How Can Brands Bridge the Gap Between Human Quality and AI Efficiency?

Katrine Rasmussen (00:02) We all also just need to take a step back and remember that AI is just a tool. It's a powerful tool, but it's really just a tool. If you have skilled creatives, right, people from within the industry, learning how to use the tool, that's how you're going to get quality work out of the tool. Just because you and I have access to NanoBanana and we think we can create something, we're not going to be able to create something great, right? It's just, we're all running around with a relatively decent camera in our back pockets, right? It doesn't make us fashion photographers. Ben Ard (00:59) Welcome back to another episode of Content Amplified. Today I'm joined by Katrina. Katrina, welcome to the show. Katrine Rasmussen (01:04) Thank you very much. I'm very excited to be back. Ben Ard (01:06) Katrina, this is awesome. I always love having repeat guests. We had such a good time last time. But for anyone who didn't get to listen to your episode, why don't you take a couple seconds and introduce yourself so everyone can kind of know who you are. Katrine Rasmussen (01:19) ⁓ sure. Absolutely. My name is Katrine, I'm CMO at Pixels. And if you've never heard about Pixels before, what we do, we take care of all the visuals on PDP, product detail pages for large e-commerce brands and retailers. So it's everything from images to video. to pretty much everything that you see of the visual content on e-commerce websites. So that's what we do. I've been with Pixels for a little over five years now. At Pixels, we work with AI for over 10 years now, but more from like in... you would say not generative AI, right? It's been very much like automations and good old AI if that's even a thing, right? But recently we've also gone into doing a lot of generative AI instead of just like retouching and editing and fixing visuals, we've also gone into like creating to be the actual creators of a lot of the visual content for a lot of our clients. So there's been very interesting times in the past couple of years, right? With all the new developments that we've seen in the tech space and the AI space. Ben Ard (02:32) I love it. That's so cool. Now I want to put a big old disclaimer on this. And what was so fun is before we started recording, Katrina made sure she's like, I don't want this episode to be an advertisement or overly salesy for our product. But we are going to kind of dive into one of your new releases because I think it brings up some really interesting subjects around AI, how to approach AI, maybe some of the philosophies. And I think you're doing it the right way. So Katrina, you just launched not too long ago a product called Digital Twins. And I think this is pretty awesome. Why don't you take a second and if you could explain for the audience, what are Digital Twins and how did you kind of approach that product? Katrine Rasmussen (03:11) Absolutely. But before I do that, let me just take a tiny step back. Because if you're not as heavily invested in this space as I am in all e-commerce, campaign, creative operations work, right? You may or may not have realized that a lot of the imagery that we've started seeing online, is actually a lot of that is AI generated. So it's everywhere, right, from socials, but it's also on a lot of e-commerce websites today. And I'm talking about campaign-style imagery, but even the PDP, the product pages, right, some of the models, the clothing, the garments that you see there, some of that might actually be AI generated. And maybe you're not even realizing it, right? Because it's that good and you might not be able to see it. So I think there's been, and I think what's interesting, right? I think a lot of you probably saw all the controversy around the J.Crew campaign. J.Crew, was it about a month ago? Maybe a little bit more than that. Well, they created a campaign It was all over socials and it was AI generated. But what was interesting was that they didn't disclaim that. They didn't go out and say, well, by the way, this is AI generated. of course, like internet is going to find out because, I mean, there was a little slop about that campaign. So internet found out and all of a sudden you've got like commenters everywhere saying, this is AI slop, AI, AI, AI. And know, guests did the same thing. There was like big spread in Vogue with an AI model. They did disclaim it, though, but still created a bit of an upheaval just because it was very Barbie plastic. was interesting. Anyway, so we're starting to see all of this AI-generated campaign imagery, PDP imagery, there. And we wanted to take a step up that on that and see how we could work with the brands that we work with in the fashion space, which is primarily where we work and help create and generate. And I want to say a bit more of an ethical. way of generating AI imagery. So that's where the digital twins come into play, right? Because if you're creating just random AI models, fashion models from scratch, because I mean, you can easily do that today, right? It's really actually not that hard. Anyone can do that. Dress them in... whatever you want to dress them in, right? And there you go, you have your campaign imagery. But I don't know, is that... I think a lot of these bigger brands, right? They don't love that approach for a number of different reasons. mean, is obviously there is the ethical aspect of it, right? Who are we cutting out of that process in terms of creatives, in terms of people? But there is also... ⁓ a little bit of a legal aspect to it, right? If we create ⁓ AI models out of scratch, where did that training data come from? Are we going to run into lawsuits? Is there going to be like a double-ganger coming up? Even if you didn't intentionally set out to recreate a person, you probably will recreate a person just because of the training data, right, and the amount of people out there. So there is a lot of... Weird things or at least things to think about before you go ahead and do that So we came up with the digital twins the digital twins product and you might have seen If you're in this space, you'll probably have seen H going out. I think probably around six months ago in the summer Announcing that they created 30 digital twins of their models and how those models are now getting compensated or the digital twins is going well The the digital twin is getting compensated for the work that they do, right? While maybe that model is doing another shoot somewhere else or at home on the couch, right? So that was an interesting approach. So we have basically created a similar product, I'd say. So we've set out to work very closely with a few different agency, model agencies. We had this idea. We went out and asked the agency, so would you be interested in creating a campaign, creating a digital twin, working with your models to create digital twins? And I don't know what I was expecting, but a lot of them came back to us and said, yes, why not? Let's try it. Let's do it. So that's what we set out to do. So now at this point, we have a model catalog with a number of models from a number of different agencies where we have created digital twins of the model. ⁓ So all these models now have a contract in place with their modeling agency, the talent agencies, works, that basically stipulates how they're getting compensated, what the approval process looks like. If any brand or any campaign wants to work with their digital twin, they can do so, just like they work for their regular modeling work. So that's how we've done it now. And we have fashion brands coming to us, working with their digital twins and creating campaigns from like 100 % digitized campaigns, which is interesting at this point, I think. Ben Ard (08:24) Now I think this is so cool. it brings up some interesting points because with artificial intelligence, everyone wants the efficiency gains that come from using the technology. But at the same time, you're right that there is this issue of where did the model get trained? How much control do we have over it? What's it, what's it going to look like long-term? Am I going to have to take this all down? Are people going to get upset? Things like that. Katrine Rasmussen (08:32) Yes. Yes. Ben Ard (08:48) So I love that you said, okay, great. Let's merge the efficiency of AI, but let's involve humans. Let's involve the models and their agencies. Let's make sure they're comfortable with this. So now the brands could take advantage of all the technology, all of the efficiencies through your system. Yet they know the models are getting compensated. They know the models have approved the material. They know that everyone involved is comfortable with things. I think this is where AI really is headed and needs to for brands. And I think there's something there where if you can just go one step further. Besides just the chat GPT prompts, but going one step further and say, let's think about this. Let's get people in the loop and make sure that they're comfortable with everything going on. Now, Katrina, I always love these stories. You got to see the behind the scenes of everything in development. How did that process start? How did that conversation go to say, Hey, rather than doing an AI, you know, just willy nilly, you can make everything you want up. How did it come to that point where you said, no, let's, let's get the models and the agencies involved so that our brands feel comfortable using the technology. How did that look like if I were a fly on the wall? Katrine Rasmussen (10:03) Yeah, I think there are actually probably two answers to that question, right? And one of them, I feel like you sort of already answered. I think right now with all this new AI out there, right? I mean, yeah, we're looking for efficiency gains. We're looking to create more with less. But I also think at this point, We all also just need to take a step back and remember that AI is just a tool. It's a powerful tool, but it's really just a tool. If you have skilled creatives, right, people from within the industry, learning how to use the tool, that's how you're going to get quality work out of the tool. Just because you and I have access to NanoBanana and we think we can create something, we're not going to be able to create something great, right? It's just, we're all running around with a relatively decent camera in our back pockets, right? It doesn't make us fashion photographers. It's very much the same thing. So when we set out to create the digital space, product. think what we realized, and we've seen a lot of this like AI slop out there, I think what we realized that AI is only as good as the people behind the AI, behind the tool. And I think that's the same thing with any tool, right? The camera is a tool, Photoshop is a tool. But in the wrong hands, we're still not going to create quality work. So... It might have been, AI might have been democratized because we all know how to prompt, right? But it doesn't mean that we can all create great content. So I think our realization was we need people from within the industry, people who know what they're doing to work with the AI. So it's for us, it's very much been teaching. people from within the industry, how to use the tool and working directly with a lot of industry people, right? Fashioned industry people who know what they're doing. And that's the approach, right? That's why what we've been producing is actually, it's not AI stuff, right? I would call it quality AI work. So that's everything from working directly with the models, having the models approve their digital twins, having the agency approve everything, right? But also to the AI artists ⁓ doing like creating, creating the digital twin, creating the digital version of the garment, prompting, combining everything and making sure that we have industry people involved all the way because otherwise the end product is not going to be great. So it's Having the approach of AI is just a tool. We need to put that tool in the hands of people who know what they're doing and not the other way around. We're not going to hire prompt engineers and try to train them to do fashion campaigns because that's never going to work. So I want to say that's part of it, right? Realizing that and that the people behind the AI is more important than what exactly what tech we're using, what tools we're using. It's not the important part, it's the people behind it that's important. And then the second thing, I think, that's going back to the brands that we work with and going back to looking at what's useful to them and what they're after. And I think definitely like the legal aspect, making sure that they're not putting themselves in risk with using random AI generated models. That's a big one for a lot of the fashion brands. At least the bigger brands out there, right? They need to make sure that whatever they're doing is legal, basically. And also that it's good quality work that comes out of it and not just a nano banana prompt. So I think that is probably the best answer. Also, I do think we have a responsibility, even though we all want efficiency, we all want to cut costs, to create more with less. I also think we all do have a responsibility to make sure that we are implementing AI in an ethically sound way. And I think we do that when we're working with people from within the industry. So I think that is the approach that we really set out to take. Ben Ard (14:14) I love that. I think that's so amazing. And I love the underlying principle of AI is just a tool and it's okay to treat it like that. But I also love that you're focused on the right people who know what the outcome, what quality looks like because they've been in the industry. They have the expertise. They're the ones working with the AI to get the results. Cause it's true. If you put me in with a photo editor and I don't know what I'm doing, Doesn't matter how good the tool is. I'm not going to be able to produce a quality and result. The same thing with artificial intelligence. If I'm in the fashion industry and I know what quality looks like, and I'm given that tool, I actually can do something really powerful with it. But if I just have chat GPT and I'm not a lawyer and I'm trying to get legal documents, that's basically a recipe for disaster. And. Katrine Rasmussen (15:03) Absolutely, and the same goes for think pretty much any industry. Ben Ard (15:07) 100%. Absolutely. If you don't have the experts running the tools, you're going to get AI slop. And I think that's where the term comes from. The slop is from people who don't know what quality looks like in that space because they just don't. And they're trying to shortcut through AI when really you could build efficiencies if you knew the right outcome. So I love that you're including humans in the loop. You're focused on expertise. All of that is so amazing. I'm so excited for technology, like the digital twin. I think it's the right way of doing AI for anyone listening that has a product. Consider some of these thoughts in that process, or if you're on the marketing team, just like Katrina said, think about the ethical way. love how you said that the ethical way of using AI use people use experts, treat it like a powerful tool, but do not. forget what quality looks like and make sure that's always your standard. I love that. Katrina, thank you for taking the time to share your expertise, to share all the knowledge and all the cool updates with this product. For anyone who would like to reach out and connect with you online, how and where can they find you? Katrine Rasmussen (16:13) LinkedIn. LinkedIn is definitely, I mean, I post a lot about AI, especially within the fashion industry. AI Slop, AI Camp, whenever I see a new fashion brand push out AI campaigns, I'll be the first one to post about them on LinkedIn. And it creates a lot of very, very interesting conversations over there. So please join me over there. Ben Ard (16:34) love it. Perfect. For anyone listening, will link to Katrina's LinkedIn profile in the show notes. So scroll down and connect with her there. Katrina, thank you so much for your time, insights, and everything else today. I really do appreciate it. Katrine Rasmussen (16:45) Thank you so much for having me.