AI’s Army of Interns: The Future of Video Production
March 13, 2025
Artificial intelligence isn’t replacing video producers—it’s revolutionizing how they work. In this eye-opening conversation, we dive deep into the practical realities of using AI tools in professional video production, revealing a landscape far more nuanced than the typical “robots taking our jobs” narrative.
We share our firsthand experiences using AI as “an army of interns” handling time-consuming tasks like location scouting, which saved days of work while we focused on creative decisions that matter. You’ll hear concrete examples of how we’ve implemented AI across the production pipeline: generating pre-visualization imagery for client presentations, creating projected backgrounds for studio shoots, and even ethically producing voice content with proper compensation for the original voice actors.
The conversation takes a fascinating turn as we explore the ethical dimensions of AI-generated media. Where is the line between enhancement and fabrication? How do we maintain authenticity while leveraging these powerful tools? We discuss our approach to these questions, including creating contracts for AI voice usage with term limits and fair compensation—frameworks we developed despite operating in what feels like the Wild West of media production.
Perhaps most compelling is our discussion about which aspects of creative work remain distinctly human. Documentary filmmaking emerges as potentially more resilient to AI disruption because of its foundation in authentic human connection—something we believe will continue to resonate through technological advancements.
Whether you’re a creative professional navigating this rapidly changing landscape or simply curious about how AI is transforming media production, this conversation offers valuable insights into adapting and thriving with these new tools. Listen now to discover how embracing AI might be the key to unlocking your creative potential rather than threatening it.
FREE KICKSTART GUIDE: https://letsbackflip.com/guide/
Topics in This Episode
- (0:28) AI in Location Scouting
- (6:04) AI for Pre-Production Planning
- (16:18) Ethics of AI Voice Generation
- (22:54) AI for Post-Production Enhancement
- (32:54) The Future of Human Creativity in AI Landscape
- (45:44) Authenticity in an AI World
- (50:26) Final Thoughts on AI Integration
Transcript
Ryan Freng:
Welcome to the Backflip Effect, the podcast that proves there’s a method to the marketing madness, specifically, a method that involves clever video strategy and a dash of creative mischief. Now, don’t be fooled by bland ideas or big budget fluff. Instead, we tackle real client questions and share how we’ve transformed businesses with strategies that actually work. Grab your headphones and join us as we explore the nitty gritty of turning everyday marketing woes into story-driven success. Let’s get started. You kind of brought up this idea or Max mentioned it, but you kind of codified this state, this idea of the state of AI in video production, which I’m super fired up to talk about, because a lot of people think AI is going to take over production. It’s going to take over from creatives and producers and videographers and all that stuff, but we’re actually on the front line using it. You know it’s like the opposite.
John Shoemaker:
Yeah, I have no idea how far it’s going to go and right now I’m just letting that thought just stay in the background, to the side, just to not be anxious, right. But for now we are full steam ahead, 100 miles an hour, going after AI stuff. So just my quick example I’ve been using AI for the past two days to location scout a nonspecific location somewhere in the country that has a front yard and a backyard that we can film in um, that’s available in a certain time of the year, uh, that has green grass, you know, and that looks Midwestern. How do you search for that?
Ryan Freng:
Uh, what you could spend a lot of time. Yeah, what search?
John Shoemaker:
terms you use. I could just describe that, and I was actually using the deep research which is available in ChatGPT. Now it’s also available in Grok 3. And I don’t know what Claude has, but there might be. You know something, and it’s just digging through. I could watch its process and what it’s doing. And it’s just digging through. I could watch its process and what it’s doing, searching through hundreds of web pages and then giving me 10 results or something, and then I could check those real quick and see are those dates actually available? Does this work for my need?
Ryan Freng:
But you know, it’s just, it’s like AI’s army of interns, like saving you tons of time, Like I think you said, like two days worth of work probably.
John Shoemaker:
That’s my guess, my guess about how long that would take to dig through and come up with that many results, and I could have that running while then I was doing something else too.
Ryan Freng:
So that’s there. Is there any concern about you know, I know AI kind of hallucinates sometimes and we get crazy results and things like that. Is there any concern about that, with what you were specifically doing?
John Shoemaker:
That that is a concern. And there’s another example that I have where it was just making things up earlier today Um, that other spreadsheet example, but uh, this one was nice because I specifically said I need the direct links to the results you’re finding so that I can check them. And it gave me links and then I could just click into there. It did mess up a few times and say it appears to be available. You know, the last week in March and I’d look and it wasn’t or something. But that’s a pretty quick, just look at that, see if it’s yeah you’re essentially verifying your data at that point.
John Shoemaker:
Yeah, so not too bad, but that’s so. I’m finding a lot of uses in pre-production just for a lot of those little detail, ticky tack things that you have to do. Um, yeah, just things that are like, I don’t really want to apply a lot of my brain to this, I just need a lot of stuff, or even what it’s.
Ryan Freng:
It’s maybe, maybe even more so that it’s not like hard brain power, but it’s just time.
John Shoemaker:
Yeah.
Ryan Freng:
So doesn’t take, you know, like we could have an intern do something, or like one of our kids help out, like that’d be a good use of that time, but like for you as a creative director, as the producer, you know, to be able to save two days and let the computers pull up the data and then you still have to verify and then reach out and do the connecting, like that seems like a huge win. And I think when people are like, oh, video production, it’s going to take your jobs, I think in general it will, you know, take up maybe 50% of our tasks, so we can then spend that time doing more important things, like you know, or calling them and seeing if they’re available, or some other producing with staff.
John Shoemaker:
Yeah, there’s so many parts of this process on this particular project that are gonna have to happen that if I can shorten the length of time on any of them, it’ll help a lot Because, yes, I need to reach out then if we get that and get all the permissions and things like that. So, without going into more detail, that’s what that one is doing location scouting.
Ryan Freng:
Also, you’re using Grok, and that’s weird to me because I’ve never used Grok. I’m aware of it, but in my mind it’s AI for Twitter, like AI for social media, but it’s just its own AI platform.
John Shoemaker:
Yeah it’s strange right now because it’s in X. It’s in the X app. I think you can download the app for your iPhone. Android is still waiting to release their app, as usual. See if I can get it and it’s available to me right now. I’m not like a premium user on X.
John Shoemaker:
And I can use it in beta. So I don’t know if it’s going to go behind a paywall at some point, but right now I’m use it in beta. So I don’t know if it’s going to go behind a paywall at some point, but right now I’m using it. But yeah, I mean, there’s just so many uses. You know, jumping off of that and getting into some other ones that I think we’re going to talk about here. Talking about pre-production, we had a shoot where we were doing this camping chair product in our studio and we had kind of a fun idea we were going to put a background in, but we didn’t want to be green screen. I don’t know where and why we came up with the idea, but it was just something that I think we saw somewhere. It was because of Jordan’s music video right.
John Shoemaker:
I think so.
Ryan Freng:
Jordan Post, our cinematographer friend had done a shoot in our studio, used a projector as a background, and we’re like that’s really clever, as opposed to green screen, which tends to be a lot of work, even though you’re like, oh, we’ll fix some posts, it’ll be easy. It tends to be a lot of work.
John Shoemaker:
Yeah, so we knew we wanted to do that and so, even from the pre-production phase, in giving some kind of storyboards I don’t know if they’re a cross between storyboard and mood board at that point I’m not the best. I could draw stick figures or take way too long drawing better things, or we could hire a storyboard artist all these things which we don’t always have budget for, on all the projects and this was a very, very small project.
John Shoemaker:
Yeah, and in this case I just described what I wanted into an AI engine. It was DALI at the time. I think now that’s incorporated in ChatGpt, but there’s again there’s many um and described I, I need a guy, and we had these gags too, so the chair is super light, and we had these gags that are like I want him carrying an armchair.
John Shoemaker:
so I need a guy carrying an armchair in a studio in front of a projected background of a mountain range yeah, and then just kick out these images for me, and it gave us a bunch of images that were like okay, that’s kind of the general idea and that’s what you need in pre-production to show the client be like, here’s kind of what we’re thinking we want to do right, including even giving me like a not like a non-storyboard image of a different angled like, like a pre-visualization of the whole studio setup.
John Shoemaker:
That I thought was like even like a map of how to actually do it. Well, not a map, but like a three-quarter view. That’s like not, this won’t be a view, that’s in the finished piece, yeah, but like as though you’re a fly on the wall in the studio taking a bts photo, sure, and I felt like that was even useful to help the client understand, because sometimes they’re asking those questions even though I don’t think they matter for the client, you know. But sometimes clients are just curious well, how do we do this? And it was like here, like, and it was kind of a goofy ai image, but it was like this kind of gives you the general idea and it was like got it?
Ryan Freng:
yeah, totally understand it well and and that’s a that’s a whole separate thing like expectations, client expectations and I have heard from clients like I didn’t know what to expect when I showed up on set Um, I think we’re going to talk about set in another podcast but, like when we did the Kelly Jackson, we did a whole stage. Uh, as you know, that pivoted a whole scene or whole set. It was a floor and a wall that we could change everything and the whole thing pivoted. It’s crazy. She had no idea what that was going to be like until she showed up.
Ryan Freng:
You know we talked about it, we I think we even showed pictures from the studio build out, but it just didn’t communicate what it was going to be like and how that was going to work until she got in and saw it. And absolutely I would have helped with that to be able to previs and be like, yeah, it’s going to look like this and yeah, you’re going to be on the wall and you’re gonna be on the floor and you know it’s going to work like that. So that was that. That’s actually a really interesting maybe byproduct, like unintentional byproduct that had helped just communicate some expectations.
John Shoemaker:
Yeah, so that was a cool one and I think the final result there too. So that was a cool one and I think the final result there too. We generated a few background images in AI, I don’t know. I mean, there’s debate right about the ethical ramifications for creatives, and where is it borrowing ideas from Right, right, and then how does that work for copyright? Yeah, is it? Where is it borrowing ideas from and right, and then how does that work for copyright? Yeah, but you know, to whatever extent it’s making things up again. Small project didn’t have a lot of budget. I need high quality imagery of mountain ranges. They don’t have to be anything specific, so, but I can’t buy stock footage, even for a hundred bucks a piece here or whatever, because I need multiple. Yeah.
Ryan Freng:
So there’s budget restraints and then very specific idea for the creative that we weren’t necessarily able to find in stock.
John Shoemaker:
Yeah, and I think I can’t remember if we ended up with still imagery for all of them. We might’ve when we did that shoot. It was a while back, so we’re light years ahead just in the speed that AI is moving. Yeah, like the video wasn’t good enough at the time, I don’t think to give us like a, give me a video.
Ryan Freng:
This was 2024, right.
John Shoemaker:
Yeah, like last spring.
Ryan Freng:
So even call it. A year ago we didn’t have AI video as a viable option.
John Shoemaker:
Yeah.
Ryan Freng:
It maybe didn’t look so good. But now fast forward. We’ve used AI image backgrounds. I don’t think we’ve actually used any AI video yet, have we?
John Shoemaker:
I don’t think so. I used any AI video yet, have we? I don’t think so. I’d have to think more.
Ryan Freng:
Maybe Max and Alexis have, and we just don’t know.
John Shoemaker:
They’re so good with it I’d have to think more about it, I mean, but that’s a good segue into where we almost used AI video with our client. Are we name-dropping clients?
Ryan Freng:
Yeah, that’s fine with ho chunk um, we had these jackpot.
John Shoemaker:
You know these gurus wearing the sequin vest and all that stuff and they’re announcing the jackpot numbers, like the weekly or daily jackpot, so they need to be able to update that frequently without the cost of a production company and there we used well, it didn’t get, it, didn’t pass the final test, yeah, but we did use a full um copy you were the first test subject right, who ran through it of the person saying and this week’s jackpot was whatever, um, but it was a little too uncanny valley, uh, and we’re probably still a little bit uncanny valley zone, although I’ve seen some that are like, yeah, mind-blowingly good, did you see?
Ryan Freng:
that one that I. It maybe begins with an h, I can’t remember. I don’t know if you saw it, luke um, but they showed like four. They’re like original performance ai, ai based off of image, ai based off of video, ai based off of full video and things like that and like all three of the AI videos are ridiculous.
John Shoemaker:
Yes, I did see that one. I don’t know that that one was out to the public yet that was like an early that’s like a GitHub.
Ryan Freng:
You got to go download code Example. Yeah yeah, it kind of sucks. So I almost did that. I went to the repo and I was like what do I got to do? And I was like I don’t have time for this, so use another package it be like how do I oh my gosh throw ai?
Ryan Freng:
so, okay, a little bit of a tangent, but I love this like using ai to help you with ai. So if, if you’re overwhelmed you know like I feel like I’m overwhelmed by all the AI stuff Like I saw it as somebody asking ChatGPT help me write a prompt to generate an image. So here’s what I want out of the image, but help me write a prompt that will generate a really great image. So that idea of utilizing AI to generate a prompt to do something else, you know, and like you’re saying like AI helped me to set up this AI, I think that is, it’s crazy. It kind of blows my mind. But that is like a key use of AI uh, you know, helping to use other tools and itself which is a little bit of a mind.
John Shoemaker:
So that’s the encouragement I think for anyone is. Maybe I don’t need to make these analogies, but it reminds me of like um, watching older friends or family members, uh, struggle with like a VCR or something. But why are they struggling? Is because they don’t want to touch any button. They don’t know which button to touch and they won’t touch any of them, whereas, like our generation, I feel like we grew up it’s just like I don’t know just which.
John Shoemaker:
Which one right you know like familiarity there’s this or even just like this, I’m not concerned about breaking it. I don’t. I’m aware that I’m not going to break it by trying different things. So there’s like a disconnect, a generational disconnect there. This is maybe just an encouragement for for any of any of you who haven’t gotten into it yet. Like you’re not going to break anything.
John Shoemaker:
You know, don’t, don’t start using operator yet, or don’t start using the ai’s that can actually do things on your computer, but just use the prompts like go in, the stuff is free, yeah, and just generate a prompt and see what happens. You won’t break anything and you just need to like try stuff. And that’s like, I think, what we’ve always done in our company. That’s why we’re like able to speed through all of this, because because it’s taken a ton of trying and a ton of like learning, but you can’t really like learn without doing, because there aren’t a lot of books on this yet you know there’s not like well and we’re small enough that we don’t need a AI department to have AI ethics like.
Ryan Freng:
We have our ethics, we have our core values and all of that informs everything we do, all the tools we use, including AI, so we can go really quickly without having to spin up a committee and go through.
John Shoemaker:
So let’s talk about ethics for a second, because here’s the example. So with the Ho-Chunk Gaming project, we bailed on the video because it was too uncanny valley. It didn’t like look great, but we did end up using.
Ryan Freng:
The final products are going out and, to be clear, it looked like 98. But what it was is like every you know, there’s just, there’s just a little bit of movement that doesn’t look right, like most of it was really good. But then you catch, like a you know, something like what was that move?
John Shoemaker:
that doesn’t make sense yeah, um so, but we did end up using uh ai for audio, so there is ai audio in the final pieces. They wanted to be able to be able to update the jackpot numbers on a weekly basis, or some some really fast you know interval. Right and so this isn’t defined anywhere, and not to toot our own horn or pat my own back or other awkward, strange ways, just toot your horn and pat your back at the same time.
John Shoemaker:
Yeah, we hired actors and then we said, okay, we wrote a contract just saying that we won’t use the because we had to get an AI print of their voice in order to do this.
Ryan Freng:
So we paid them a little extra than the day rates yeah, essentially a licensing fee for a very specific use of their voice.
John Shoemaker:
Yeah, and then set a term limit and all of those things. All of that was self-imposed. We decided to do that, knowing that, oh, we should do this. It’s you know, it would be better for the actors to do it that way, but it’s kind of the wild west and like the other way to go yeah, the other way to go is people are just like haven’t got, you know, saying no or didn’t get into it.
John Shoemaker:
Although there’s some hollywood, they’re like you know, bruce willis, like licensed, like into it. Although there’s some Hollywood, they’re like you know, bruce Willis, like licensed, like everything.
Ryan Freng:
Well, and that’s because he’s sick and he’s dying and what’s his face? Darth Vader, james Earl Jones did that before he passed away, which is awesome for Darth Vader, I think. But yeah, the ethics, the ethics are interesting. I want to go back to the old people using things. The other part, too, is like you don’t need, like they don’t need to ask a friend or a young you know, like a grandchild or their child for help or anything. They could literally ask the AI like hey, how can I use this better? Or how can I use you to do this, or how can I work with you, maybe say please and thank you, just in case they take over, but you can ask it how to help.
Ryan Freng:
Because I think that’s a paradigm shift that I am still, you know, like in the midst of, because, like, google is like such dumb search and I hate it, even though it is crazy awesome in terms of search and maybe a knowledge system, you know, surfacing knowledge. But anymore I don’t want to go to Google and like ask a question and then get 10 nonspecific results that I have to look through. Like I just want to go to ChatGPT as an answer engine and type my question and get the answer and be like can you source that, and it will come up with potentially a better, more relevant answer than me having to search through google. So I ideally still get to really good content, but it saves me time yeah and um, they’re adding features all the time.
John Shoemaker:
So in uh, chad, gpt, you can. So I’ve been playing around with deep research, which is a limited feature and I’m almost out of uses of it, for my whole month already. But, um, I think in one of the models there’s one called just deep search. Um, I’m just pulling it up now, or maybe I’m mixing this with Grok, I don’t know. They’re getting all kinds of things. 10 deep research available I think Grok was deep search Chet.
John Shoemaker:
GBT was deep research. So Grok was deep search and then Chet, gbt is deep research. So maybe it is that the deep search is to some extent these AI models have knowledge already or they’ve been building up knowledge right from things that they’ve been fed. The deep search stuff is better for what were the examples? It was like current conditions. So I used the deep search as well to say, um, what areas are most likely to have grass? You know, green at the end of march for this, like cert the location, and that’s not exactly the same every year. There’s like um, there’s like averages of like, when it is usually like that. But this could also look at a current data you know and say, like these are the current conditions of what’s going on, and like I want you to actually go look at web current web pages, recent ones and read what’s going on and then give me that information back.
Ryan Freng:
Yeah. So that’ll be really interesting too, cause I know Pete Chad GPT makes stuff up. You know, when I’m doing some research on a topic I’ll be like give me some content around here on this from this document and cite your sources, and you know I’ll get kind of a good summary and then quotes. And then I look at the quotes I’m like those quotes don’t exist. And then it’s like, oh sorry, yeah, I paraphrased and combined a couple different pieces like no, no, if your quote you, it has to be a direct quote, so cite only this one. And eventually I got to a prompt that was very specific about cite only this, um, this document. And eventually it did come up with quotes that were cited from that document.
Ryan Freng:
But a little bit of the hazy. What is that called Hallucinating? Easy for me to say the ChatGPT hallucinating. But it looks like deep research. It’s optimized for web browsing and data analysis, leveraging reasoning to search, interpret and analyze massive amounts of text, images, pdfs on the internet, pivoting as needed in reaction to information it encounters, which is interesting. Because we’re like isn’t that what all chat GPT does? But no, like original models, it’s just like you know, you take just content from all around the world and build something out of that content in an intelligent way.
John Shoemaker:
That’s the original learning language models LLMs but now we have reasoning models which are different yeah I, I can’t even get into the, you know, trying to understand it from that point.
Ryan Freng:
But um, yeah. So maybe back to some of the, the ideas of ai in production. You, you had mentioned voiceover, so that’s one that we use seamlessly to improve audio when there’s a problem with audio too.
John Shoemaker:
So we’ve done full voiceovers with just a full on AI voice and had clients be like, yeah, that sounds excellent, I like that, and gone with that. We’ve also used it. Sorry, vor, I know I mean it’s a changing industry and I don’t know there are going to be areas that go away. If you’re a famous person, your voice will still be valuable. If you’re not a famous person and you just have a good know, a good low voice like I don’t know, I don’t know what to tell you.
Ryan Freng:
That is like really one of the only industries so far that I’m like I don’t see how that stays. Yeah, Because of the way the technology is so good.
John Shoemaker:
We’re not quite there yet, but stock photos and stock videos are also on the chopping block, like they’re not going to last.
Ryan Freng:
Good thing we don’t make money that way, yeah, and we can save money in fact, yeah.
John Shoemaker:
Yeah, so, but not full voiceover. But, like we do a lot of documentary stuff interviewing people, and sometimes somebody’s talking and they don’t really finish their sentence, you know, and you’re like, oh, that was a great, a great spot to. You know, but they kind of like went up at the end and we can plug that into is it 11 labs that we’re using? And just give it the instruction I want that to end and, in that person’s voice, change their inflection so that when they are talking and they finish their sentence and it drops.
John Shoemaker:
You know, uh, editors know this that sometimes you’re like searching through everything looking for an s that you can attach to the end of a word because they forgot, they accidentally said it singular and it was supposed to be plural, right, and then you’re doing all kinds of goofy editing and trying to make that sound natural. And and then you’re doing all kinds of goofy editing and trying to make that sound natural and blend them together. It takes a lot of work. We can just say we can just take that input, the word uh, change the word, add an s and export it and just have them in their voice saying it plural or saying a different word. Right, again, ethics. Ethics need to be followed to the T, and you know we don’t have any issue with that, like we don’t have any reason not to be ethical.
Ryan Freng:
Well, I mean it’s. It’s also just like you’re describing it. It’s just editing. We’re not doing anything to make them say something they’re not. I mean the news does that. The news will soundbite something and make somebody say something they’re not, but what we do is we try to enhance and make somebody look and sound their best. So, like you said, if we’re just editing to find us to actually finish a sentence, that’s the same thing, but we do it with less time and it sounds better with the AI voice tools.
John Shoemaker:
Yeah, I was going to go on another. You know it’s kind of a tangent, but it is going to become a thing. It’s already a thing and we’re talking about ethics. And we’re talking about ethics and I mean, if you have sources that are not ethical or you’re not sure or whatever, we’ve moved into an area with all this that people, average people, have no clue. And I think, being so close to it, I’m aware of it, we’re aware of it in our company, but other people are’m aware of it, we’re aware of it in our company, but other people are not aware of it. I’m like, already we we’re already in a spot where you, you truly can’t believe what you see if someone’s not being, I hate that on like social media if somebody’s not being ethical, I can make.
John Shoemaker:
Right now I can take any person, public figure, actor, politician, and I can make them say anything I want. And usually the ones that you see that you catch on to are goofy. You know it’s just somebody doing something crazy. But if you didn’t make it crazy and you just tweaked it slightly and put it out there with like no other fluff and it was just like looked like a boring press conference.
Ryan Freng:
Yeah.
John Shoemaker:
Right out there with like no other fluff and it was just like looked like a boring press conference. Yeah right, well there’s, there’s already people doing unethical things without ai.
Ryan Freng:
Yeah, the way you cut it, the way you soundbite it, the way you share. You know, I was listening to the joe rogan podcast and they were talking about it was joe and elon musk and they were talking about how the news is blasting them and Joe, in one of his commentator roles at the UFC, he did something and put his hand up like that. I don’t know if it was to stop someone or high five or whatever, but they show that and it looks kind of like Nazi, yeah, like a Nazi salute, and they keep showing that in order to show him like he’s extreme, he’s bad. All that stuff like that’s obviously nefarious, yeah and that’s not even ai.
Ryan Freng:
So I’m like you can’t trust anything that you see or read. I mean not even this, like the only like I can trust anything you guys say in this room, but beyond that you can’t trust any of it.
John Shoemaker:
Yeah, that’s scary so, okay, yeah, I said I, I said those are the parts that I just like. Well, set that aside, you know, for now, leave that in the back of my mind and for the meantime, just run as fast as we can into this world of ai. I mean selling Luke earlier. I’m like I don’t, I don’t know how long this will last. Next year, this could, we could just be doing this via AI. I don’t know how fast it’ll go. Like you, think about where we were last year or two years ago. Go one or two years forward and there’s probably a model where I sit in front of a camera, I read a script, I I turn my head this way and that way and I give it all the information that it needs. I’m like, okay, great, use that model. That’s me in the podcast, and then that’s ryan, and then we just like yeah, here’s like hours of my our random babbling we don’t even need that.
Ryan Freng:
We could just say what is a hot topic right now for video production that people are searching for. Yeah, write a podcast in our voice yeah with ryan and john talking.
John Shoemaker:
Yeah you know like, well, that’s what I meant. It’s like you could train it.
Ryan Freng:
Yeah, like here’s, here’s the source material of us talking, and then and then say, yeah, write a new one I saw an agency video today that was a vsl like video sales letter and it’s like here’s who we are and what we do and how we do it, and had graphics and things like that, like they typically do. But I’m pretty sure it was an ai model. Yeah, you know, because the, the company was like a older man, you know, maybe in his 50s or 60s, is like the CEO and founder, but then it was like this nice young lady telling us all about the business and you can kind of tell a little bit, but without a doubt, people are doing that now to varying degrees of success. So you know, the question kind of returning to marketing, like what is what is good? I mean, if it saves time, that’s good. If it helps convert, that’s good.
Ryan Freng:
The human in the loop, the creative that humans have there’s something that’s still unique and special so far that AI hasn’t captured? I don’t think. But in as much as we can use these tools to create great messages and get them to the right people, I think that’s a great use of it, and I don’t know where that line is.
John Shoemaker:
And we can definitely like the AI topic. There’s all these fascinating things, but yeah, clearly we end up philosophical, because that’s like where I don’t really know either. Like right now we’ve been using it to just, uh, augment things that we’re creating. You know, I just I just need a background and I don’t want to like search everywhere and try to find the exact right thing, shoving it into this part of the conversation. We did another one for M-State Interact and we needed backgrounds and we hadn’t pre-selected the backgrounds before shooting green screen. So if you’ve ever shot green screen and you do that kind of production, you know that you want a light to whatever the background is, where the light comes from, the angle, the height of the camera, all that stuff.
John Shoemaker:
Well, max was able to have these AI image generation programs kick out backgrounds and just keep asking for tweaks and stuff to get the angle where he wanted, to get the space the way that he wanted and and get the lighting coming in from the correct angle. You know, I want the light coming through a window on the right side of the frame or whatever, and then he could put that in the background and make that fit pretty well, which is, again, not the ideal way to do it. You know we were kind of working in reverse on that one, so we’ve been doing it to augment things and add things. Another example before I answer. Finally, the philosophical thing. I really thought it was cool.
John Shoemaker:
I saw and I won’t name a name because I might have it wrong, but I saw somebody that we follow on linkedin using some of their photography and then they plugged that into like a video generation engine and got their photos sort of like re done, revisualized, as like short video clips that are like in slow-mo or something. Yeah, that’s wild and it looked so cool and it was like a really cool demo. So I’m like that’s opportunity for b-roll. Think of like the opportunity for documentary. You know, you see like people showing examples. Now imagine an example of a documentary. All I have is like some like granule photo of my grandpa, you know, but I could get him like turning and kind of looking you know I love that so much because people are always like here’s, here’s photos, put them in there.
Ryan Freng:
I’m going to send you digital photos. Yeah, like oh my gosh, digital photos in like a video are so terrible. But if you could do something like that I mean we have done. If you can composite the 2D, 3, you know, 2.5d effect, but if AI could help with that, then be like yeah, send me all the photos. I’m going to make them look baller.
John Shoemaker:
So I think I think it’s all good, as, while we are still achieving authenticity and that’s the part that I don’t know, like, where is that line? Where does the line go too far? Where is it blurred, like if I’ve prompted this thing to a certain extent and then I have, um, you know, I’m writing something, you know how much can it write, you know, with the guidance of my prompting, before it’s like not authentic to me or to my voice, and then as a user of it, as a consumer, I know, from my own perspective, I’m I’m less interested in like, reading, like something written by AI, if I can tell. But that’s the key thing, if I can tell, like, if I can’t tell, I probably keep reading it, I’m interested by it. And then if somebody told me later, hey, you know, that was written by AI, I’d be like, oh yeah, interesting, you know like, well, I mean, to to take that even further, thinking about animation and 3d animation, when it was originally maybe more, less rigged and more manual.
Ryan Freng:
But then now you can completely rig a character and then make them move around in a space, and how their body reacts and their clothes react to the things is just the system that’s generating. You know gravity and force and wind and pressure, and certainly they can tweak it, but you know it’s more and more computers doing that and I yeah.
John Shoemaker:
So I did see one example there too. Um, so if video right now is struggling because of uncanny valley and it’s like looking a little off animation you can totally do animation. There were some examples I saw with. Was that on? So I think Sora had had had the animation examples or maybe it was that same link, the same video that you had, where they were breaking down the guy in four panels and it was like here.
Ryan Freng:
It is just with the arms.
John Shoemaker:
Here it is with the right. Yeah, that one had an animation example and it was like here’s your input, you know, a 3D animated character, just a still image, and then animate that and that. That. I mean what? Where’s the issue there? You know there’s. That’s already like you’re not going to catch up on or not going to catch uncanny valley from that because it’s an animated character, right? So maybe that’s another. You know, desolate future for the industry of like animation, right, once you can give it a model you know, and have the AI sophisticated enough to not don’t change the look of the character as you take it through this stuff, you know, but I mean even the animators.
Ryan Freng:
You know voiceover artists. I can’t see how they don’t get taken over. But an animator like somebody still to guide and influence the prompt you know. So you become less of a computer technician and more of a prompt engineer, prompt technician. You still have your creative eye, you have your creative modality. Like you’re still driving. It’s still a human creativity that is driving the tool. But I, like vo and I have friends who are voice over artists. Like I don’t know, like I stopped responding to vo emails, I’m like I don’t have time to manage hundreds of people when I can either use ai or I can go to a service and the service can just find people for me. And even those services have AI services that they’re providing.
Ryan Freng:
I don’t know if we’ve used any of those yet I’m sure they have to.
John Shoemaker:
Yeah, there’s the taste tastemaker, you know, is that?
John Shoemaker:
what you know just like that, that is maybe the last you know. Is that what you know? Just like that, that is maybe the last you know. The last line that we hold as all things happen is like, well, who will make the decision about where the key light’s coming from and where the right, so that that does become, yeah, the final bit is like designers, people with an eye, people who know what they’re looking for, who know kind of the rules of I mean, even this could change but sort of the rules of cinema, the rules of good photography, the rules of ratio and stuff like that. To some extent it’s almost it’s kind of subjective, but there’s almost an objectivity to those rules, which AI could pick up on.
John Shoemaker:
They could pick up on that? I’m sure they could, yeah, but that would be a way, you know, that’s a way to stay in it for a while, you know making decisions about what you’re having. Ai do.
Ryan Freng:
I wonder if there’s, like the industrial revolution was a huge change where it took a lot of people’s jobs. Computers to some degree took some people’s jobs. Now AI, and AI is a hot topic and it’s in everything. But I was reflecting with somebody else last week that it’s, you know, it’s a tool like those, like computers, where, when they kind of came on the scene, and microprocessors and chips and RAM, each of those inventions revolutionized computing. But then it cascaded to everything. You know there’s almost nothing that is not computerized at this point or mechanized from the Industrial Revolution. And now AI is in everything and it’s so hot because it’s so new and so transformative.
Ryan Freng:
But I imagine within some amount of time maybe it’s longer than with these other technologies, but some amount of time we won’t be talking as much about the technology anymore. It won’t be about AI, it’ll just be yeah, my phone has a great assistant capability, or now there’s these just generative video tools and it’s less about AI, this AI, that AI and it’s just the underpinning of the technology. So I think that’ll happen and if we keep a perspective like that, then it’s a tool that can enhance the work that we do. And there’s a lot of unanswered questions of like VO artists. Or how will it help in X field? It’s unclear until somebody solves that problem.
John Shoemaker:
Yeah, so I have always enjoyed documentary and I love the fact that we do documentary because it has been the thing that the conclusion I’ve come to is that that will hold on for a long time because you can’t well, you can fake it, but it’s not real, it’s not authentic. Like if I want to know what another person thinks about something, there’s no other way to get that than documentary. So, like that’s, there is still that, that value again, you could fake it, but then it’s just not real. It’s like well, yeah, you faked it, you know well, and then what is?
Ryan Freng:
you know, what is the purpose of it? Is the purpose of it to have the visuals and the audio and just the combination of all those things affect someone in a certain way? Or is it for that person to see another story in that? Because, again, a lot of it probably could be simulated, but there’s something to the real people and real stories and I think that’ll continue to resonate through all of this. You know, we have like computer generated art and we’re like, oh, that’s really interesting and unique. But then you see something done by a human and you’re like, wow, yeah, you know, it can just blow your mind in a different way, I guess.
John Shoemaker:
What’s the most interesting part of Ready Player One?
Ryan Freng:
Steven Spielberg directing.
John Shoemaker:
I mean, and I didn’t come prepared for this idea, I just had this idea.
Ryan Freng:
I do love the book.
John Shoemaker:
It’s when you start peeling back the layers and understanding that there are people behind those avatars. It’s like discovering the real person. Sure so, connecting with the real people, connecting with reality, even though they’re living in non-reality or in the virtual space or whatever. The whole interest and intrigue is that who the? People are. That’s true.
Ryan Freng:
There’s something there and that’s interesting, in a narrative story, in a fictional story that someone made up, that other people pretended to be the characters in and pretended such that we could understand these concepts, or something.
John Shoemaker:
It’s like the most intriguing thing about like AI, I think for a lot of people I watch my son do it all the time is just asking questions, trying to figure out is it actually thinking or thinking things, or have some consciousness, or is it? Just a tool spitting out you know whatever like. You can use it as a tool, but the real like question is like what’s what’s there, what’s behind?
Ryan Freng:
it. I mean, the reality is, you should probably say please and thank you just again for when the robots take over I also think that’s.
John Shoemaker:
Have we talked about this before on podcasts? I don’t know.
Ryan Freng:
I don’t know. I don’t know what you’re. You haven’t said the thing.
John Shoemaker:
I think it’s valuable just for the state of the person’s own soul to say please and thank you. You know like it’s good for’re to not get into a habit of just like, like give me this now you know? Yeah, I was thinking about that, like you know in what realm is that ever okay?
Ryan Freng:
yeah, yeah, it’s just good practice. Yeah, even if even if it is not sentient, yeah. So what’s our takeaway? Ai is a tool. Use it, abuse it. Maybe don’t abuse it. Use it to save you time on tasks that are lower value to you. Or maybe you know, for me, like if I’m on phone calls, talking to people or reaching out to people and connecting and prospecting, that’s very high value to people. Or reaching out to people and connecting and prospecting, that’s very high value. If I’m searching for emails, or searching for contacts or creating, you know, generating an image for a mood board, or creating an image for a mood board versus generating one. You know, it’s all these things. Like I can drive and create the thing, but to get to that thing is a lot of time and a lot of work and I need help, and AI can certainly help in there. So, utilizing AI in as much as you can. Don’t use it for quotes, because it’s going to get a lot of them wrong.
John Shoemaker:
Yeah, well, and you mentioned this, I forget where you said you found it, but the AI, artificial intelligence, or AI army of interns, right, that concept? Yeah, that was that Madison Chamber.
Ryan Freng:
Yeah.
John Shoemaker:
Yeah, when, you know, if you had an army of interns working for you, they could generate a lot of great stuff, but you’re still the expert and you still need to, you know, check their work. You know like, Okay, is this good, Did you? You know, cut corners? Did you make something up here? You know?
Ryan Freng:
and, you know, eventually we could probably train an AI to do that as well. So just try to try to be on it, try to work in the AI, try to continue to use it to help you, help augment your work, I guess.
John Shoemaker:
Yeah, I had the sorry. This is not a takeaway, it’s one more point I mentioned that to somebody today was yeah, get your output from an AI and then, just for the sake of variety, give that to a different platform. Like, if you’re working in chat GBT, give that to Grock or Claude and then ask that to proof it for you, or just to take a look and see. You might not be 100%, but another one will probably catch something because looking at a different, you know oh, this quote doesn’t appear in the document.
John Shoemaker:
For example, Right, Something like that. It probably would work.
Ryan Freng:
I love it because then, yeah, it’s like different peers reviewing things or different people quality assuring each work each other’s work.
John Shoemaker:
Cool, but yeah. So my takeaway a better takeaway is that it’s changing the industry right now. It’s changing everything, and the way to win in the end is not to like run away from it, because it’s just, it’s, it’s everywhere, it’s taken over and you, we have to use it as long as we can and as long as it makes sense, and what our company looks like two, five and ten years from now, um, you know, we could be around because we have changed everything. We won’t be around if we just try to keep doing things the way that we used to. Mm-hmm.
Ryan Freng:
Yeah, embrace our AI overlords. Exactly All hail chat, gpt or Grok.
John Shoemaker:
Or Grok, or Claude, or we don’t know who’s going to.
Ryan Freng:
That’s the thing we get well, the chinese one apparently, is faster and better right deep seek. Yeah, like it. A 20th of the cost to get to be better than where we are, but we were.
John Shoemaker:
You know it’s also communist, also managed by communist well, it, it will tell you it itself is communist. They trained. I didn’t know that they trained it. But then if you ask certain things about, like what happened in tiananmen square, and stuff like that it will not give you the same answer you’ll get from uh, it’s great to live in america um gemini.
Ryan Freng:
Is that google, oh?
John Shoemaker:
yeah, yeah, forgot about that one gemini.
Ryan Freng:
I feel like they’re doing some good stuff, but claude is claude like based off of jet gpt no, it’s a different one it’s its own yeah, by who?
John Shoemaker:
claude sonnet?
Ryan Freng:
I don’t know because then amazon has rufus, which that’s the dumbest thing ever. Sorry, sorry to all the rufuses out there, but yeah, I don’t want rufus like rufus sounds like a moron the Rufuses out there. But I don’t want Rufus Like Rufus sounds like a moron who’s like you don’t want him to be your assistant.
John Shoemaker:
Yeah, what’s Apple doing with? Is Siri still alive or is she going to? Oh, so they have Apple intelligence.
Ryan Freng:
So, yeah, Apple intelligence, but I don’t have it on my phone because this is old. On the 16, I think that’s the latest one you can actually connect Siri to ChatGPT and get ChatGPT answers in Siri. Now it does have Apple intelligence, but it’s kind of a narrow focus in what it can do, Kind of assistant like oh, somebody said something about this, Want me to put that on your calendar, which is kind of helpful.
John Shoemaker:
Google’s still struggling with that too, it’s.
Ryan Freng:
It’s funny, I thought that would have been the first thing they need to quickly abandon.
John Shoemaker:
I think it’s like is there infrastructure built up in the old assistant? That’s like not fully ai sure and so they’re like well, we have all these tools, we can’t just throw them away. And I’m like you need to throw them away. Just Just put Gemini in Google Assistant. I don’t want any answers that are like it can’t understand what I’m saying. Give me the AI that understands what I’m saying and give me the result I’m looking for.
Ryan Freng:
And interconnect it all. So if I get an email and it’s like, hey, somebody wants to do this, okay, schedule it. You know Google has a little bit of that and Assistant has a little bit of that, but I just want it all integrated, instead of like let me hop over to this app and type it in here and then copy and paste, and then let me book this and then let me reach out. No, just show me what you want to do, do it so you still have approval, so it doesn’t do anything mistakenly.
John Shoemaker:
For now, we still have approval. All right, that’s what we got.
Ryan Freng:
Thanks for hanging out.
