screens to be different every day.
35:22 And to to how they responded to. These were screens that you could gesturally play with if you were in the right spots in the lobby and you could audit, you could play with them by through your voice. And we came up with, you know, machine learning things that allowed the code to put together different graphic looks on the spot. And this is 2010. Wow. Yeah, that was very. Early and this and it's been a hallmark of a lot of the work we've done, especially on the permanent installation side of creating what we dream to be semi autonomous art pieces, you know, art pieces that would that would think for themselves or do what do do do things because of environmental. That's interesting.
36:07 Yeah. What the? Weather was what the traffic's like around the building. You know how many people are moving through a lobby and then suddenly this LED sculpture was behaving differently, showing different colors, doing different patterns. We repeated that in the in an installation at a Salesforce building up here. We repeated it in a bunch of different installations. And then in 2018, when we designed GM World, which is a visitor center in the lobby of General Motors Global headquarters, where we had 17 moving LED screens, we came up with a thing called the Interactive Content Engine, which would basically give an input from the GM people, but also environmental input and computer vision input. It would put together new pieces of content every day for those screens.
36:58 And that came from the chief marketing officer for GM, Tim Mahoney saying, Bobby, how can we make sure that people's experience of this area is different this Tuesday at 9:00 AM from next Tuesday at 9 AMI don't have a budget to pay an agency to be constantly making new content. Yeah. And I was like, well, I think we have an answer. And that was because we have been doing all of this work. So we built the interactive content engine which which which does this, it comes up with whole basically GM is making new content everyday, like terabytes of it. So we were having all of that go into a bucket and we had one machine learning piece of code that would look at every piece of content that that GM contributed. So an ad for, you know, an ad for Chevy Silverado, training video, marketing videos, etcetera. It would go through all of these videos and it would look at the videos and it would tag those videos. So we put meta tags on it to say this piece of content features a Chevy Silverado.
38:09 And I can see that it has this trim package and I can see that the the truck is on the beach and there appears to be a family around it of a mother and a dad and some kids. And so it would then Add all this metadata to those clips. So all of the media going into the bucket now had a way to, you had a way to identify, you know, through metadata, what it was. Then we built a different machine learning program and this was the interactive content engine, which on a given day, let's say it's February, it's truck month, the Chinese manufacturing representatives from the factory that GMs in Shanghai are visiting and there's there's more traffic around the Renaissance Center than usual.
38:55 The interactive content engine would know all of these things. It would know the stuff about who was visiting because GM would tell it, but it would know what the weather was. It would know what the traffic was. It would know things like it's truck month and then it would go into the media buckets and say, oh, I've got a clip of a Chevy Silverado. Oh, it's on the beach. That's good because it's cold outside. So we want to put nice warm images on the screen. And I can also use it because the Silverados are drive trains are manufactured in, you know, and I'm making this up, by the way. This is I don't do, you know, not actual content, but they're writing engines in China. They probably aren't. But yeah, you know, it it would know, oh, these pieces of that truck are made in the Chinese plant. So the Chinese contingents going to be happy to see that. And then it would put together a whole content sequence based around that.
39:50 And it does that, and it and as the more it does it, the more it can build its model out and it gets better. So it was learning. So we were doing that in 2018, which is, which is, you know, relatively speaking, like centuries before we were all talking about what we're talking about now. So we're not, this isn't something that we've come lately to. It's kind of been something we've been doing from the beginning. Now would the would the new tools that have came come out with the chat, you know, on the large language model side, just to give you an example like we use. ChatGPT affords you the opportunity to create custom GPTS. They call it some other people call these agents right, which is basically you can train the version of the model to become specifically knowledgeable about something you would like it to be specifically knowledgeable about by feeding it a lot of information, which it then adds to what what open AI is already trained it on and gives it better context.
40:50 So we, for instance, at Mode, we've been working to train a custom GPT into being an experiential strategist, right? So it helps us with our corporate world like we can. You know, we feed it all the news about brand activations and what the big agencies are up to and what installations were South by Southwest this year and what you know what, what popular installations out there and the news around them and data around it and stuff. We've been working to build this custom GPT, which has now become quite an expert at, you know, so now I can come up to it and say, hey, I'm going to do, I'm going to do a big engagement for a tech company.
41:39 What would be a great interactive, immersive, temporary engagement to show clients or to show guests the power of using, you know, using a particular software suite to get this done in their lives? Yeah. And this custom GPT will then think about it, and it'll go into all the training we've given it. And it'll say, here's some great ideas you should set up, you know, the high tech TP on the ground and you should like, you know, put these, put these installations in it and these engagement points in it. And here's the strategic reasons why that is good. Here's stuff I know from metrics on other activations, corporations have done that have been successful or not and why this might be a good idea to do it. So in one way we're using it.
42:33 We're using artificial intelligence in our work as a creative agency to help us with coming up with new ideas and strategy stuff, in particular in that corporate activation market. So that's one way we're using AI one second. Go back up one second. So like, I know what a custom GPT is. I know how a custom GPGPT comes about. But can you explain like just a very simple custom GPT that somebody might build to do something and and what that would entail? Definitely, definitely. So you know, let's say that you run, let's say that you run a major lighting distributing company or distributorship, let's say, let's say you're you're Ben Saltzman and Act Lighting. Or let's say you're, you know, you're Colin Waters at TMB or something like that.
43:28 Yeah. And you've got a lot of people on your staff, right? You're managing a lot of employees. But you're from show business, right? You're not necessarily the best HR guy, right? Yeah. And but you've got employees who want to know, is it appropriate for me to tell my coworker at work, wow, those jeans make your butt look good today or, you know, whatever. Right. Like or whatever. Right. Yeah. You could phrase. I could totally see Ben saying that. Yeah, but I could not see that. Exactly. But I could see other people maybe. Of course, of course. I think I might have said that in my career 100 times or so. You've been around a while. I'm going.
44:08 To be controversial about it, let's say that you are running a company and you want, you want your employees to be able to find out like can I take a personal day or what are the rules about if I'm late to work, right? Or. And you don't want them to have to go find an employee manual that you gave them when they got hired that they probably stuck in a drawer and they don't know. So for instance, you could take your employee manual and you could take your HR guidelines and you could train a custom GPT on that and make that into a chat bot that exists on your company's Internet. So it's a great application. To know, hey what is my company's policy on PTO and they can engage with the chat bot and say how many, how many days off have I allowed in a given year and what for and that and now that Custom GPT is like your ultimate unbiased HR representative. Right.
45:08 Yep. Who your employees have 24 hour immediate access to. Yep, that's a good example of. It is, yeah. That's a great rather low tech example of why and how you would create a custom GPT. So thanks. Thanks for going through that one. So where were you going next that you were on to something when I. Stopped. So there's another. So that's one kind of AI which are these, which are these kind of very intelligent chat bots. The other kind of AI I think that a lot of people are seeing and talking about is the generative art AI. So things are creating the pictures like Dolly or Mid journey or programs like that, you know, and now also the ones that create video like Runways Gen. 3, the one, the one that you mentioned that you just used. There's a spectrum of these these apps as well. And what these are is these are these are, these are diffusion based transformer apps.
46:17 And I don't want to, I don't want to make people's eyes roll into the back of their heads, right. But basically what these apps do is they've been trained on, that is to say they have been exposed to a vast volume of, of, of visual information, pictures, videos, etcetera. And then what they do, and I, and I want to be careful because I want to make sure that people really understand this. They don't absorb all that and it sits in the database. And then, and I think this is a a misapprehension people have that like they then go into mid journey and they say, give me a picture, you know, of, you know, a lady walking down the street in the start, you know, you know, as if it were a painting made by Monet. Yeah.
47:08 Those programs then don't go look at that moment at Monet's art. And they don't then like sample that and put paste together a collage of existing stuff. Here's what happens. And it and it's, and it's much in the same way like people who go to art school, the programs are trained on the materials. So the materials end up resulting in a database that contains no pictorial imagery whatsoever. It is a, it is a literally a four dimensional hang on there. It is A4 dimensional mathematical space that basically has a bunch of data about composition, line weight, color, aesthetics, styles.
47:59 There's no pictures in it whatsoever, whatsoever. If you were to look at it, if you could look at it, because looking at a latent space database has problems of its own, it would look like a massive table of numbers. Yeah. So basically when you enter a prompt into a program like Mid Journey, which is a very popular image generator, it looks at what you've written, right? Yeah. I want a picture of a blue painted picture of a crazy purple haired, you know, entertainment designer, you know, sitting in a room full of clouds, right? It turns that verbiage into tokens, right? It translates what you've written into tokens. Now each of those tokens, it relates to a specific thing, a color, a descriptive term, male, female, you know, etcetera, etcetera.
49:04 And then then the program takes those tokens which basically give it numbers, data points, it goes off into its latent space database which has been generated from the training and it matches the tokens to areas of the numerical matrix. Yeah. And so it understands that bananas are yellow when it understands that clouds are puffy and white. And it understands that, you know, hair looks like this and etcetera, etcetera. And it goes into that space, and then the app begins with noise, right? It begins with literally what you might call static if you were looking at it.
49:51 And then it sharpens it and sharpens it and sharpens it until an image begins to appear based off the tokens and based off the data it's gotten from its latent space database. Now, the best ones, the best versions of these programs are actually 2 AIS. One AI is doing this work. The other AI is sitting there serving as a real time critic and going that's not a banana. Yeah, I don't know about that. Right. And and basically it's help, one AI is helping the other AI to tune its results. And that fuzzy static picture gets clearer and clearer and clearer and clearer until it matches the verbal request, the typed request you'd give it. Yeah.
50:37 So it is not sampling. It is not copying. It is not directly going out, scraping the Internet and putting a picture together. It has learned about. It's learned about, you know, a Monet painting. Monet paintings are soft, they're impressionistic. They're painterly. They usually have a more pastel palette, right? And so now the computer program knows what pastel means. It knows what warm means. It knows what soft and painterly means. And instead of sharp and graphic, right.
51:17 And so when you prompted to give you something like that, it doesn't go find a Monet painting and bend it and twist it into something you want. Yeah. It makes, that's such a great point. Completely custom picture. Yeah. That that references the aesthetic, the composition, the style. It's just like me going to art school with which I did I, you know, I attended RISD for a while, Rhode Island School of Design. When I went to RISD, I showed up and the first thing they did was teach us art history and we learned about the great masters and we looked at their work and we went into the museum and they would make us try to recreate their work, right?
52:01 They would. And at the same time, they're teaching us techniques. You know, they would teach us the techniques of these artists so that we could do things like what they did, having looked at what they've done before and been inspired by that, and then go out and, and then they taught us how to synthesize all these things together to come up with our own individual style. You know, the way these apps work is a lot like that. You know, they get exposed to things. They get taught why those things happened and how they work, and then they get taught on how to create things like it. Well, you know, it's such an amazing point because this is one of the greatest fears. You know, it is just the the fact that it's going to, you know, copy music, it's going to take people's likelihood and do things to it. It's going to do this.
52:50 It's going to all of this copy replica stuff is such a big concern, but it's no different than, you know, with music. There's only so many notes in music. And so yes, there's a good chance the song's going to have an A in it and it might have AG or AB minor or something, right. And but that doesn't mean you're copying another song that's got AB minor and an A in it. You know, it's and no different than than lighting design. You know, you're going to use a similar color of red. You're going to use a similar color of blue, perhaps doesn't mean you copied, you know, John Featherstone or, or Cosmo Wilson or Mark Rickman. You know it. It just means that you're doing a design. And when you do that design, you're taking all those experiences, all the other things you've seen and learned over the years.
53:35 And you know, you're, you're fitting them together in a way that matches this particular performance, right. And so I think AI does a lot of that as well. And, and you know, the way you just explained it is incredible. Like, I don't know that I've ever, I've seen 1000 YouTube videos and I don't think I've seen anybody explain it quite as well as you did just now, Especially the two AIS. You know, when you're saying 1 is basically creating it and one is critiquing it. So how do things like 3 arms or three eyeballs happen? Then what? Is that you know AI doesn't understand context, right? Like it doesn't, it has, it can tell you that it no, like it has a general sense of what a hand looks like.
54:18 You know, it's a big BLOB with some sticks coming off of it. Kind of like it knows from a compositional standpoint what a hand looks like. It doesn't understand because no one that you know that it's not the AIS aren't trained with context. And This is why there's always going to be a place for artists in this, because artists and creatives, I mean, part of what makes up me anyway creative is that I have the, you know, a ton of books behind me, right? If you could see the office around me. Yeah. All books. Yeah. Right. And a lot of what I bring to my creativity is context. Yeah. And the AIS don't understand context. So yes, it's true.
54:58 Like the classic thing of like, you know, you can tell an AI image because everybody's got 7 fingers. Yeah, they got the wrong number of fingers and toes. The AIS don't they understand that they're making, they're going to, they're going to sharpen a diffuse thing into a picture of something that they that it's been told is the shape of a human. But the AIS, you know, that these creative AIS haven't been trained in deeply in human anatomy. Yeah, you know, they don't know. So there's, there's like so much that they don't know about yet, you know, and I say yet quite on purpose, because, you know, with every time they refine these models and train these models, for instance, the whole phenomenon of too many fingers on hands has gets, has been reduced and reduced and reduced has come because because it became something that people noticed. And so the people training the models were like, oh, I can fix that now, you know, and and that's the thing, you can just like I talked about training the large language model to become an expert in experiential strategy, you can train a graphics model to be to get much better at a specific thing you want it to be much better at.
56:17 For instance, when we did Blake Shelton and Blake the back to the Honky Tonk tour, we wanted to build a big virtual piece of scenery in the video that was kind of supposed to be like Blake's barn on his property. That was a honky tonk and that one of the things in there would be all of these honky tonk posters of Blake, right? Like Blake plays, you know, the the the Riverfront Bar and Grill and Blake plays this and Blake plays that. Well, you know, and AI doesn't really know what Blake looks like. Like it probably has seen Blake a few times in the materials it's been given. But you're going to get hit or miss results out of it for sure, right?
57:04 So one thing we did was we used a technique called Laura LORA, which is a way to give a generative image AI the ability to gain extra knowledge about a specific way something looks or feels. We trained a version of Stable Diffusion, which is a different generative AI art thing, to be an expert at making pictures of Blake. Oh, interesting. Added. All of these pictures of Blake from his, from his younger days to his older days in all kinds of different, like we, we had access to all the photographs that had ever been, you know, that Blake's people had. And we trained this AI to be really good at making images of Blake.
57:50 It was already pretty good at making images of stuff that would look like honky tonk posters. So now we could prompt it make a series of Honky tonk concert show advertisement poster posters featuring Blake Shelton, and it made hundreds of them for us to choose. From Oh, that's awesome, so that. Was the way that we trained Stable Diffusion to get better at that we, you know. But don't a lot of these models don't a lot of these like things like Dolly and, and different image generation models or video generation models, aren't they really afraid of like the sort of anti plagiarism and, and you know, so aren't they avoiding that?
58:34 Like, aren't they purposely not allowing you to say, give me a picture of Taylor Swift with a mustache or whatever? 100% and I think that's that's probably from a. It's judicious of them to do that, of course. Yeah. You know, being accessible to a larger public, there are risks of misinformation, Yeah, you know, and. And the. You know, the kind of the bad things that can happen if. You fake? Yeah. Which is why we trained Stable Diffusion is an open source model. That is not, it's not, you know, I have that model here in the studio on our own computers. It's not accessible to the public.
59:15 OK. We train this model on Blake. It's not out in the world. You can't just you. Can't go to stable. You can't download Stable Diffusion now and it's, you know, and we've made it an expert on Blake. We've made our own version in our server closet. Very good at making Blade Blake pictures. That's cool, really sad out into the universe. So you could do Blake, You could do Blake Shelton songs, you know, deep fakes from your computers. Yeah. We can, and intentionally, like we've been granted the permission by by Blake to do that. Yeah. And now we've created something that is like with anything and is for us it's a highly, you know, that's a piece of proprietary technology we.
59:58 Have. And our agreements with with Blake Shelton and with Blake Shelton's management is that's not publicly accessible. Yeah. Of course. Yeah, of course. And in fact, they own they own the that. All the image rights and everything. Else that version of it. So yeah, if you ask, for instance, mid journey to give you a picture of someone famous, yeah, they've trained it to. Yeah, trust me, I've tried. Seen this true of Dolly, Same as true of many. They've set up guardrails to make sure that can't happen. Yeah, but if you have to know how to code, you have to know how to. It's not easy to grab one of these open source AIS like Flux, yeah, or Stable Diffusion and just use it like I have computer programming.
1:00:51 Yeah, yeah, yeah, yeah. Experience. So for me, you know, and even though I had to puzzle out a bunch of stuff and yeah, afternoon, I mean Python And you have to, you have to, it's not. Yeah, mid Journey and mid Journey and ChatGPT and all of these are end user types of programs that a knucklehead like me can go on to and, you know, figure out how to do some, some prompt engineering and get some results out of it, get some pretty good results out of it, which I have. But no, I mean, I've tried all of those things that that you're talking about, like getting images of, of specific people, specific places, specific whatever, and it doesn't do so well at most of those things. So music blows my mind, though the music ones blow my mind.
1:01:35 Like I've, I've written some really good songs on AI and just gone, whoa, this is nuts, you know? Yeah. And then now there are guardrails showing up on that because of course. Of course you know the. The, the record labels and the people that control the rights to music are, are, are, are legally well equipped as we know. And they've, they've you know, and so now you can't. It used to be that there were some of these opens, there was some of these music AIS you could go and you could say I want a song of like a like Kendrick Lamar singing, you know, song. By The Beatles, right? And it would do it and it would blow your mind. It would be like, I never thought of the Eye, you know, And it, and I think it's really cool, but the industry clamped down on it immediately. And the reason is because they don't want the public doing this.
1:02:25 And I think there is still a bunch of great tools out there that if you are musically inclined or a musician, you can still use these AI tools to create music. And it's super cool, but you can't do that that thing. But I'll tell you who can is the labels themselves. The reason, you know, the reason is that the the the record labels and the same goes for the major any of the major media companies on the planet. You know they're going to be and some of them already very much are. They are training their own versions of these A highs so they can have the ability. So they can sign an artist to four albums and do 12. Yeah, yeah.
1:03:14 I mean, they own the the, the likeness of the artist or whatever. And, and yeah, I mean, that's scary. That's definitely some scary stuff right there. It is scary, but, but I think there were a lot of things that were a lot of people were scared of a there are a lot. The big overwhelming, overarching scary thing that a lot of people thought about was like, oh, are we headed into Terminator times? Is there going to be an artificial general intelligence that basically decides that that, that humans are optional and probably should be eliminated? We're not close to that. I'm here to tell you, really smart sounding chat bots are nowhere near being an artificial general intelligence, right? It's a different, it's a whole different, I know from programming.
1:04:04 But did you hear Ray Kurzweil? I think what where was he? Was it on Joe Rogan? Or maybe it was, I can't remember which podcast it was on, but he says 2029. I don't think it's quite that close. I think, look, the kind of code that results in an AGI depends on a very different structure. Yeah, you know, and the chat bot is a basically is basically what it's really good at is guessing the next word that you want. Yeah, right. That's. When you get right down to the root of what it's doing, that's what it's. Yeah, right. You have to build a very different code base to be just good at everything, right?
1:04:51 It it's a real neural network, right? The, the large language models are not, they are neural networks, but they're a different kind of neural network. I think we're, I think we are going to see AGI. Yeah, I think we, you and I might see it in our lifetime. Like maybe. Somewhere, I think we definitely. Would probably. I don't think it's that far off. I mean, when you look at how fast things have happened over the last, what is it 18 months or something? We've gone from pretty much no one knowing not no one but the normal human walking the street knowing nothing about AI to now they've got it on their iPhone or or Google phone or whatever, right? And I think things are going to continue moving very fast. I think the low hanging fruit with some of the stuff we've been seeing, the translator models and the the web coding models and you know, personal assistant types of models and things like that, writing is obviously gotten pretty good already.
1:05:52 Oh, yeah, yeah. And I think that, you know, I think, you know, I don't think that we need to talk about politics or anything, but I can tell you that the incoming administration has a very sympathetic viewpoint in terms of deregulating AI and allowing the tech companies to go as fast as they want in the directions they want to go. Yeah. I mean that that. May be, which is fine. I don't have a Yeah, you know, I don't have a horse in that race. I. Think The funny thing is Bob Elon is very much the opposite of that. So I would think Elon in, in other people's ears is going to be because Elon is, is very, I wouldn't say afraid, but nervous of where things are going and how quickly they're going. Marcel Elon says that.
1:06:44 Yeah, that's true. But then he spins up Grok. Yeah, which is an open source large language model. Yeah, and freely distributed it to the world. The Chinese now have Grok. Yeah. And what they're going to do with it may or may not align with whatever sympathies Elon has. So one is saying one thing, but at the same time, he's building the world's biggest training data centers that exist, and he's freely distributing the result. Well, he also says that by the end of the year he's going to have GPT 44-O power. December I think is when he announces Grok 3. I believe it is and.
1:07:27 Like he has, he's definitely made huge strides. What I can tell you is a training a big large language model like GPT 4 is hugely expensive in terms of times and time and resources. But also, it's not like Elon Bolt believes in this very this, this, this work philosophy of the of intensity and that that, you know, they look, there are things in his companies that that I admire a great deal. Like I I am a huge fan of SpaceX. Yeah. And I think that because SpaceX is vertically integrated, and I think because of Elon's philosophy of how to tackle engineering, it has afforded them this huge advantage. Yeah, but it's one thing.
1:08:11 But I also know this about Elon. He says things like are gonna be something by something. Oh yeah. And they're that. Never happens. Yeah, that never happened, you know. I. Have well, it's like I have AI have a friend who has a deposit on the the whatever the the sports car is called. Yeah, exactly. The Roadster I think it's called or whatever, but I have a friend who's had a deposit on that for like 3 years and is like, are we ever going to see this party? You know, and look, we'll see about what he has by December. In terms of Grok, Yeah, Yeah. But I want to circle back to that fear factor. I mean, artificial intelligence is one thing. I don't think that that's what we need to be worrying about immediately right now. But but one another thing that people worry about is that that AI is going to put people out of work. Now look, AI is going to change the nature of the workforce. And there already has been huge impacts in journalism, in particular if you're a professional writer, like there used to be a decent living to be made, for instance, in aggregating news about a special topic. Like, let's say you were a tech journalist. Yeah.
1:09:23 And you write reviews about tech stuff like, you know, like phones or like gaming consoles and stuff like that. Forget it. That job's gone. That job is gone. Yeah, there are AIS who are just, they can much, they can look at everything on the web instantly. They can and they can summarize it and repackage it. So you know, so you know. Have you used notebook LM yet? Notebook LM Yeah, I've tried. Yeah, I've worked with Notebook, with the with Google. It's pretty freakish and cool. I love it. I love it. Here's. Another, you know, and the job of the paralegal is to go in the most part, or a big part of that job was go find existing information, find the important parts of it that relate to this particular case, summarize it, organize it, put it together, give it to the lawyer. That job's gone.
1:10:19 Yeah, it's gone. Well, you know, some low end lawyers, some lawyers who do like contract law or, you know, very basic, like, I mean, if you got to do a lease for an apartment that you're trying to rent out, are you going to go to a lawyer? You're going to go to ChatGPT? I mean, it's so easy to get those documents done now a. 100% like, and then this goes like for instance, we do a lot of work with big corporations, right? Like work with GM. To work with GM, you have to jump through a lot of hoops. You have to have, you have to have compliance documentation. You have to have all kinds of documentation for yourself as a company, your emergency response documentation. Because for instance, GM wants to know that if there is a hurricane or an, or where I am an earthquake, what, what are our, what are our emergency response procedures to make sure that we're still going to be able to do business the next day.
1:11:13 Interesting. This our GM account right? So for instance, so there's a lot of documentation you have to write as a company if you're doing business with people like that. Yeah. That kind of stuff, like, would you, we've been able to use ChatGPT to do emergency response plans, compliance plans, you know, computer security plans, you know, Yeah. Me too, I that simplifies so much of my day so. But there's also been a big, there's been a big within the art community, in the arts community, there's been this backlash of like this is going to put artists out of work. I don't agree. It's another tool.
1:11:53 It's another thing we can use. The effects industry did not put regular movie making people out of work. It just became part of the workflow. You know I want. To give the flip side of that, because like, you know, what about like Tyler Perry who cancelled, what was it an 800, 1000 square foot studio or something that was being built? And you know, because he just said we're not going to need studios anymore. We're just not going to need them. Tyler Perry personal opinion. I think that was calculated much more off of that was an excuse all of the big networks and film studios and most probably streamers stopped buying content. Yeah.
1:12:42 Right, because they're writing content down because they're trying to trying to drive share, share price up by reducing debt. Here's the fact of the matter. It is there's a huge, huge, there's a huge fall off in production of television and film right now. And none of it's got anything to do with AI. Yeah, it's got to do with media companies trying to figure out how to monetize their content in the age of streaming. I think consumption consumption was changed. Wasn't about to build a billion dollar film studio to sit empty. Not because AI had replaced it, but because there's no market for that shit.
1:13:24 Well, do. You know, do you know, would you know Glenn Rogeman from AED in Nether, Belgium? No. He'd kill me for saying Netherlands because he has a, he has a lighting sound, all kinds of companies, but he's got a, a studio as well. And I talked to him about this and I said, does that make you nervous at all? And he said, not really, you know, we're going to stay busy. But the problem is here's what he says. Like, you know, I know I paid a lot of attention to Sora. I haven't followed some of the newer ones that have come out, but Sora was doing some incredible stuff, like really incredible stuff. And but he said, here's what's going to happen.
1:14:04 Movies are going to get to here. They're going to get really good. And but he said then what's going to happen is all of this content is going to be right here. It's all going to be really good. And so really good is going to become vanilla, and everything will become sort of ubiquitous and boring. And he just thinks that, you know, the nuances and the good and bad combined together is what makes, you know, film discovery or movie discovery, show discovery so good. It's not like everything is an A. Well, what about some A pluses and some B minuses and whatever? You know, you need some of that stuff, right? But well.
1:14:48 I, I, I'm of, I'm of various opinions about that. I think that that I think that misunderstands and maybe skips over the idea that the AI should get better, right? Like they're going to feed context into these AIS. The AIS are going like the generative AIS are going to be taught that everything being perfect leads to homogenization, right? So they will introduce little human. Point. You know, just to appeal to us. So they also haven't yet even begun to train these AIS to produce stuff that is targeted to demographic psychographics and audience segmentation. Right. Good point.
1:15:28 Yeah. And they're going to get really good at it. And in the culture has become short attention span theater. You know, I know that my daughter, she gets, she does most of her media consumption not in this, not in a, in a, in a movie theater. Yeah, my son too. And. Only barely in any long form, anything she sees on TV, Yeah, she mainly looks at TikTok, Instagram stories and stuff like that. Yeah. And you know, or maybe like half hour episodic stuff, right? Yeah, Yeah. So I mean, look, I agree. I just want to say the AIS, yes, what your friend said, I think that there's some value in it, but but I think it skips by the idea that they're going to teach the AIS to to get better at not.
1:16:16 Being that's better, Yeah, that's true all the time. We didn't discuss that, yeah. And then the other part, this is kind of a three prong thing. The other part of it is I once had an executive producer told me, and this is a guy who's made billions of dollars off audiences. He, he said, Bobby, never underestimate the, the, the, you know, the, the appetites of the general public, which is to say they don't really care. You know, most of mainstream culture, they don't give a shit about quality. You know, they look at video in bad MPEG 4 on their phone all day in little bits and pieces and they they aren't necessarily captivated or captured in a conscious way by art.
1:17:06 So. Troy, I heard, I heard Justine Bateman who is the president of of Screen Actors Guild, she said. I think it was in a Variety article this week that there was going to be a mass cultural backlash against AI produced content 'cause people would get tired of it not being great or, or being just vanilla. Yeah, They're not. No, they're not. No, I think a lot of the audience is perfectly happy with vanilla. And I am not. I I am an artist. I want people to be inspired and to feel deeper feelings and to go deeper with things. Yeah, but I think a lot of the mainstream culture, they just want to hit a dopamine when they want to hit a dopamine.
1:17:53 Yeah. And they're and they they don't much they don't get attached. They they don't think to the extent of like, wait a minute, this all looks a little bit just too homogeneously. AII don't want to watch anything made by. Yeah, I saw one of those things today. I saw, I saw a video that had a notebook LM. You know how you can do the podcast thing with notebook LM And so it had the notebook. It's actually AI was going to show it to you, but boring. But it's it's a a video about the Jeff Bezos. What's it called PI0? Have you seen this thing yet? Jeff Bezos backed it. So Pi 0 is basically giving robots brains that you can tell a robot what you want it to do.
1:18:41 And the video shows like folding clothes and lot of house chores, right? Household chores, breaking and folding clothes and cleaning up and doing things like that. But I mean, Bezos round I think headed up around of like 1/2 a billion dollars in funding or something for this, this π zero thing. But the, the video is a video of this robot doing its thing. And the audio in the background is an LM notebook, LM podcast. And there's a few commenters who have like 1 follower who are like, how dare you insult me with an AI generated podcast. Really.
1:19:21 You know, are you serious right now? I think a lot of people would be perfectly happy to listen to the Notebook LM podcast. Their voices are great. I actually find that they they have a very interesting dynamic, those I love it, you know, P Here's what you know a lot of audiences are perfectly happy with good enough. Yeah. Right. And I and I and I am constantly seeking to help escalate audiences to a point where they can see something extraordinary and really be moved by it. That's where I am as an artist. But but but but as a professional designer and creative and strategist, who advises?
1:20:03 On these topics and who's I spend a lot of time thinking about audience segmentation and demographics. Yeah, what do audiences want, right. And I could tell you that a lot of audiences are perfectly happy with good enough. Now there are, there is an audience segment, there is a demographic that appreciates, handcrafted, they appreciate and they want and in this era of AI they are going to seek out real human created art just kind of like at the independent film movement. Right. Yeah, yeah, like. People. People have an appreciation for smaller stories told more intimately in an intimate way.
1:20:48 There's always going to be a place for real human handcrafted art. And by the way, that's probably in the marketplace going to escalate to a fairly expensive place where people are going to be. Those people who want that are going to be willing to pay more. To connect, yeah. You know, absolutely going to go out of existence. And here's the other thing I have to say, Mode Studios, since we started widely deploying AI, we haven't laid people off. We've hired more people. We have more designers working here then now that we're two years into deploying AI in all kinds of places than we did before. So it seems to me that like the evidence I have in hand isn't that it's eliminating jobs, it's changing jobs. You know, with the with the advent of things like follow me and Robos, you know, robospots, we no longer have rigs full of stagehands in spot baskets, you know, move. That job has changed because of technology, but it hasn't gone away.
1:21:56 Moving lights didn't get rid of follow spots entirely, right? Yeah. You know, things, technology changes the way things are made. It changes the way that that vocations happen and it changes what kind of careers are possible, you know, But the fact of the matter is that right now in the marketplace, AI, you know, we are searching for people with AI talents. We also want those people to be talented in using software like Unreal Engine or 3D software or other complimentary tool sets that allow us to put all these pieces together and make great. Contact but going back to like this whole fear thing and you know, like the common thing in our industry is just like I don't want anything to do with this AI. It's coming from my job and it's going to take away my job and you know, I used to my my go to statement on that used to be you know, no, it's really not.
1:22:55 It's going to provide you tools to make your job better and to make you more functional and more efficient. And and so that at the sort of lowest level, that is what it's going to do. It's, it's today, you know, like I had Kadena on the podcast and we talked about AI at one point and, you know, he grabbed the tool that I had shared with him and went out and started using it to create JavaScript for him on some of the things that he was doing and it. Very good at creating code. Correct. Yeah, so, so you know that I think at the very lowest level, that's the stuff that it's going to do. It's going to help you. But when you look at these robots, you know, both Elon's robot, this, this Pi 01, there's a Chinese one that I read about that's incredible and very close to launching right now. And you know, these things are twenty, 30,000 bucks and they can be housekeepers, they can be, you know, they can work in your shop doing tedious things.
1:23:56 I'm sure at some point, you know, they'll be able to work on lighting consoles or or sound consoles or whatever it is. Cuz I used to say, you know, the people who are always gonna be safe no matter what are the plumbers and the, you know, the people who do physical jobs. Yeah, I don't think that so much anymore. Like I think that's agree with. You. I think that, I think there will be a massive deployment of automation. Yeah. Of robotics, yeah. I think it is gonna change things. I think before we see robots behind lighting desks, the first thing we're going to see is lighting desks deploying more AI functionality. How you approach a console to create effects, for instance, Yeah, or even like you is like, you know, I know because I'm. I've been talking with people at various of the companies that that, that are making lighting console. Yeah, me too.
1:24:49 They're already working on being able to have a prompting model for lighting. Yeah. Like, so rather than saying, oh, I want group 5 at 50 in Group, you know, eight at 75 in Group 1000, and you're working towards making, you know, a blue mysterious night time queue, you're going to be able to prompt the console. You're going to be able to say, you know, create a blue mysterious night time Exactly. Focus mainly on stage, right? Speaking in English going to be able to do that. Right. Yeah, yeah. Look, there's going to be an impact. I, you know, and I, I think, you know, we're going to see it. We're going to see AI deployed in a whole bunch of technological ways. Yeah.
1:25:31 This is inevitable. And I mean, I think that I understand the people pushing back. Can I understand the fear? Change is hard. Humans are not wired to be constantly adapting. We crave finding cycles, finding rhythms of things we can do. You know, we're, we're going to plant the seeds, take care of the crops, harvest the seeds, make the food, put the, you know, like we, we have certain things that are genetically coded in US and behaviorally coded in US and holy crap, we are in a hockey. Stick curve big time, yeah. Technological change. That's really uncomfortable, right? But at the same time, like I'm so excited about it because of the control that it gives the humans. So like for example, you know, I don't remember what it was, but last week there was something bothering me health wise. You know, I just turned 60, so there's always something bothering me.
1:26:26 And you know, I would have normally gone to the doctor and said, here's what's happening. Bah, Bah, Bah. What should I do? Should I take a different vitamin? Should I do this? Like what, what is it that's causing this problem? And instead I use ChatGPT. And within 1015 minutes of prompting and, and going back and forth a little bit, I had the answers and they were really good answers. And it worked, you know? And so some people will say you're a maniac, that's dangerous. You know, how dare you trust your health to what the hell do you think doctors have been doing for the last 20 years? The same thing they. Go into Google. They either know things or they or they go to their bookshelf.
1:27:06 Yeah. Like, and they like. Oh. Judging by your symptoms, yeah. I I I'm pretty sure that you have a violent case of herpes, Marcel. No, but I mean, all we're dealing with, if we do it ourselves with Chachi, it's just a larger language model than what the doctor's been using, you know well. Exactly. I'm like just mentioned that they're really good at code. What? What AIS are really good as, as anything. That's a question of deriving a result from aggregated information. Correct. Super good at yeah. Right. Correct. So the nature of jobs is going to change. You know, is the lighting programming job in danger? It's in danger of changing. Ultimately, we might move the ultimately, and I'm talking, you know, probably maybe a decade out, we might be at a place where, you know, a production designer is now sitting next to a director and prompting the media servers and prompting the light just verbally talking to his interface. Yeah.
1:28:09 And there aren't as many programmers involved. Yeah, I think. That that might happen, yeah. It's hard to know. It's hard to know where it's going. I can tell you that right now it hasn't changed a whole lot. It is going to change things moving forward. It's even going to change them dramatically. We talked about at the beginning of the episode about how my daughter just went and got deep training in MA Three. Yeah. You know, judging by the crawling pace of lighting console innovation, correct. Yeah, correct. We might be also here, you know our industry is small enough where it's relationship based and all of those things.
1:28:49 I don't think anybody's going to want to be the first console that program programs itself, you know, because that's a scary thought because you're going to be boycotted and cancelled and all of those things when you're the first one coming out. There is that you have to get by that resistance to really bring. Into there will be resistance. Time there are going to be people who are super interested in that and who are even if they're not people working at the, you know, ETC or MA lighting or stuff. There are people who are interested in I've seen a product, I get assaulted with it all the time on my social feeds. Are these guys that have made a little box? Yeah, that was the one that got blasted at LDI last year, Yeah. And I think they did get blasted. But you know what?
1:29:33 That company is going gangbusters. Yeah. They are selling lots of those little boxes to little clubs. Yeah. Does it make sense, you know, to DJs and to people playing in certain things who just want some basic, you know, generative lighting to happen on tempo and, yeah, you know, moves of the thing? There is a marketplace for it. So one other thing. Stopping, you know, trying to stop technical innovation from changing what we know is like trying to is like going to the beach with a mop and trying to keep the waves. It's just it's. You're right. No, you're right. There's no stopping this thing. Like you either get on the train or it runs over you. And we decided to get on the train and, and I have had, I have had discussions with people who literally have come at me and said you're a vile, you are, you are architecting the end of art.
1:30:25 Well, while that may be true, I and I happen to be with you on the AI thing. Like circumstances beyond my control and time and the advance of technology. I've always been an artist interested in deploying technology since the very beginning. Yeah, since the very beginning, I and I can't ignore this. I want to, I am curious. I want to understand this stuff. I want to understand how it works in the process. Well, I want to understand how I. Remember how bad, how hard people fought against time code, how hard people fought against, you know, programming cues and then running them in sequence and things like that. Like, you know, advancements always tend to have people fighting on the other side, right?
1:31:06 It you run into to resistance always. But one of the things I want to spend a few minutes talking about because I know I don't have you all day is this this term prompt engineering. And so you know, I know that from my own experience when you take the time to learn how to and you can go out and Google and there's 1000 YouTube videos and probably 100,000 TikTok videos and stuff on how to do the best prompt engineering and all that stuff. But one of the things I've discovered recently, and I'm sure you're very aware of it, is anthropic. Claude writes prompts. So you can basically tell Claude to write a prompt for ChatGPT to do blah blah blah whatever it is, and you speak to it in English, it writes the prompt in prompt world and then you copy and paste it into ChatGPT and it works amazingly well. Yeah, and in fact, you should know that if you're using Dolly 3, which is Open a Eyes image generator, when you prompt Dolly three, it sends your prompt to ChatGPT for ChatGPT to rewrite it.
1:32:19 Really. Better prompt. Oh, that's interesting. And this is all behind the curtain. Yeah. You don't see it. And then ChatGPT feeds the better prompt. Back to Dolly three. Yeah, You know, Yeah. And the, you know, and I know some people who, who deploy this in a way I use, you know, the tools I use, like I'm a big mid journey user. My workflow is I tend to make images in mid journey and then if I need to then move to video, I'll take anchor images that, that, that are like the look in the feel, the aesthetic in the composition. I want to make them in mid journey. And then I go into Runways Gen. 3, which is a fantastic video creation tool.
1:33:00 Yeah. And I will use an image prompt. I will use the picture I made in mid Journey. I will put it into Runway. I will add a little bit of detail. And then Runway makes a piece of video. Yeah. The thing about prompt engineering and like and like using AI to help you prompt other AI is a the models move really fast, right? Yeah. And how you prompted, like for instance, in Mid Journey version 4, you benefited from making really long prompts, right? Like, like the more description you gave it, the better. When Mid Journey went to version 5, because of the way the model worked, it preferred you to be very direct.
1:33:43 Like it wanted you to deliver like a couple of sentences and then it filled in the gaps. You know, version 6 works in a very different way, which is kind of structured more in a way of like version six wants you to work in a in a very specific like it wants to you to start with style, tell it about subject, give it detail about settings, then information about composition, then information maybe about exterior forces like lighting. Yeah. And then add additional info like commands about aspect ratio and stuff like that. OK now it has format it wants you to follow and if you feed it a ChatGPT prompt into version 6 and mid journey your results might be hit or.
1:34:33 Miss, you're going to get crap. ChatGPT has not caught up with the format that mid journey. Prefers, right, So when we're talking about prompt prompt engineering, so like ChatGPT, if you go to it and you just say, write me a blog article about, you know, the live events lighting industry, you're going to get shit right. You're going to just get. You're going to need to do, and I do use ChatGPT to help me write things, Yeah. But usually I start with something like, you know, if I and then I'll feed it to ChatGPT to help me. That's me too.
1:35:13 Polish it and. Or vice versa. Or I'll do it the opposite way. Out of ChatGPT and then I go back and that's. What I do and then yeah, yeah, that's what. I do. My work with the large language models is very back and forth iterative. Yeah, like if you just say give me a brilliant article, you know, about the current state of the lighting industry, you're going to get garbage. Well, you know, it's like I had people tell me that Grok in the fun mode on Grok, it wrote a lot better things than ChatGPT where you could actually copy and paste right out of Grok and and use that article. I don't think so. Like Grok, Grok tries to use AI humor and it's just so stupid.
1:35:54 You know, it's just like it, it's nonsensical, right? It just doesn't work, doesn't work. It makes me throw up, not laugh, but but yeah. So I mean, you know, the thing about AI, I guess in closing is that that I say to friends all the time, just start using it like find if you want, like I tell people and I'm telling people again here. And I'm sure you would do the same if you come up to me at LDI or you send me an e-mail or a text message and say, what are you using for spreadsheets? What are you using for writing articles? What are you using for images? What are you using for this or that? I'm happy to share that information with you, you know, Yeah.
1:36:36 I mean, it's it's like, like people ask me, you know, it's just you and Sarah doing the podcast. How does Sarah do so many posts? Like she's always posting YouTube Shorts and Instagram reels and does she not sleep? And the bottom line is she sleeps really well. She's probably sleeping right now while we're recording this. And but we use AI apps to do some of those things. So like there's, there's one called Opus clips that we use pretty religiously, which takes a YouTube video of our podcast and it grabs like 20 or 30 clips out of it. And then you look at them and you find the best ones. But it's looking for things that might become viral, things that make a lot of sense, things that, you know, it thinks people are going to relate to and, and it works pretty well.
1:37:22 And, you know, so there's a lot of really logical, really good things that I use on a daily basis that I think people would benefit from and at least get comfortable with it so that when it moves a little deeper into your world, you know, you're not afraid of it. That's my advice to people too. It's, it's like, look, if you're a professional technician, if you're a, if you're a designer, if you're a creative director or creative producer, you can make the global decision, do I want to use this stuff or not? And I honor that, whatever your choices are. But the fact is that this technology is coming. It's not coming. It's here. If there's only going to be more of it, it's only going to be more pervasive.
1:38:04 Yeah, you can, if you're interested in participating in this industry and being a professional in this industry, choose to learn about it and use it or not. Yeah. And, and again, I honor whatever decision you want to make. I've always had the predisposition to want to figure out how to use new technology and what I do. I was one of the first people that figured out media servers and using media servers. You know, I was, I, I, you know, I've been along for the ride in terms of like, you know, moving lights coming into the picture and then, and then, you know, and then the advent of LED and now this diversity of fixture types that do all these kind of different things. I've always been an artist that wanted to incorporate technology into my design.
1:38:50 So I'm curious about AI. Yeah. And since the beginning and, and, and as you heard in the podcast, you know, we've been deploying it really since like 2000. That's. Crazy. Yeah, I didn't know that. That's such a cool story. So I, you know, I'm not, I'm not slamming the brakes. Yeah. And keeping up with it. I spend a lot of time educating myself on it. We've certainly been able to do things we've never been able to do as a creative agency. We've been able to swing way above our weight because we've been using using AI to help us put together creative pitches, to help us put strategy stuff together to give our work underpinnings that are stronger. We've been able to compete with agencies with way bigger resources than ours because we have. That makes so much sense.
1:39:36 Listen to us as a small business. It's your superpower. It's your superpower. I'm interested in it. I'm curious about it. My feeling on it is if I'm talking to people that are new or just coming into the business, learn about it. You can choose. Whether or not you want to use it. But learn about it. Learn how it is used. Learn how you can deploy it. Learn how you can deploy it to sell yourself. Sell your work. Improve your work. Yeah, be more nimble. Save time, right? Save money. Well, there's a part of your job that you hate doing and there's a really good chance that there's AI tools already in place that will enable you to do less of that, whatever that is. So, you know, that's, that's the bottom line for me is, is AI, like when I saw the, the clothes folding thing from, from Bezos yesterday, I was like, oh, really? It can fold your clothes for you. I want one of those, you know, that's perfect. So are you doing anything at LDI as far as like are you doing any any speeches or talks or anything on on AI stuff? I will be I will be talking on the 10th. OK.
1:40:49 Along with some of my friends from Tate. OK about placemaking. We're gonna have a panel that's part of the the DSE, part of LDI, the typical signage people. Yeah, we're gonna be talking about interactivity and technology used in creating immersive experiences for permanent installation, so I'll be talking about that. I'm also apparently one of the judges for Battle of the Busk. Fun. So I'll be, I will be there as well. And then I'll be walking the floor and. So there's no AI? There's no AI talks or anything that surprises me. I thought they'd do something. I think there is the there is the X live part of. LDI yeah, yeah.
1:41:31 Where they I think they are having and Jake Pinholster, OK is leading that this time around. And I think that Jake does have several panels about AI and how it's being used. Cool. I'm not talking about it this year. I've been, I was, I've been, I was too busy this year to commit to, yeah, to doing a lot of stuff at LDI, but I will certainly be there. I will be talking on the 10th and I'll be around and about if anybody wants to talk to me one-on-one. So. Yeah, buy him a beer at the Circle Bar and he'll tell you whatever you want to know about AII. Don't know, you know, having been having been clean and sober. For about don't buy him a beer, buy him a. Coffee.
1:42:14 I don't know that I'll share the beer, but we have fizzy water with a slice of There you go. There you go. Well, that's probably about $27.00 at the Circle bar, so. That's the problem. That's a that's a nice fee. All right, Bob, thank you so much for doing this. I appreciate all your time. You're you're certainly very, very up on this. And I'd actually love to buy you that soda water with a, with a lime and, and have a deeper talk because I'm not a technical guy, but I'm really an AI guy. I love the tools that it's providing me. So I I play with them every day. Thanks.
Apple Podcasts
Spotify
YouTube