The Vergecast - The AI-generated, oddly colored future of art

Episode Date: November 16, 2022

Today on the flagship podcast of the difference between CMYK and RGB colors: 02:19 - David talks about the future of Photoshop with Adobe's Chief Product Officer Scott Belsky. 13:37 - Verge senior reporter James Vincent joins the show to discuss generative AI art and all its possibilities and complications. 43:05 - The Verge's Kristen Radtke and Jess Weatherbed chat with David about Pantone's new subscription service and what it means for artists and designers. Email us at [e-mail] or call us at 866-VERGE11, we'd love to hear from you. We are conducting a short audience survey to help plan for our future and hear from you. To participate, head to vox.com/podsurvey, and thank you! Learn more about your ad choices. Visit podcastchoices.com/adchoices

Welcome to the Vergecast, the flagship podcast of the difference between CMIK and RGB colors. I'm your friend David Pierce, and I am at Ace Hardware shopping for paint, which is both very fitting for the episode we're about to do, and also just a life necessity, because apparently the fact that I'm having a baby soon means I am required to repaint every single surface in my house. So I'm here trying to figure out the difference between Maui mist and moonlit snow and Austrian ice, which are apparently all different colors, despite. by looking absolutely exactly the same to me. Anyway, we have a super fun show coming up for you today, all about the future of art. We're going to talk about Photoshop and what it means to make a 30-year-old piece of software work for the next 30 years. We're also going to talk about AI art, things like Dolly and Mid Journey, and all the

possibilities and complications that those bring up. And then we're going to talk about a strange beef in the world of colors and how it came to be that Pantone is the authority on what colors actually are. It's a crazy world out there. All that is coming up right after the break. I need to actually like pick a paint color and buy something. I'm leaning toward Arctic Dawn personally, but also I have literally no idea what I'm talking about.

Here goes nothing. This is the Vergecast. See in a sec. Support for the show comes from Retool. Too many companies run critical operations on duct taped spreadsheets, Slack workflows, and whatever else they could cobble together. Not because they want to, but because building internal tools

means weeks of waiting on someone else's backlog. That's where Retool comes in. Build custom internal tools just by describing what you need. Prompt something like, build me a revenue dashboard on our Salesforce data. And Retool actually builds it on your company's data in your cloud with enterprise security built in. Go to Retool.com slash Verchcast. We all need to retool how we build software. What's up, y'all.

I'm Skyler Diggins, seven-time WMBA All-Star, Olympic gold medalist, and mom. And I'm Cassidy Hubbard, host and reporter for me. nearly 20 years covering the biggest names and stories in sports and mom. And this is Am Mom, a community for athletes, game changers, and moms of all kinds. Dropping May 14th. Tap in with us. Welcome back. As I said a minute ago, we're going to spend a lot of this episode talking about digital art.

Because in a world where everyone's a creator, most of our life happens on screens, and we're all buying and selling JPEGs for millions of dollars, it turns out art is pretty important. And you really, really, really cannot talk about digital art without talking about Adobe's Photoshop. Photoshop is to art what Google is to search or what Kleenex is to facial tissues, what, I don't know, Mario is to games where you jump on top of things that look like turtles. I'm getting away from the point here. The point is, Photoshop is a mission-critical, essential tool for artists around the world. And it has been for decades. You know what's wild, by the way?

I went back and looked at some fun videos of Photoshop 1.0, which was released in 1990, 32 years ago, and it's kind of incredible how much has stayed the same. There was even this really fun video I found from a couple of years ago, where a graphic designer named Will Patterson tried to use Photoshop 1.0. This is weird playing with Photoshop on here. Right, let's just get on with the design anyway. And he made it work.

There we go. We got our first design in Photoshop. shot of one. It doesn't look great. It's legitimately impressive, and it's good evidence that Adobe got a lot of things right, even really early on. But in a way, it's also kind of a bad sign. Technology changes. Users change. The things that we want and can do with technology change. And apps that don't change along with all of that tend to turn out like Microsoft Word or Lotus Notes, and they eventually just kind of go by the wayside as a new generation of tools emerges. It's this eternal struggle for a lot of successful products. How do you make an app super

successful and then change it to appeal to new users without irritating the existing ones? If you were to start Photoshop from scratch right now, it probably wouldn't have a thousand menus and all those icons that haven't changed in more than three decades. But if you made those changes now, you'd piss off millions of people. Scott Belski is the person charged with threading this very careful needle. Kelsky, chief product officer overseen creative products at Adobe. Scott told me he knows very acutely how important it is to not screw up Photoshop, which is good. You don't want to screw up Photoshop.

And the way Adobe has avoided doing it so far is pretty simple. It just does what customers want, which is mostly, you know, don't touch anything. People want more features, but they also really, really, really don't want to have to change the way that they use a tool for no good reason. Photoshop has one of the greatest product market fits ever. It is a name brand that is also in some ways a verb. And when you talk to a customer who loves Photoshop, what do they say first? Don't change anything.

Like don't screw with it. Like don't redesign the interface or whatever. Now, of course, like as a modern product builder, I look at the interface and the way that it's been built, you know, there's 30 years of decisions that have had to accumulate. And, you know, it's almost like the difference between going to a European city that has been built over hundreds and thousands of years of architecture. And then you go to like a modern American city that was built from a desert or something. It's just completely different. The more I think about that city's analogy, the more I like it.

Because like those old cities are amazing, right? Nobody thinks about it and goes, ah, Venice versus Phoenix. Give me Phoenix all day. But if your goal is just to get where you're going, I suspect. a more modern city helps. There are no obvious answers here, especially as people keep wanting more and more of these new modern things.

But that's where it gets tricky, right? You give people everything they want by just sort of shoving new features into Photoshop. You eventually build an app that is big and powerful and borderline unusable. It's like if you had a city that was both Phoenix and Venice, I'm not sure you'd have a super good time there.

And I think a lot of designers would argue that Photoshop has been teetering on that line for a long time. One way Adobe has tried to combat this is by splitting off apps like Lightroom, when it became obvious that a huge number of people really needed a way to manage their photos, but doing that in the confines of Photoshop along with everything else Photoshop can do didn't really make sense. So Lightroom became a thing.

More recently, Scott told me the rise of social media creators has gone kind of the same way. An observation we've made over the last 10 years or so in Photoshop is that there were so many customers coming in, creating things for social media. And the subset of features that they were asking for were easier interfaces, ways of making for mixed media, because most platforms, every social platform these days is mixed media. You come in thinking you're making a graphic and then you realize you want to add some animation and some text and maybe a little motion and then you put a timeline, you throw in some music. And before you know it, it's like a fully featured mixed media creation. So we realize that Photoshop is not the right tool for that.

And that actually was one of the reasons why we embarked on Express. And the previous incarnations of Express, because we recognized that we needed to build something specifically for the social media creator that probably had a lower willingness to pay, had a lower tolerance for a learning curve, and wanted mixed media creation. And so that's a story repeating itself all over again. Ultimately, the business model here is pretty clear, right? What Adobe really wants to do is introduce you to its suite through tools like Express and Figma, and then slowly upsell you all the way to Photoshop and the whole creative suite.

But one thing Scott said to me a few times was that Adobe has to let Photoshop be Photoshop. But even that means something different than it used to. Adobe is now bringing Photoshop to the web. It's embedding some of Photoshop's functionality inside of other apps, and it's thinking about what collaboration looks like. Again, technology changes, users change, and Adobe has to figure out how to keep up while letting Photoshop be Photoshop. Thankfully, it doesn't sound like Scott wants to do the whole let's all edit a photo together thing, which honestly sounds like my nightmare.

Yeah, I've seen very little evidence of customers saying I want to have three people editing my image at the same time. So I don't think that that's where we're going to take Photoshop. What we want to do with Photoshop is we want it to be easy to work with every stakeholder. You know, every single creative of our products, unless you're an artist working solely for yourself and you're never going to show your creations. Everyone has stakeholders, copy editors, clients, marketers, you know, you name it. And currently the workflows to engage with those stakeholders are extraordinarily antiquated. That basically means making it easier to show stuff, easier to share stuff, easier to leave comments on stuff. Which, fair enough, I can get behind that.

Anything that means fewer email attachments is a huge win in my book. To me, the biggest question for the future of Photoshop comes from AI. advances in AI have made all kinds of image editing the stuff people used to do professionally in Photoshop suddenly easy, even automatic. Think of the magic eraser on the pixel phones. All you do is outline what you don't want in the picture and poof, it's gone. Or the iPhone's action mode, which automatically tries to steady your video even as you run and jump and move around.

Normally, these shots would only be possible with a ton of stabilizing equipment, but with action mode, we didn't need any of that. Photoshop has introduced a bunch of automatic filters recently. It's trying to help people make their stuff better without requiring a huge amount of repetitive work. And if you've ever spent what feels like a thousand hours carefully masking around a part of your image just so you can edit it, it's definitely good news that you can now just click on the tennis ball you're trying to remove and poof, it's gone. You can immaculately apply color to a black and light photograph or you can just touch the button and it makes a color. I mean, why would you not do that and then make final edits and changes?

Adobe is also starting to really lean into generative AI, the tools that you can use to not just edit your photos and art, but actually create stuff from scratch. What if the future of Photoshop is just like a text box? You type what you want, the AI creates it, and then you apply the edits you want it. The whole text box thing is probably a bit much, but that idea, I think, is actually where Scott and Adobe are headed. I think that's right. And I think if you look at our tools into the future, they will all be AI-assisted tools. But at the end of the day, the creative control resides with the person. And the AI is really their co-pilot.

It's working for them layer by layer by layer or suggestion by suggestion. And the creatives will either take it or leave it. It will make them feel more productive. This sounds right to me in a lot of ways. And also sounds messy and complicated in a thousand ways we haven't even gotten into yet. Generative AI art. Generative AI in general is going to change things in bigger ways than I think we even realize. So actually, let's leave Scott and Photoshop and Adobe for now, and let's get into that.

We're going to take a quick break, and then we're going to dive into the weeds of all things generative AI. We'll be right back. Support for this show comes from Shopify. Starting something new isn't just hard. It can be really scary, too. So much work goes into this thing that you're not entirely sure will even work. But here's a better thought. What if it did all work?

What if your instincts were actually right all along? Shopify wants to help you get there. They're the commerce platform behind millions of businesses worldwide and nearly 10% of all e-commerce in the U.S. From established brands like Allbirds and Hines to companies just getting started. Their design tools make it simple to create the exact online presence you're envisioning with hundreds of ready-to-use templates available.

And with built-in marketing tools, you can launch full email and social campaigns in just a few clicks. So you can connect with customers wherever they are. It's time to turn those what-ifs into with Shopify today. You can sign up for your $1 per month trial today at Shopify.com slash vergecast. You can go to shopify.com slash vergecast. That's Shopify.com slash vergecast. Support for the show comes from LinkedIn.

If you're a small business owner, you know, that every hire counts, but time and resources are limited. Finding, connecting with, and screening the right candidates takes up valuable time you could be giving to your customers. That's where LinkedIn Hiring Pro comes in. It's built to be your hiring partner, helping you find the right candidates faster. That way you can hire with confidence without turning it into another full-time job. Hiring Pro streamlines the entire process from drafting your job to shortlisting candidates, and conducting AI-powered interviews for initial screenings. Its updated conversational interface lets you describe what you need in plain language.

Nearly 60% of hirers find a candidate to interview within a week. With Hiring Pro, you spend less time searching and more time connecting with the right talent. And instead of getting buried in resumes, you get a focus shortlist that actually moves your hiring forward. Join the 2.7 million small businesses using LinkedIn. to hire. Get started by posting your job for free at LinkedIn.com slash track. Terms and conditions apply. Welcome back. Like we were just talking about, generative AI is kind of having a moment. Everyone's out there using Dolly to make weird pictures of like the moon made of spaghetti

painted in the style of Andy Warhol or whatever. And stable diffusion and mid-journey and a bunch of other companies and organizations are building genera. narrative tools of their own. It's weird and it's messy, but it's really interesting stuff happening out there. So is the future of art just writing words in an AI prompt? And if so, how does that change the way that we think about art and artists? This is all fascinating, and it gets really complicated really fast. So I asked James Vincent, the Virges resident AI expert, to come wade through all of it with me. Hey James. Hi, David. How you doing? I'm good. I have been excited to talk to you about AI art for a long time because this is to me like the most interesting

and deeply bizarre space in tech right now. I have no idea what to make of it. So you're, you're going to make all of this make sense for me. So thank you in advance for doing so. I'll do my best. The first thing is like, let's just like lay the land a little bit here. Everybody talks about Open AI and Dali, but this is like the sort of generative AI space turns out to be like much bigger than I realized. Scope this industry a little bit for us right now. Yeah, I mean, generative AI is sort of this term that's become buzzy very recently, and we've sort of recond what were a lot of older systems, which were doing this sort of thing. But generative AI technically includes stuff like GPT3, so generating text, stuff like GitHub co-pilot, generating code, a lot of experimental music models and art models. But the art models are definitely the most visible, and they're the ones that are, they've become touchstones for a lot of the wider issues in the industry, which I'm sure we will talk about stuff like copyrights, stuff like safety, bias, etc., etc. For the generative stuff, it looks like these models have come out of nowhere,

but there has been a lot of work going on in this area for many years now. You know, the latest sort of generation, if we're looking in a multi-decade span of AI art models, probably goes back to 2014 or something, where you have a type of method known as GANS or generative adversarial networks, which become the big new thing. And now in the years since then, we've had a lot of different methods pop up that have been sort of tried and tested. And you might remember some of them. Do you remember Google Deep Dream?

Oh, yeah. And that was all the way back in 2015. But that was, you know, one of these early forays where researchers went down, what turned out to be basically a sort of a bit of a cul-de-sac. But you have to explore those things. And then the latest wave, which was very much kicked off by Open AIs Doarly, has come about because of a method called diffusion, which is just, it really fits with our current capabilities and it's proved to be very, very rich in possibility. And it's from that, which sort of opening eye was first demo, that we've got all these other models like stable diffusion, like mid journey, and all the

rest. Okay. So obviously, I'm very smart and I know things and I definitely for sure know what diffusion is. But let's just for a second assume that I don't. Like what is it about diffusion that is so new and different, because it really seems to me, and I've only been following this from sort of a, like, end user perspective over the last several years.

Like, something clicked in the last, like, 12 months where this went from, like, nifty science project that people would demo on stage to, like, real thing, doing real stuff in real people's lives. And it sounds like diffusion,

which, again, I understand extremely well and know exactly what it is, is responsible for that. So, like, is diffusion the thing? Like, what is this thing? And how did it sort of change that? I would say there are lots of factors here. Some of them are to do with these new methods, to do with diffusion.

Some of it is sort of, as I say, stuff that's been building for a while. You know, we've got a lot of things that people have been working on in terms of computing capability in terms of data sets that have accrued over the years. Diffusion, though, has really improved things, definitely. To break diffusion down in its sort of simplest term, is it basically, you get models, you feed them an image on one end. Imagine an image on one end, a complete, you know, whatever, it's Van Goghung, Sunflowers. And then on the other end, you have static. You have TV static. You have nothing

discernible, just random patterns. How do you get from Van Gogh Sunflowers to complete static imagery? You get it through a series of steps. And what a diffusion model does is it takes that image step by step by step by step by step, you know, hundreds and hundreds of steps. And it detunes it to nothing. And then, this is a clever bit, is it goes in the other direction. It tries to learn how to get from sunflowers to static to sunflowers. It looks at every single step and it tries to learn about the image by doing that. Now, again, that sounds like, well, why is that better than any other method of making these things? And I think part of the reason is it fits what has become the sort of default approach to artificial intelligence at the moment,

which relies on a lot of parallelized computation. So, you know, people talk a lot about, for example, the utility of gene. EPUs and how they have been really useful in pushing forward artificial intelligence. It's because they are good at doing lots and lots and lots and lots of calculations simultaneously. Now, if you've imagined you're doing this diffusion tango from sunflowers to static to sunflowers, that is a process that involves lots and lots and lots and lots and lots of little calculations simultaneously. So that's sort of the broad strokes picture, the big picture, to use a very appropriate metaphor, of why we've taken off now.

But yeah, people saw Open AI and Dolly do this, and they were like, right, we're going to try that as well. And then everyone else has sort of improved upon their methods and tweaked them in new ways. Part of what's interesting to me about this now is it now seems like we're in this phase of going from, like you said, sort of proofs of concept to like there are lots of slightly different takes on this same kind of thing. Is there an underlying race here to be sort of the technology provider for whatever is coming next? Is everybody still just kind of experimenting to see what's going to happen? And like, what's building out of all of these different projects here? So the really interesting thing with the current, with what we've seen over the last year, is that we've had a new sort of approach evolve.

And it starts with text image models. But it's something that it will certainly affect the wider AI industry. People sometimes call them the first generation of AI labs. And this is deep mind and open AI. And Google Brain. You know, those are the big ones that have made an impact. They had a certain approach to this technology, which is a little bit more cautious, shall we say. So you have OpenAI.

They create an API for Dolly for businesses, and they put in a lot of safety guardrails in there, and they don't give people access to it. Now, the second wave of AI labs have sort of come and look to this model and gone, right, how do we beat these guys? Because they've got so much money. They've got so much computing resources. They've got such a head start. What can we do that's wildly different?

And they've gone, you know what, we're going to build these and we're going to give them away for free. And this has upended a lot of what people thought they knew about how this particular part of the industry should work. The standard bearer for that approach is a company called Stability AI. They promote a model called stable diffusion. Now, a lot of people say they made that model. I want to be a little bit more specific about that. They funded that model. The model was a lot of the research was done by a university in Germany.

and there is a reason for that which will become apparent later, but it's essentially to do with legal liabilities. And it means that stability AI isn't legally liable necessarily for what people do with this model and this sector. So that is important. But their approach has been, why don't we make the model and give it away? And that means our model will get everywhere and we'll collect business as part of that.

So they do run a commercial version of stable diffusion called Dream Studio, but they are doing much more and they're having a much bigger impact by giving this stuff away. And I can really vouch for that firsthand and that I've covered just in the last two months so many companies who have gone, we've got a new text image model and we're doing new things with it. And you asked them, oh, did you make this yourself? And they go, it's stable diffusion. And I'm talking about huge companies like Canva, for example, you know, the Australian design app,

they came out just this week. They had it in beta a month ago. now it's sort of available to all the users, a text to image model. And it's just stable diffusion. You know, they've built some new UI stuff into it. But really, Stability AI have really sort of upended the industry with this, let's just make it free and let's see what everyone else does as a result. Interesting. Okay. And to that point about all these things sort of starting to show up in these products, it seems like there's been this run on it over the last few months. And my sense in talking to some of these product folks is there is this like sense in the

that this is something big, right? Like, this is going to change things. The idea that you can make, especially the text to image thing, is like, everybody seems to grok that there is like something powerful here. And so you're seeing in Canva, you're seeing it in TikTok, like, shutterstock is doing stuff, which I want to come back to. But as far as I can tell, there are these, like, big, meaty questions about how this is supposed to work and how we understand content veracity and how we understand ownership and how

we understand. And we've solved none of it, as far as I can tell. So there's just a lot of people who are like, who are like, oh, these are interesting questions. By the way, it's available in our app right now. And it's like, this feels slightly backwards to me. Is there more progress being made than I'm giving it credit for? Or is this just like total abject chaos? That's a tricky one.

That's a tricky one to answer. I mean, I think there has been progress about this stuff. And in a way, some of the debates that you alluded to about, for example, copyright, infringement, about safety. are debates we've had before in different guises. So, for example, the copyright debate is one we've sort of had with search engines. You know, when they use a snippet of someone's text, is that fair? Is it a transformative use? The debate we've had about content veracity is one we've been having for years with deepfakes.

And there have been, you know, approaches put forward by people like Adobe. You know, they have their content authenticity initiative where they're discussing, you know, watermarks, where they're discussing ways to sort of embed veracity into metadata. data. So there are big things happening, but we have been talking about some of these problems for a while now. Well, yeah, I guess the difference to me is, I think, like, the deepfakes thing is an interesting one, right? Where it's like it's been possible to do for a long time, but it's not like TikTok had a deep fake filter, right? And now we're getting to the point where this stuff is becoming like first class features of these things and you're actually

being, like, they're easy to use and being made easier to use all the time and being just made so available to you, it feels like before we've answered some of these questions. And part of me wonders if there are going to be big new answers about copyright and ownership, and all of a sudden, all of these companies are going to be like, oh, dear God, you know, Stability AI owns all of the art that exists because some judge decided. So we're so screwed. Yeah. I think you're right that the problems are not actually necessarily as different as they've been

made out to be. It's just that, like, they are being put in front of people at such greater scale. quickly. Yeah, absolutely. I mean, let's take copyright infringement as a sort of example of that. So the big thing for these models and copyright is that they are trained on usually vast amounts of data scraped from the open web. Now, we don't always have a clear insight into what that data is. Open AI, for example, has never, has never disclosed the complete contents of its training data. And I have asked them multiple, multiple times, and they always are like, why do we need to tell you? We don't get anything from telling you, which is whatever. It's their prerogative, but I think it's, you know,

deeply suspicious and also completely counter to their name. Opening eye, that's been a misnomer for a long time, but don't get me started. Don't get me started on that, David. So they scrape huge amounts of internet, of internet content, and they scrape it from, from personal blogs, from stock image sites from all these places. Their defense, these companies, is that using this material in this way is covered, particularly in the US, my fair use doctrine. Now, it's funny, you should ask about this because I've been working all week on a report trying to answer that question, is this covered by fair use? And I've spoken to several lawyers over the past week and technologists and risk analysis. And the general consensus is we don't know for sure, which is a really

interesting place to be in. Okay. We're pretty sure that the training of these systems, that collecting this data and creating a model is covered by fair use. Most people are pretty confident on that. Where the questions come in is how are you using these systems? So, for example, you know, if I train an art model on 100 million images and I use it to create a new picture that looks kind of like nothing, that's fine. Like nothing in the training day set, that's fine. If I train it on 100 images by artist X, who happens to be living and I tell the model to produce more images that look like artist X made them, then we have a much stronger case for this not being fair use and for it being copyright infringement. What lawyers told me was that this is stuff that's

going to be settled in the courts eventually. It probably will be anyway. And the interesting thing is we've already seen the first legal case launched in this area, the first sort of legal case aimed at a generative AI model. And it wasn't at an art model, but at a code model. It was a model built by OpenAI and Microsoft and Microsoft subsidiary GitHub. It's called GitHub Copilot. And it basically, think of it like an auto-complete for code. You are typing your programming away, and it will suggest what the next line should be.

The problem is, it's trained on a lot of public repositories of code, many of which have licenses that say, if you would reproduce my code, then you need to credit it to me in some way, because this is the sort of ethos that the open source community is built on. Now, co-pilot does not do that. It never credits where it gets the code from.

And so this lawsuit has been launched saying this is in violation of the DMCA. The lawsuit says a lot. It also says that there's a conspiracy, that there's unfair competition. But this could be something that really changes the entire generative AI landscape. If they get this certified as a class action lawsuit, if it goes ahead, which it may not. We don't know about that yet. Then it could really rewrite the rules about using copyrighted material. Interesting.

And it seems to me that part of the problem or just the mess of all of this, is that it's so hard to know, right? Like, you really can't draw a straight line from, I fed these 10 images to this is why the output looks the way that it does. Like, we have some sense, but all of this stuff is like black boxy to some extent or another. Okay, so have you heard of a tool called Clip Interrogator? No.

I've been meaning to write about it, so you can blame me for not knowing about it, because I should have written about it by now. It was a fascinating thing. So basically, you feed Clip Interrogator, clip is an acronym that is another AI image generation tool, a module component. You feed it an image and it tries to extrapolate what words were fed into the system to create that image. So if you feed it a picture of yourself, it'll be like, how would I describe this if I wanted

to generate this image anew from the system? And so people are now suggesting that with something like that, you could feed someone's art into that, that they've generated with the help of mid-jurney or what it's stable diffusion, whatever it is. And you could get a clue about whether they'd say, I want this to look like a drawing by Greg Wittowski or Rebecca Sugar or whoever it is. Now, whether that will ever be useful in court, whether that would ever be admissible evidence in any way, we don't know. And again, this comes back to the big thing here that's like a lot of this is up in the air. Right. That's really interesting. When you say Open AI doesn't want to share training data, my immediate response is like, of course Open AI doesn't want to share training data. Because, A, that's like, it's like a trade secret to some extent. And also, like, there's a lot to be gained for these companies.

by having it appear to be sort of unknowable magic, right? And I think part of what we're trying to figure out as a like society grappling with this stuff is like, we just need to understand this stuff better, right? Like, yeah, we need to be able to understand how algorithms rank the stuff that we see on social networks. We need to understand why this image looks the way that it does and where it came from and whether it's original. Like all these things are just so much more complicated and so much more intentionally,

like opaque to us. And we actually need to go in the other direction. But like, if I'm open AI, nothing is to be gained by helping you understand how it works. All they want to say is just like, look at this nifty thing that it does when you type in a frog riding a scooter in the middle of the ocean. Like, look, isn't that great? Why do you need to ask any more questions? And I think that tension to me is going to be so, so hard to resolve over time. Absolutely.

And yeah, I should say, like, the reason I press open AI on this stuff is not necessarily because some people think journalists are very like they want to break down this technology. criticize it in some ways. And it's like, no, it's exactly the reason you've just said. I just think it's beneficial to everyone if we knew a little bit more about how this stuff works. And for OpenAI, a company that is going to make a lot, a lot, a lot of money and that has a lot of big techs money in there. You know, Microsoft put a billion dollars in Open AI, and that's what we know about. There could be, you know, more big investments like that. This is why it's important to be able to have some insight into these systems. Now, interestingly, the one thing that OpenEye has told us about this training data, is that it bought hundreds of millions of images, license them, along with

the metadata from Shutterstock. This is fascinating because a lot of people have been saying, what's the first casualty of text to image system is going to be? It's going to be stock photo companies. Because if you can just generate pictures of women laughing with salad or whatever it might be on the fly rather than licensing it, you're going to be in Clover. So it's interesting that Shudderstock has sort of preemptively done a deal. And as part of this deal, they are now going to add text to image systems to shutterstock's website. So you'll be able to generate new stock photos. And they've set up, or they're going to set up, we'll see how far they get with it, something called a contributors fund where they are going to pay royalties to the creators of

copyrighted images used to train Dolly. So they're going to pay people for using their images in this work. Interesting. So the way they're thinking about it to some extent is like your art is still helping make the art even when the art is AI generated. So you should be compensated in some way. Yes, exactly. But then the artists, the photographers, they've said a one-off payment, a royalty, how is that going to replace the loss of my job? There's lots and lots of unanswered questions here. I personally think the contributors fund from Shutterstock is not going to satisfy people. I think there's no way it will be able to give out enough money in order to make people think or feel and to have it be worth their time, that their job has been cloned by this

machine. And it's very interesting because they stand in opposition, shutterstock, to Getty Images, which is, of course, the other big stock image site. Getty Images has gone the other way. They have banned it entirely, AI generated imagery. And I've spoken to their CEO about this, and he's been very critical of these companies. He's saying that we would never sell AI generated imagery on our sites because we think it would put our customers in legal liability that we don't know about. So they've gone exactly the other way. And they're sort of want to. to differentiate themselves here. But it sounds like by that logic, then Gettie is not saying, we think this is a bad idea and bad for, you know, the business and the art community. So we're out.

They're saying, who knows where this is going to land? We'd rather not be roped into whatever weird stuff is going to happen in the meantime. Yeah, they're protecting their customers rather than their contributors, shall we say? Yeah, that's very fair. So yes, what are you seeing in terms of like sensible guardrails on how this stuff is being done? Because right now, we're at this moment now that feels very much to me like the sort of chatbot era when everybody was like, here's a chatbot, do whatever you want with it. And everybody was like, I'm going to make it racist. And they did.

And it was like, well, this was obviously a bad idea. And so everybody has sort of tamped it down and made them smaller and these things are much narrower now. And now they will like book you a flight, but it's very hard to make them be racist while they book you a flight. But now it feels like we're in the kind of here's the text box, do whatever you want with it phase with a lot of the image stuff. Are you seeing people sort of putting race?

on this in ways that make sense? There have been a lot of guard rails put in, and the guardrails really differ based on the company doing it. So I mentioned earlier that we had this first-gen wave of AI labs, which included places like OpenAI, and they have become, you know, because they are now the incumbents, right?

They've now got a reputation. They've got a huge amounts of investment in them. They've become much more cautious, and they do a lot of things in terms of safety. They filter a lot of keywords, you know, so you can't explicitly ask for, naked pictures of people or whatever it might be. And they filter a lot of stuff that would result in obvious bad images. And they also have a look at the output as well. And they also study,

it's not clear how, again, they're not very open about it. They study to some degree what users are asking. So, you know, if you put in a lot of dodgy requests that keep on getting bounced back, they might have a look at your request and go, well, this guy's clearly trying to mess up our systems and they will boot you out. And not coincidentally, I have been booted out of Dali. because I was trying to stress test the system. Damn it, James. I was trying to break it. And they banned my account.

That's good. That feels like the correct response, I would say. Yeah, no. I mean, in fairness, fair play to them. That was the right thing to do. But on the other hand, you have places like civility AI making stable diffusion. They put stable diffusion out there with a sort of very basic tier of filtering. And I know it's basic because, A, I've used it myself and B, I'm speaking.

to Canva, this company that integrated stable diffusion, and they were like, it was trivial to overcome their filters. And we had to add another layer of our own because we just thought it was so easy to bypass them. So that's not me saying it. That's Canva saying it. You know, they're putting their tool in front of school kids. They were like, this is not enough. We got to protect these guys. The thing with stable diffusion is that it's very easy to turn off these filters altogether. And their CEO has said, we just think that this is, you know, it's up to individuals. It's up to people to make the right choices that, you know, they come up with the same argument you hear a lot in the tech industry, which is technology is neutral. I'm very suspicious of that

myself. I think technology always has affordances that guides people in one direction or another, but it's, you know, it's something of a mantra. People say, people are going to do good stuff with it. They'll do bad stuff with it. Amad Mosdak, the Stability AI CEO, he says, I think the good's going to weigh the bad. And also, if the model is out there completely accessible, we'll be able to create better filters more quickly. Now, you know, we'd yet to see what this is going to play out like. I personally think we're going to see some nasty stuff over the coming months. One thing that is sort of just beginning to become an issue is a model called Dreambooth. Now, Dreambooth basically lets you fine tune very easily, stable diffusion, on a specific set of

images and you can use that. A lot of companies have already implemented this, or a few companies anyway, as selfie generators. So you feed the system, you know, 10 of your pictures and it'll do a range of pictures of you looking like a soldier, you looking like a monk, you looking like a whatever it is, a cyberpunk villain. However, this is going to be very easy for people to use for targeted harassment. And I think this is when a lot of the discussions about deep fakes, about non-consensual pornography, these are going to start coming back into prominence again because we're going to see a lot of people mocking up nudes of people, essentially. It's not going to be pretty at all.

Yeah, and I think it's that kind of stuff. Like, I just, I'm permanently suspicious of the argument that people use that are like, let's just figure out what happens and then we'll back our way into successful answers. And we hear it from everybody in every industry all the time, right? Like, let's let all the bad stuff happen so we can learn all the bad stuff. And that doesn't work. It turns out. It's like, it just doesn't work.

But I do think you're right. And I think we're about to enter this really interesting. phase where like we as people are going to start to have to have to figure out how we respond to this stuff, right? Like we're still in the phase where like people just do a dolly search and like take a screenshot of it and share it on Twitter and all of our responses like, that's sort of funny. But this stuff is going to start to end up in like, like it's the new clip art in office is

working with this kind of AI art. AI art in business presentations is like a whole weird thing that we're going to have to figure out what we think about. And art is really interesting because it seems to people like it's very low stakes, but it's actually everywhere and in a lot of ways it's really high stakes. And I think figuring out all the different things it means and how we're supposed to approach each one feels like a big challenge to go through kind of all at the same time. But it's going to be super, super interesting to watch. Yeah, absolutely. But I think you mentioned this stuff is clip art. And I think that's a really

important thing to remember. And specifically, you know, Microsoft has created this app called designer that's part of the office suite. It uses Dolly. And they want to, as you said, use it to generate clip art. Whenever I think about this technology, there's, you know, I flip between, in this sort of quantum superposition, between thinking this stuff is the maddest stuff there is possible, you type an image and it exists, blah, blah, blah, going to change the world. And then I look at what people are doing with it right now and I go, it's clip art. They've just redone Clipper, you know, and it doesn't seem like that bigger deal when you put it like that. And I think our job as journalists is to sort of say what we see, right? It's to just not get too carried away by

what could be and just say what is. And what is right now is a lot of clip art, right? There's definitely stuff that's going to, bad stuff and amazing stuff that's going to happen with it. But actually, the way we are going to get the impact of it really soon and right now is going to be just people putting new images and presentations. And that's going to be really good in many ways. I think that's going to be really fun. You know, if I was a six, seven year old right now and I was making a mock up about, you know, why turtles are the best animals in the sea. And actually, I do that anyway on a weekend. But that's, you know, as one does. Yeah. It has one. does, then I would be so overjoyed that I could like ask for an image of a turtle wearing a crown.

And I would be like, yeah, great. You know, there's a lot of good stuff in there that I think is going to be beneficial and is low stakes. So we should remember that, definitely. Totally. I agree. Thank you for doing this, James. I appreciate it. My pleasure, my pleasure, guys. All right, we need to take a break. When we come back, we're going to dive into the deeply confusing and surprisingly very important world of digital colors.

and why designers are so mad at Adobe and Pantone right now. We'll be right back. Support for the show comes from LinkedIn. If you're a small business owner, you know that every hire counts, but time and resources are limited. Finding, connecting with, and screening the right candidates takes up valuable time you could be giving to your customers. That's where LinkedIn Hiring Pro comes in.

It's built to be your hiring partner, helping you find the right candidates faster. That way you can hire with confidence without turning it into another full-time job. Hiring Pro streamlines the entire process from drafting your job to shortlisting candidates and conducting AI-powered interviews for initial screenings. Its updated conversational interface lets you describe what you need in plain language. Nearly 60% of hirers find a candidate to interview within a week. With Hiring Pro, you spend less time searching and more time connecting with the right talent.

And instead of getting buried in resumes, you get a focus shortlist that actually moves your hiring forward. Join the 2.7 million small businesses using LinkedIn to hire. Get started by posting your job for free at LinkedIn.com slash track. Terms and conditions apply. Support for the show comes from MongoDB. If you're tired of database limitations and architectures that break when you scale, it's time to think outside of rows and colleagues. Because let's be honest, you didn't get into tech to babysit a broken database.

You got into it to actually build something. MongoDB lets you do that. It's flexible, developer first, asset compliant, enterprise ready, and built for the AI era. Say goodbye to bottlenecks and legacy code. Start innovating with MongoDB. There's a reason it's trusted by so many of the Fortune 500. And that's because it's a platform built by developers for, developers. MongoDB. It's a great freaking database. Start building at MongoDB.com

slash build. Welcome back. A couple of weeks ago, a bit of small seeming news kind of shook up the art world. Pantone and Adobe announced that if you wanted to use Pantone colors in Adobe products, you now need to subscribe to a service called Pantone Connect, which costs $15 a month. Adobe kind of blamed Pantone and sort of backhandedly accused it. of money gouging, while Pantone said that Adobe wasn't supporting its latest stuff, and so it needed to go its own way. Whatever the reason, this was a big deal for designers, many of whom have been relying on Pantone colors to do their job for years. Personally, all it did was confuse me aggressively. Aren't colors just like codes on a computer? How did one company come to

own those colors? How do you even own a color? Like I said, I'm very confused. So I grabbed Kristen Radke, a creative director at the Verge, and Verge reporter Jess Weatherbed to help me figure it out. Hi, Kristen. Hi. Hi, Jess. Hi. This is going to be delightful because we are now entering a world I know absolutely

nothing about. So bear with me as we go through this. But I think the place I want to start is I think, Kristen, we have to do like a weird amount of defining like what colors are in order to understand what's going on here. So can you just sort of walk me through like what pantone is and what sort of purpose it serves in the design community? So basically in the 60s, Pantone created what they called a universal color language, which is basically just a giant and very expensive swatch book of colored ink. So Pantone is only used for print design, like if you're making a poster or a product

and you want to know for sure what the color is going to look like before you go into production, Pantan's the best way to do that because other ways of printing produce a lot more variation. Okay. So the idea is basically like what I think of as a blue and what you think of as blue are different. But if we can agree on like Pantone's version of blue, we can accomplish the same thing together. So I mean, like in any version of color, whether it's on the internet or in print, there are a million different ways to make that color. If you look at something like black, for example, the range of black is extreme. There's like true black, there's rich black, and then there's a million other varieties because black can either be built

just out of black or out of all the colors kind of shoved together, differentings put together. So Pantone is a way, like there are a huge variety of black. There's a huge variety of blue and Pantone, just like there's a huge variety of any color, but you can pick the exact one so you know exactly what it will look like because it's an ink. It's not just like a combination of other inks. It's a specific ink. Got it.

Okay. And it's like figuring out how what Pantone sort of owns and doesn't has been very confusing to me because I've like, I've read enough about this to know that like you can't own a color, which is a thing. Yes. But Pantone is sort of the, we've just all agreed that it's. is like as good as standard as any, so we kind of went with it, right? A standard only for print.

So I think that's one thing that some people don't necessarily realize is that Pantone is really only used for print. So like, would it be helpful if I just like told you about different color profiles and different color spaces? Please. Okay. So there are like three main color spaces that designers use most often. So like if you're using a color for the internet, if we're designing something for the

verge, we're going to use a six digit hex code that signifies a mixture of red, blue, and green. So that's within the RGB color space because it's built out of a formula of three primary colors. But the way things look on a screen is a lot of times super different from how they look when they're printed. Screens are backlit. Screens have the ability to kind of communicate a certain range of color. So sometimes that means there's a wider range and sometimes it's more limited depending on the color that you're using. So that doesn't translate exactly to physical printing. So there's also like added variables like what kind of printer you're using and what kind of paper,

because different paper sucks up ink in different ways. So for print, a lot of times we use a more advanced color space, which is called CMIK, and that stands for the four colors you're using, which are cyan, magenta, yellow, and key. And everyone's always confused what the K stands for. It's just black, and it got its name from in traditional printing. You would call the black plate, the key plate, because you'd run it through last, and that's how you get all the detail in the art. But then we have Pantone.

And Pantone isn't so much a color space, like I said, but it's a specific kind of ink. So it ensures you know exactly the kind of color you're going to get. And it also lets you print a lot of stuff that CMIK can't do. It can be a lot brighter, a lot more vibrant. It can do neon. It can do metallics. And a lot of times it just looks better. Okay.

This all is very helpful. It also makes me realize, like, we are down the deepest rabbit hole of things I have never. If you had asked me what CMYK stood for and we just, like, sat here in silence for an hour, I would have never come up with it. But Jess, walk me through this, like, fight between Adobe and Pantone. because as I understand it, there's like some slight backstory to this beef that predates this news. Yeah, we don't necessarily have the entire story. We only have kind of like the sides that they're telling us from their perspective.

Well, perspectives and that. So as far as Pantone is concerned, they have been in a partnership with Adobe now for years and years and years. They do not feel that Adobe has either allowed them to substantially update their product within the Adobe Creative Suite and then subsequently the Creative Cloud. or that Adobe has not pulled its own weight and done those updates via itself. So they basically turned around and said, we no longer trust Adobe to maintain our product. We are removing it from Adobe in order to preserve it and keep it maintained to our own state of standards.

Adobe has, for whatever reason, not fought against that and gone, okay, sweet, I can't see that anyone else is going to have an issue with this, buy, and it will come to a head. There seems to be kind of rumors speculating that either Pantone has turned around and asked Adobe for more. money or vice versa. Either party has then turned around and gone no, and now they're just expecting the

consumer to absorb that cost at face value. Okay, so where this lands then is now if you want to use Pantone colors, if I understand this correctly, you have to pay an extra subscription on top of your Creative Cloud subscription. And if you don't, was I reading this right earlier that basically everything that use Pantone colors, if you

stop paying, it just turns black? That is like the most bonkers, heavy-handed execution of this. They're like, oh, what do you want? Colors? You get no colors. It kind of goes against what Pantone were promising, because they were saying that any kind of legacy documents that already used Pantone colors would preserve the access to those files or whatever books that they were using at the time. That wasn't the case, whoever got that initial update. I have no idea whether those have been restored. I haven't seen any instances of people saying that they've come back. But yeah, when that update went live, existing legacy documents that were opened that were using Pantone colors initially are now just displaying complete black where they were. Kristen, how big a deal is this like in your day-to-day life? Like you work mostly digitally where we're not a print magazine.

So hopefully this doesn't like ruin your day-to-day. But like in your world, how big a deal is losing this thing kind of overnight? Okay. So anyone who's like who's working as a full-time designer or who's working at a, you know, a company where you're doing a lot of print products, obviously they're going to pay for the service. It's going to be no big deal. I think it is a much bigger deal when you're a freelance artist, partially because

working with pantons is already a pain. When I work with artists for a print product, like for example, we've been working at the verge on this big package almost all year about the 20 year anniversary of the Department of Homeland Security. That's mostly online where we're using this really great electric blue, but we are doing a print product that we really couldn't replicate that blue in CMIK. We couldn't get it as bright as it was online. So that's when it might turn to something like Pantone. The problem is we're working with a lot of artists who are used to working in digital spaces and converting art that was built for the internet into Pantone is like the nightmare that you can't believe. Like it takes more time than makes in his logical sense. And I feel like that's always been a broken part of the Adobe Suite,

that translation, because the way you're making something might change, the application might change. You might want to shift and do something for print. And so it's always been really broken. It's always been like there's been this separate program within side of Adobe Photoshop or Illustrator or whatever you're using. So this just makes that a much more complicated mess. Yeah, this is the part of all of this that has been the hardest for me to wrap my head around, which is like, I think of it as like there is a whole spectrum of colors and all of them are identifiable with hex codes, right? And it's like I have the slider in RGB that I can use and I have hex codes. And between those, I always thought I could access all the colors that like goof around with that enough and you should be able to accomplish that. But now we're in this place where it seems like what we actually have is a bunch of standards that are overlapping but slightly different and none of them actually understand each other. And now somebody is saying, well, this one is ours and we are taking it away from you. And that, like, I can see why that would be a big problem for people. Which has always been kind of a problem with Pantone. Like, Pantone updates their color books, like, every year or two.

And they do that so they can charge, like, another $175 for a swatch book. And that means that sometimes if your swatch book on your Adobe suite is out of date, you have to pay to download that new color book. Now what's really happening, though, is that Pantone is just turning into a subscription service rather than something you do every couple of years. Jess, this seems like the kind of thing that it's like a funny matter. to when Adobe went to Creative Cloud in the first place. And everybody got really up in arms about it.

And they're like, well, I used to just pay for a thing. And then I would have the thing. And now you're basically telling me to like pay you a large amount of money every year or like the tools I need to do my job disappear on me and I don't own them anymore. Like, is this just the inexorable path that all of this stuff is heading and to be a designer or anyone is just going to be a million subscriptions that you basically have no choice but to pay for? Like, is this where we're headed?

It kind of feels like that is the way it's going. There are a couple of services recently that have. also switched from a one-time purchase model over to subscription. I can think of kind of like a clip paint studio off the top of my head a couple of months back. Their fan base felt quite significantly betrayed by that. But Baha kind of stand in their ground and saying, well, we as Adobe don't want to front this cost for our consumers, you can now pay for it. They've kind of shot themselves in the foot by proving that the software as a service model is as bad as people were worried it was going to be because inevitably people that still have legacy Adobe products. If you

still have a copy of Creative Suite, for example, CS6 knocking around, that will still have access to the Pantone Color Books within there because it doesn't receive any legacy updates. So if you have Adobe customers to be an outdated product that's no longer up to industry standards, you're actually better off if you're still using Pantone colors, despite the fact that they might be out of date by a significant margin. So are we going to get like a black market for old versions of Adobe products? Like this is just, this all gets so crazy to me where it's like suddenly several years old software becomes a really useful thing to have.

It's already kind of happening. So there's a couple of things that noteworthy here. So one is that Adobe software in general is just one of the most pirated software is in the world alongside Microsoft Office suite. The two of them are just like they are the top ranking in terms of what people will go and legally crack online. For those reasons, they don't want to pay an ongoing subscription fee for the product. Yes, it might be an initial hefty cash lump sum to begin with.

But if you're paying that and you're consistently working with the creative industry, I think it's within like three or four years or something, you would have already paid for the most premium monthly subscription tier of Adobe. And at that point, you're just handing Adobe money for a continual product that may not be necessarily getting updates that are worse your time. The other thing from that is that people are finding that if they either had a pre-update version of Adobe products

or an outdated version of CS6, they can pull the Pantone color bucks from those, I'm not condoning this, by the way, I'm just noting it's happening, but they can pull the Pantone Colorbox from those versions of the software and then manually implant them into the updated version. It may not work forever. It may not work well.

But it is what people are doing in order to try and avoid paying Pantone a monthly subscription. The other thing we would need here is a better version of this kind of standard system. Because I think part of the reason the story is interesting to me is this is the kind of thing that has come up over and over across a lot of tech. It's like everybody looks around and it's like, oh, we've relied essentially on a for-profit company to provide what we see as sort of a necessary service across a lot of the things that we use. And then that company turns around and says, oh, we'd like to make more money.

And everybody goes, well, that sucks, but also fair enough. Like, that's what they do. And it seems to me, like, if I'm an enterprising designer right now with some, like, I don't know, coding skills and time on my hands, like, somebody should start building, like, open source color repositories that start to replace some of this stuff in more functional, accessible ways. Like, is that a thing? Is that happening?

Yep, that's happening. There are other kind of colour books, whereas we consider Pantone to be the industry standard because of its longevity and how many people already use the service, how many printers are already calibrated to use Pantone inks. There are other colour books available. They're just not as popular. You've got like focal tone, true match. An interesting development recently, there's an artist called Stuart Semple who has kind of gained notoriety for beefing with another artist, Anish Kapoor, over gatekeeping art. He developed a paint called like the Blackest Blank. and then sold that because Anish Kapoor, notably, painted the rights to Vantam Black at the time, and he didn't think that was fair. He's now done a similar thing where he has created a open source color book of his own, and he is making it unavailable for Adobe and Pantone as some kind of like protestant. But it's quite easy to develop these color standards. It's getting everyone else to use them. That's the hard part. Yeah. Kristen, what are you seeing?

What do you make of all the other stuff that's out there right now? I think the most important thing here is to remember that Pantone is only for. for print products. So I think sometimes like the confusion over how colors can be used, like if you're not designing for print, this is kind of like a totally irrelevant thing, which is I think why because fewer and fewer people are designing for print products over time, I think it's probably, that would be my guess is why Pantone is now trying to gather more money from the people who are using and printing that way. I mean, Pantone also similar to Adobe has a lot of pirating. There's a lot of fake Pantone ink out there

that's being used at various printers across the world. Do you see any of these kind of new color books and tools out there that have a chance of being, like, just said, the ones that everybody actually latches onto? Because that seems like you're saying, the biggest problem here is that we need one that everybody looks at and goes, okay, this is now the one. And as we've seen, that's a very hard thing to do. I mean, the hardest thing is to, as Jess mentioned, is to get all of the printers to stock all of this ink. I mean, printers across the world are using this as the standard. So it's about kind of like that shift happening in an industry that is not often quick to make changes.

Yeah, as we've discussed before on this podcast, relying on the printing industry to catch up to things really fast is a recipe for disaster. Yeah. Is there a version of this in the digital world? Like, as we move toward digital stuff, like do hex codes and RGB sliders solve this problem for people who work digitally native? Or are we due for another one of these at some point in the future? I don't know.

I mean, the main most important thing to remember, I think, is to understand the distinction between RGB and CMIK if you're working on the computer, because you're basically only going to be working in RGB, which is using hex codes. So those colors will continue to exist. Those are just codes. You know, there's nothing. There's no ink attached to those things to those colors. And that seems like good news, right?

That's a hard thing for someone to decide that they own and take away from you. It's just a bunch of six-digit. Yeah, I suppose unless Adobe was like, if you use more than seven colors, you have to pay us an additional $9.99 a month or something like that. Don't give them ideas, Kristen. Don't give them ideas. Jess, what's your sense of where this goes next? This fight is, I think, it's a really interesting one because, like Kristen's saying, this applies to a smaller and smaller group of people, but this is the kind of thing that a lot of

people care really deeply about. And Adobe has already been under fire for years for what I think people perceive as, like, nickeling and diming people out of a lot of money for the tools they need to do their jobs. Like, is this coming to a head in a meaningful way? Is this just the cost of progress? Like, where do you think we go from here? I don't necessarily, as much as I'd love to say that I think that something is going to change in the next few years, I think that.

everything is so standardized at this point and the captive market that this affects is so small and to an extent kind of so able to absorb those additional costs that it's not really going to affect the industry as a whole. So like illustrators and photographers aren't really going to notice the Pantone colors are disappearing because they simply don't need to use them. But if you own a huge manufacturing business and you work with several big notable brands, you need to use Coca-Cola red. You need to use like whatever like brand specific colors. You have that additional subscription charge every month on top of what you inevitably are already paying for their physical colour book swatches and like any colour matching systems that they provide is negligible.

And it's frustrating that companies will have to absorb those costs. But I think Adobe and Pantone, as much as they're beefing with each other, they are aware that the people that this affects can afford to absorb that. So unless it starts to affect the vast majority of creatives, we're not going to see any kind of difference here. There are kind of discussions of smaller manufacturing business that are, moving over to using different color matching systems like CMYK, because with certain stuff, you can get it close.

There are inconsistencies. You can't necessarily guarantee that you're going to have consistent results with whatever you're printing. But if you are a one-time printer and you just need to, I don't know, make a couple of screen printed T-shirts, then you can get something as close as you need to. But in terms of like big, yeah, big branded projects, I don't see them moving away from Pantone any time soon.

And a lot of times you're printing with CMIK and Pantone. For the example, for the Homeland Project that we talked about, you know, that's how we print it. That's how I printed it. And, you know, when I worked in books, when I worked at other magazines before coming to The Verge, you would use the Pantone if you needed to accomplish something, you couldn't in CMIK. So that's something, like Jess said, that's a really recognizable color, like a Coca-Cola. That's something that's metallic. That's something that's neon, or that's just, like, really bright and vibrant. Those are things that you cannot get with CMIK. But you can maybe print, if you were printing a poster, you might need some neon elements that you print in Pantone and then the rest is CMIK. So a lot of times, So that means you're running something through the press five times rather than four. The inclusion of brands is actually quite important here as well because whereas the system's like CMIA can replicate those colors accurately, Pantone will actually let brands. And patent, they're branded color. So in Pantone so that no one else can use it and provided that it's something on a gamut that none of the other color matching systems can reach, no one else can use that.

Coca-Colololored or like whatever other branded color you have. So it's within a brand's best interest to keep Pantone as a color standard. Like, yes, you could use everything else, but there aren't necessarily brands that want you to do that. Interesting. All right. Well, you have both succeeded in successfully convincing me that colors don't make sense and aren't real. And no one actually knows anything about what a color actually is. That's true, actually.

No one knows anything about color. Yeah, apparently. Yeah. All right. Well, thank you both. I appreciate it. This was really fun. Thank you.

I'm going to send you some color books, David. All right. That's it for the Vergecast today. As always, thank you so much for listening. There is tons more coverage on everything we talked about at Theverge.com. It's a website. It's cool. You should go check it out. You can follow all of us on Twitter. Jess is Zombie underscore Wretch.

Kristen is Kristen Radke. James is J.J. Vincent. And I'm Pierce. This show is produced by Andrew Marino and Liam James. Norie Donovan is our executive producer and Brooke Minters is our editorial director of audio. The Vergecast is a Verge production and part of the Vox Media Podcast Network. If you have thoughts, feedback, feelings, ideas about colors or just six-digit hex codes you want send me. You can always email Vergecast at theverge.com. And if you have tech questions,

call the hotline. 866, Verge 1-1. Send us all your big thoughts and questions about all things tech. We're going to do a bunch more hotline between now and the end of the year. So please keep calling with all your questions. We'll be back on Friday to discuss Elon and Twitter because that just keeps happening. Plus the ongoing crypto craziness, new GPUs, and much more. We'll see you then. Rock and roll.