Latest / The Interline Podcast / An AI Likeness Free-For-All, Social Impact, Cataloguing, Sorting, And Persuasion
Transcript
- speaker-0: Hello, welcome to the edit from The Interline, the show where we run a quick fire analysis on our pick of the most important fashion and beauty technology stories from the last seven days. I'm Grace, the social editor, and I'm joined by Ben, the editor-in-chief. And together we have less than 25 minutes to give you our analysis on the stories that we think really matter. speaker-1: Hey Grace, good to see you again. It's hot again. This is the second the second heat wave recording I think we've done. So we are under time pressure and environmental pressure as well this time. speaker-0: Yep, still powering through, but today we do still have a lot of stories so we can get into. speaker-1: Go ahead. Fire away. Hit me. speaker-0: So as always, AI was really dominating the headlines this week. And they were kind of polarizing with all the stories that we found. But this is going to be a very AI heavy episode. And the first story that I want to talk about actually relates to a couple of other stories that we've spoken about on the edit before. So just to remind everyone of these stories. One of them was the story about Francesca Pujot's, so the model who sued the retailer about using her likeness to generate new AI images without her consent. The other story we've spoken about was about the outdoor retailer REI and the fact that they fell foul to Meta's new AI ad program that they didn't opt out of. This week it's another very similar story. So an Australian Nigerian model has basically sued the retailer Peter Jackson, saying that they used AI to alter his face, so change his skin, change his eyes, and also change his facial structure. And he's like accusing them of racial discrimination with AI. Similarly and equally as concerning, it's also been announced this week that Meta has revealed its new Muse AI image generation model. So this is going to allow users to generate images using the likeness of other Instagram users who don't actively opt out of this program. To me, these stories just highlight that there's so many questions and unresolved things around how people use AI with commercial imagery. And also it highlights how it's still such a grey area. So I just wanted to get your take on what's going on with all of this. speaker-1: Yeah, I'll do them together then in reverse order, I think. So we are firmly in the ask forgiveness, not permission stage of the AI rollout. So you mentioned the REI example from last week. You mentioned the Franchise Superior Jource ones. Those are both instances of companies of very different scales. So a fairly small to mid-market retailer up to Meta, one of the world's biggest tech companies, effectively saying, well, AI is a It's a license for us to basically do what we want. The frontier is untested, the legal precedents are not set. So let's let's take our chances. Let's let's play around with these things. Now, Meta in particular seem committed to just running down the evil genius playbook, basically, at the minute. I think if your brand is as what's whatever the opposite of beloved is. If your brand is as kind of out of favor as the meta brand is, part of me understands the the that being an opportunity to say, let's just try and get away with as many things as we can because people don't like us anyway. Now, there's no world where it's considered to be socially acceptable and a good idea to allow people to generate images based on publicly available Instagram profiles, right? Because ⁓ if I understand it correctly, Don't use social media in my personal life, but Instagram profiles are public by default. ⁓ influencers in particular thrive on putting as much content about themselves and featuring themselves out there as they possibly can. You have that interest then clashing with Meta's interest, which is saying, Well, we have all of these images on our platform. Why don't we use those to allow people to include them at inference for this new ⁓ Muse AI model, by which I mean include them at the time that you actually use it. Presumably they're also included in the training data ⁓ for from Muse as well, because Muse is is actually seemingly a reasonably competitive image generation model, but the bit that people are up in arms about is the permissionless side of things. Now, just like the REI story from last week, which was ⁓ based on Meta's programmatic advertising stack, this is opt-out. So, if you are an influencer and you don't want people to be able to generate images using your likeness, you need to go manually set a flag to ⁓ to do that. The model story here is actually a little bit more complicated because it falls into the same zone of you have a retailer who is taking pre-existing material that they have shot, presumably. They've they've done a campaign with this guy before, they've done ⁓ a PDP photo shoot with this guy before. And they're treating that as just fair game for intake. So ⁓ at the point of input, they're presumably going, we've got some new products, we need to ⁓ generate some new images. Let's use the guy that we've used before. However, let's use him as a basis for iteration ⁓ in this case, and let's start to adjust his features, his skin tone, ⁓ his height, all these kinds of things. Now, there's a sliding scale beyond which that guy would not necessarily know. That he was used as the input, right? I'm I'm assuming here that there are some giveaways that reveal the fact that he was the input and then that the outputs still superficially resemble him, but with ⁓ with some transformations. You could do the same thing, or you could transform some that you could transform somebody in a deep way, you could change something more profound about them, and they would probably never know. The big issue that this raises for brands and retailers is you have to have full chain of accountability and custody and visibility into how you generated your images. So Assuming you're using like a node-based workspace, you need to know at every step of that journey, you need to be able to reverse engineer all of it to say what you actually did or didn't do in this case. Because ⁓ on the merits of this, it strays into some really uncomfortable areas around racism and ⁓ why people would want to adjust somebody's skin tone to appeal to or appease a particular demographic, which is not great. ⁓ but it also gets to if you're gonna use things as intake. for this, if you're gonna say y you need to either sanitize your inputs and know that they are above board, or you need full accountability so that you can open your doors in the in these cases and say, look, I what this guy's claiming isn't true, or in this case what this guy is claiming is potentially true. speaker-0: So, like I said, AI really dominated the headlines this week. And the next story is staying on the same vibe of backlash against AI. There was a really interesting piece in The Guardian that was talking about how the government has chosen Larnarkshire in Scotland as a key AI growth zone. So the plan is to build large AI data centres. And the plan with these data centres is that they're meant to help. All of the AI models as AI becomes a bigger part of our economy. Unsurprisingly, the local communities were not happy about this. But to me, it was just kind of a stark reality check that AI really is having a physical impact on communities and also the environment. So I wanted to get your take as fashion adopts AI more and more. Do you think fashion is gonna contribute to this problem? And also, how does fashion just generally fit into this issue? speaker-1: So I think ⁓ people seem very confused, some people do, about where the backlash for AI is coming from. I've had conversations as recently as this week. I was a guest on another podcast which is coming out at some indeterminate point in the future. And in that, we the host was kind of saying, well, why do people hate AI so much? What is it that makes it fundamentally different to 3D or makes it fundamentally different to the web or any other kind of technology revolution? There's two bits to it. One is the permissionless training piece that we just talked about. And the free-for-all ⁓ image ingestion and text suggestion, everything else at the point of inference. There's nothing to stop your creative work becoming part of a training data set unless you simply don't publish it on the internet. And so that's one of the reasons that people object. The other is that it is having this like tangible physical effect on communities. Now, data centers are not Solely the preserve of AI. No, AI is behind a lot of the data center build out because data center capacity and usage has been fairly static for a decade or so prior to this. ⁓ fashion is also not responsible for all of that. I would argue fashion is probably no different from any other industry. Maybe at like an image generation quantity and frequency, maybe it's higher than other sectors, but it's going to be negligible difference, I think. It's a whole society problem. Now, the issue with these is data centers need a lot of space. They need a lot of power. And they under the kind of traditional cooling method, there are like closed loop ⁓ cooling systems that you can do that are not full of water, but most of them use open loop ⁓ like water cooling. Basically, you're taking a massive thing, putting it close to where people live, and that thing consumes a lot of power, consumes a lot of resources, can potentially raise utility rates and things in those areas, is a nice ore, is apparently loud in a lot of instances. But That's almost the the the easy part and that's almost that not the worst part. You have data centric build-out companies and energy companies promising that these things will be beneficial. So they will say, Well, this will we are gonna amortize and absorb any raise in utility rates, so you won't feel that. It's going to create jobs. ⁓ it's gonna s help to charge the local economy and everything else. What ends up happening is there are ⁓ there's very little job creation. ⁓ there's a whole lot of mess and noise and ⁓ and utility use and everything else. This is part of a pattern that makes me sound like a like a crusty old, like left-leaning radical, which I'm not fully, which is that a lot of this sort of objection to AI comes from this idea that the public society must bear the cost, and private enterprise is what reaps the reward. You can frame this in like pure anarchist terms as like. Socialized the losses and privatise the gains. and this has that sort of sentiment to it. As I said, fashion has hooks into this. It's not solely responsible for it. But when people talk about the environmental impact and the social impact of AI, these specific kind of community issues are just as important as I generate an image and it uses X amount of water or uses X amount of electricity. speaker-0: The next story is actually a bit more positive, you'll be happy to hear. So this week there's been a lot of coverage about wearing and the fact that it secured seven million dollars to really ramp up its AI capabilities. And this is from eBay and also Google's AI Future Fund. So for those of you who don't know, Wearing is basically a digital wardrobe app. And it helps people digitize their wardrobe, organize different outfits, and really see what they have in their wardrobe. With the funding, they're gonna use AI to create these personalized outfit recommendations based on the person's mood, where they're going, and also even the weather. They're also gonna be able to create more like visually rich outputs as well. So it's all just leveling up. I know that there's been a lot of digital wardrobe apps in the past, but now that wearing is incorporating AI. it feels like it might actually reach mass scale. So I wanted to get your take on what you think about this and what you think the future holds for digital wardrobe apps like this. speaker-1: Yeah, so this is one of those categories that I recognize is growing and compelling for a lot of people, but I don't have a personal use case for myself just because I dunno try not to have that many clothes. but no doubt there are There are some suits in my wardrobe that I have not worn in ten years that probably shouldn't be there. But aside from that, like there is this this is limited kind of first hand use for me. I did want to ask you actually, is this something that you would use? Is this something you would consider using? Do you use it? Yeah. speaker-0: I think in the past I wouldn't have used apps like this, but now that the quality and user experience has improved so much, I definitely would use it. And I think wearing seems to be the best example of it. So I would be open to it for sure. speaker-1: Yeah. And it it kind of dovetails with so Google, interestingly, was the company that did that whole like, okay, we're gonna build the clueless closet thing automatically from your Google photos. So they they went through that whole process. Really smart tech. ⁓ and again, a a use case that's not for me, but presumably a lot of people ⁓ enjoyed that. Now, these are the kinds of apps that feel like they are B2C first. Right. ⁓ and save your wardrobe is another good example of this. They feel like they are consumer plays because they are in the first instance. What you're trying to do is you're trying to get people to download this and create behaviors ⁓ before you can then build out the B2B ⁓ and kind of brand facing and retailer-facing side of these things. Now, I think AI will definitely help with the behavioral side of stuff because the biggest barrier to somebody digitizing their wardrobe. It's gonna be the effort involved in pulling every piece out, taking meticulous photographs from every angle, labeling everything, looking at the ⁓ c wash care labels, looking at the tags to see if they can figure out the material composition, or all of that kind of stuff, guessing at what their retail price was, if this is ⁓ like a resale play as well. So anything that you can do to reduce that friction is what helps to create the user behavior. Now, I don't know for sure that wearing has a ⁓ B2B play to come in this. Presumably they do. And the way that I would think of it would be this. People talk about AI as a when when AI kind of mediates between consumer and brand, when it's part of this sort of agentic like product discovery and shopping journey, they talk about it passing highest in t the highest intent and the most complete kind of customer over to a retailer. So rather than somebody just rocking up and saying, I'm looking for X, help me. In this case, they're saying, well, I've already been through multiple rounds of queries, and not only am I looking for X, here is a very finite definition of what X constitutes. Add this to the mix, and you can also say, well, actually, this pairs well with what you also already have in your closet. Or you can say, Well, there's actually an opportunity to upcycle something here, there's an opportunity to replace something here. If you tie this into take back programs, if you tie this into AI commerce, all of a sudden it becomes much more interesting. It's the same philosophy behind how I feel about RFID, which is if you have item level visibility, so you have a tag that tells you where a piece of clothing is going in the supply chain, in the store, in the fitting room, and so on. Not that exciting on its own, but you can build some really cool stuff out of it. If what you start to get here is effectively the same thing, like item level visibility for what is in people's closets, what they have, what they're looking for, then you can start to architect some really cool stuff on top of it. So I agree, a more positive application of AI all round. speaker-0: So we've just spoken about AI helping consumers organise their wardrobes. But there was also some really interesting stories this week about AI helping organise secondhand fashion. So there's a London-based startup called Fleek, and they've just secured an additional twenty-five million dollars worth of funding. And they're gonna use this funding to digitize the infrastructure of second hand sorting. So Fleek explains that there's so many garments that go from donation bins to sorting centres and the process of grading Pricing and organizing these garments is still very offline and very analog and very manual. So they're gonna use AI to completely revolutionize this. At the same time, we also saw a story from the Berlin startup reverse fashion. And they've also secured funding to develop an AI-powered textile sorting technology. So both of these companies are trying to replace labor-intensive subjecting sorting processes with AI. And this could potentially make the secondhand industry more sustainable and more efficient. So I wanted to get your take on this. Is it a big leap forward or just what do you think about it? speaker-1: So I think this is the kind of thing that can scale depending on what the hardware requirements are for this sort of thing. Because it's very easy to say you have AI that does this. It's harder to know off the top of your head where that AI lives. So the textile sorting one, for example, feels like it's part of the same class of technology as ⁓ like fabric quality inspection AI. Which is something that's existed for a while using machine learning. There are now other companies that do this for high value fabrics. So if you think about like high quality leather, for instance, you want to be able to identify any defects in the hide before you cut it so that you can cut around it. And then that increases the value of the final garment and it reduces your wastage. Now that AI has to be tied to sensor hardware. ⁓ it's not just you're not just pointing your phone at a piece of leather. You're actually having to put it through a scanner to to get those kinds of results. So how scalable, how portable, how kind of durable this sort of setup is is really dependent on how much hardware infrastructure you need to have in place for it. Now, I think provided you can do that, ⁓ and but most of that hardware is not especially expensive, I think the AI sorting of textiles is fascinating because it does very quickly get you to much more efficient recycling, much more efficient upcycling, much more efficient repurposing and ⁓ and so on. And it's not an untested or unproven avenue for AI. Like machine learning AI have already been deployed in this space for a long time. So I'm optimistic about that side of things. The sorting of secondhand goods, similar deal. So we've just talked about how the importance of consumers digitizing their own wardrobes, creating that behavioral pattern to then enable brands and retailers to take that into first hand sale. Now if you think about the same thing and you think, okay, I have an account with vintage, I have an account with somewhere else, like whatever whatever platform it is that I use to ⁓ to to sell garments when they're done, if that platform can start to be more proactive ⁓ about these kinds of things, and then you can centralize it as well. So if you can say, give me a bag of all the stuff you don't need, you don't have to sort it, you don't have to separate it into monomaterials and blends. You don't have to separate it into tops and bottoms. You don't have to determine what is worth recycling and what is not. Send it to us and we will do all of that automatically. And we will centralize all of that process. And then presumably we will build partnerships that allow us to then shift those things afterwards. Optimistic about both. I think I think both of those have got some good runway in front them. speaker-0: Now to wrap up this week's news, we're ending with yet another AI story. So this week OpenAI has just announced that it's going to update Upgrade its voice mode. So the new GPT Live experience is built on a full duplex architecture. And this means that it can basically listen and speak to keep the conversation going. And I'm pretty sure OpenAI said that it was really meant to mimic a natural human conversation. It also introduces features like agentic capabilities so the mode can delegate different tasks to other AI models while still keeping the conversation going. To me, this feels equally parts, exciting but also concerning. But there is no doubt that this is a massive leap forward. So I wanted to get your take on what this means for fat. speaker-1: Yeah, and I think it my take on this one is really quick and simple. So it it's a big leap forward in voice interaction. OpenAI have always had the best kind of voice mode because if you if you've ever tried it in Claude or Equivalent, it works on like a turn by turn, speech to text and text-to-speech basis. And it introduces a bunch of lag, it introduces a bunch of inaccuracy. OpenAI have always had a really kind of engaging sort of back and forth ⁓ conversational dynamic. And swapping it to the duplex architecture is it just it just supercharges that. ⁓ it's probably gonna be very similar to the e sliding scales of expressivity and stuff that you can build into the upcoming ⁓ Siri AI, which is due out later this year, along with ⁓ iOS twenty-seven, I assume it's called. So that's that's fine. So the the voice mode has been upgraded. The the big part is that as as compelling as the voice mode used to be to interact with. It was always paired in the background with one of their less capable and older LLMs. ⁓ so I think it was GPT four. ⁓ for a while. I'm sure it got upgraded at some point. But if you've seen any of those viral kind of videos of ⁓ that guy who talks to ChatGPT and he's like, set a timer for me, I'm gonna do a run a mile, and then he comes back a second later. and it has it has no concept of the passage of time, it has no it it It's just it's it's a very interesting human seeming interface paired with a pretty dumb ⁓ LLM on the back end. So they what they seem to have done is to upgrade this now to where it's connected to the like GPT 5.5 series, I think, or whatever the 5.6, the the the planet, Sol, Terra, and Luna stuff. the interesting part for fashion is that then sh in at least in theory, you should upgrade it to where it's connected to a lot of the ⁓ brand discovery, product discovery and agentix shopping stuff. So if you look at some of the video demos, they show ⁓ they show kind of weather search, World Cup scores, things like that. But it can present you with visual information. So I don't think there's any impediment to it kind of companies like ASOS and companies like that that have done brand catalogs and immersion. Is there any impediment to having that show up in voice interactions? That is a bit Of a dual-edged sword, though, because from a brand point of view, yes, it puts some cool new surfaces out there. It also really intensifies the question of as long as AI models and applications continue to be optimized for engagement, as long as they continue to be optimized for can I do this for you? Can I help this with can I do this with you? They the more of that there is, the more chance there is of. AI starting to actually railroad people into making purchases, starting to act in a salesman kind of role. And the most that compelling persuasion tools that human beings have at their disposal are ⁓ at their disposal, sorry, are ⁓ your voice and ⁓ your ability to kind of ⁓ generate pictures and things as well, ⁓ to to visually represent things. Adding a more compelling voice to this mix, I think makes this much more. dangerous potentially. And to bring us full loop to what we talked about at the beginning, again, it feels like a ask forgiveness later rather than ask permission situation. I think we'll see some stuff emerge from this. Nothing quite as pronounced as the light AI psychosis side of things, but I feel like there is a looming study and I think we might have to do it if nobody else does it. There's a looming study about persuasion and prescriptive selling from AI ⁓ still to be done. Let's see how that pans out. speaker-0: Amazing, that's all the news. Great. speaker-1: Grace, good start to you as always. Let's hope it's a little less hot l next time we do it. Although for ⁓ the people who watch this on YouTube, the next one of these, I will have a different background because I'm gonna be in Spain for a couple of weeks. ⁓ so it's gonna be hot there. Let's ⁓ let's let's see if we manage to ⁓ get it under twenty five minutes with that additional heat as well. Grace, always a pleasure.