Scribd & Created By Humans Co-Founder, Trip Adler: How Can Creativity Thrive in the AI Era?

0:01Do you want to be in a world where every single piece of media you consume is created by robots? I don't think any of us want that world. I think it's great if robots are creating media, but I think we also want to have humans creating media, and that's what we're trying to figure out. We want to create that world. Hello, and welcome to Giant Ideas with me, Cameron McLean, and me, Tommy Stadlin. We're co-founders of Giant Ventures, which builds and backs purpose-driven companies. At Giant, we're lucky to meet extraordinary people with Giant Ideas that are changing the world. This podcast brings you behind-the-scenes access to those ideas and the inspiring stories of the people behind them.

0:35We explore how one giant idea can kickstart a billion-dollar company, shape culture, and transform life as we know it.

0:45Welcome back to Giant Ideas. Today we're joined by Trip Adler, co-founder of Scribd, a company he founded with Jared Freeman, who's now a partner at Y Combinator. Script is a digital document library that hosts over 195 million documents with billions of views a year. But today, we talk about Tripp's next act, Created by Humans, where he is the co-founder and CEO on a huge mission to champion AI rights for authors. Giant proudly led the seed round of Created by Humans, and we invited Tripp to talk about this giant idea that affects so many people. How can creativity thrive in the AI era? In this episode, we talk about what happens if we live in a world where only machines create art.

1:23Are we in the pirate Napster era for creativity work, soon to be structured like Spotify? And his early days pivoting at YC, why persistence matters and the unusual way he earned his first revenue.

1:36Trip, welcome to the podcast. Thank you. So good to have you here. We're going to just dive right in it. So you've now created a new company. It's called Created by Humans. Tell us about that. Yeah, so Created by Humans is a company that's aiming to solve copyright in the AI era and get a good balance between AI and human-generated works. And we're doing this by building a marketplace for what we call AI rights, where creators can license their works to AI companies, and AI companies can get the high-quality data they need. So do you think creatives are kind of being ripped off at the moment? Well, I think that it's a tricky situation right now, where a lot of creative works are being used without the permission of rights holders.

2:14and we think there needs to be a new model built where rights holders can have control over how their work is used and get compensated for it. And it's just a tricky problem to solve. I mean, I think AI took a lot of people by surprise and I think it's just a problem that we need to solve. So maybe just taking a step back for our listeners, how do the LLMs currently work? How are they using the work of creatives in things like OpenAI and ChatGPT? Well, these LLMs need a lot of data to learn from and to train on. They need lots of data and it's just very complicated to like. at all. So they've basically just taken the approach of just trading on everything for free.

2:48But I think long term, we need a new model that works for the AI companies and works for creators. And that's what we're trying to put in place. So maybe a stupid question, Tripp, but how does it actually work? So I wrote a book in 2013, put out by WH Allen, a publisher. Where would, say, OpenAI have got that book from? You know, these companies don't talk a lot about where they get their data. It's kind of, it's something that they sort of, yeah, they just don't talk about too much. There's a lot of these various pirate sites and various places where books get out there. So I mean, I can't even really say.

3:18But what we believe created by humans, there just needs to be a really easy, simple, legitimate way to get it. And one of the reasons these companies haven't been able to get them easily is just the rights are really complicated. The publishing contracts for the last 70 years have not mentioned AI rights. So it's not clear who has the AI rights. And the general consensus is that they're co-owned by the authors and the publishers. So in order to unlock the AI rights for a book, we need to clear that with the author, the publisher, sometimes multiple authors. There's often a literary agent involved, an illustrator.

3:45It gets really complicated. So a lot of the work we do is just clearing the AI rights so that AI companies don't have to worry about that. Okay, there's quite a lot to unpack there. So let me start with the first part of what you said. It's not clear where the AI companies actually get this work. That to me sounds quite strange. You think about these companies, you know, Google is one of them, right? OpenAI is another. You think of these as incredibly shiny professional companies. that sounds almost Napster-esque that they have been what kind of using pirate sites to get PDFs of books and it sounds very kind of wild west yeah well I think Silicon Valley is very wild west and I think the world AI is very wild west so um yeah I think everyone's just figuring this out as it goes along I mean I just I don't I don't work at one of the big foundation models I just can't comment and it's a it's a legally sensitive situation so I don't I don't want to spread rumors uh but there's various rumors in the media for like where they get the data and um yeah I I mean, we just think that, sure, it started as a Wild West, but I think longer term we need a legitimate model that protects the creative industries and gets these ad companies the right book.

4:42Because I don't think they want to use pirated versions. I think they want legitimate copies. They want to make sure they're getting the highest quality data. They want to make sure they're getting it directly from rights holders. So I think everyone wants to do the right thing. It's just a tricky problem that we need to solve. So maybe one way to frame what you're doing is you're trying to move, particularly, I guess, the written industry books from the Napster era, which is sort of currently now, where it's a bit of a Wild West. They're using these works without sort of proper authority and without compensating people.

5:10Move that more to a Spotify kind of model where people are generally getting compensated for the work that's being used and listened to. Yeah, I think that's well said. We're kind of in the Napster era of AI and we need to move to the Spotify era. And I mean, this has played out in music and other media types before, and it's just playing out again in AI. And there's just new nuances that need to be solved and addressed. And that's what we're doing at Created by Humans. So let's explain the rights that you were talking about there a little bit more. So as a writer, you write a novel and traditionally with a publisher, they'll buy the rights and then it'll be really clear who owns the rights to the book.

5:40If there's a movie made about it, for example, maybe that's involved in the contract. But you're saying there's basically never been anything in these contracts between an author and a publisher, which says if an LLM wants to use it, if it wants to be used for AI, this is who owns it. And so it's completely unclear whether that's the publisher or the author. Yeah, if AI rights were never mentioned in the contract, generally the default is those rights go to the author. But in some cases, you can argue it's a split, right? Where the publisher and the author both co-own it. And that's why we've built a system where authors and publishers can kind of together work this out.

6:10Because there's just so many permutations of authors and publishers and asking every single deal to be negotiated is just really complicated. So our platform just makes this really simple. So we can kind of solve these AI rights in a scalable way. How have you found that bit? So, I mean, clearly, you know, publishers and authors are on one side of this argument and the LLs are on the other side. But within the publishers and the authors, is that also complicated where you've got to kind of individually barter between them. Yeah, it's really complicated. I mean, we're solving sort of disagreements between content owners, ALMs, and also within the publishing industry.

6:41So it's really complicated, but we're making very good progress. We're pleased with the progress and we think we're on track here. History doesn't repeat, but it certainly rhymes. It sounds like a similar paradigm for technology. Yeah, it does. And I think there's a lot of learn from the past year, but also the past gives us a lot of confidence that we're going to figure this out this time. taking a step back, why do you think this is such an important problem to solve for society? Why is this a giant idea? It's important that we create a world where humans are creating things. I think it's great that AI is going to start creating more and more media, more books, more videos, more music.

7:16But if it's just AI creating things, not humans, that's kind of a terrifying future. So we want to create a world where there's still an incentive model for humans to create things. And if humans are creating things and AI are creating things, these two kind of coexist. and enforce each other. I think that's just a, it's going to be a great world. So that's our, that's our dream idea. That's what I wanted to pick up on Trip actually, fundamentally today was it's really important to compensate existing creatives for their work, right? It's just not fair that they've been ripped off and that they need to be compensated clearly.

7:44But do you think there is actually a future for future creatives to be compensated or because we're getting to a world very closely now of super intelligence? We were together at an event of ours earlier this week and the DeepMind founder was there and he was saying, look, yeah, of course, artificial general intelligence is coming very soon to the human level, but actually it's going way beyond that. And that applies to creatives too. So do you think there will be any creatives who are needed and therefore needed to be compensated? Well, I'd ask you guys, I mean, do you want to be in a world where every single piece of media you consume is created by robots?

8:19No. I don't think any of us want that world. I think it's great if robots are creating media, but I think we also want to have humans creating media. And that's what we're trying to figure out. We want to create that world. So I hope that for hundreds and thousands of years, humans are still creating things. And we're trying to build the system that will allow that to keep happening. My view is that, of course, we want humans to continue to be creative. And I think as more and more of the content gets generated by machines, I personally believe that originality and true human insight will be differentiated.

8:50because if we're training these models on what's come before, then by default, it's going to produce a median of what's been produced before. And that's not, although all art is derivative in some way, I think all art that really captures the attention of people typically is novel in some way and original in some way. So I think just by default, if you're training on what's come before, you're going to produce something that's the average, but perhaps that's me being idealistic and hopeful. I did hear some stat the other day that sort of shocked me as a musician where that was 18 % of the music on Spotify is now AI generated.

9:20And apparently Daniel Ek was saying, I heard on sort of good authority that he thinks it could be much more, but that sort of human society is not quite willing to kind of accept that yet, which made me a bit depressed, to be honest. I really hope that's not the future. I think also, Cam, you know, you were talking there about everything reverts to the meme with AI. A couple of months ago, you know, we all saw that the internet was flooded with these Studio Ghibli type images, right? From the OpenAI's image generator. and there was this big outcry about, oh my goodness, this is terrible. Studio Ghibli spent literally hundreds of hours creating one of those images and now it's been ripped off.

9:53But actually what I felt looking at that was, this is so boring. The whole world is creating the same type of art and actually is an example of why individual real human creativity will always be somewhat more interesting than the mean. Did you agree with that? Yeah, I mean, I agree. Human creativity is, there's something invaluable about it and it needs to be protected at all costs. And yeah, I think it would be, it would be terrible if the world just sort of humans stop creating and AI just builds off of, you know, human creations until now. I hope humans keep creating things for decades and centuries in the future.

10:25And then I hope that AI keeps learning from that for decades and centuries in the future. And I think if we, if we, you know, we build the, build the pipelines to make that happen, it's going to be a great, a great world for everyone. And maybe in the world in which AI does more and more, the things like human creativity, like compassion, like intelligence, like core human qualities that at least today can't be replicated in the same way by machines, perhaps they will become even more valuable. That is one sort of counter reality we could imagine, right? I think that's totally possible. Yeah, I mean, I think that as AI gets more abundant, you know, human creativity will likely be, yeah, more valuable at the time.

11:03Maybe if it's okay, just maybe taking kind of a step back into your journey, what led you to build Created by Humans? I think it's, when we heard the pitch, we're proud investors, we thought it was a fantastically original idea and hugely needed. So what took you on your journey to launch this company? Yeah, well, I just saw the problem very clearly. I saw that all the content owners were suing the ad companies, and the ad companies didn't seem to have a clear answer to this besides just responding to the lawsuits. And in particular, I knew a lot of the people involved. I mean, from my previous company, I knew a lot of people in the book world, and they were coming to me asking for advice on what to do about the situation.

11:41And then the people getting sued were people I knew in Silicon Valley. and I just thought like I wanted to solve this problem. And actually, I saw it as like an interpersonal problem. I was just like, well, if we can just end the lawsuits and get people talking to each other, we'll figure this out. So I just started talking to, you know, my folks I knew in the book industry, folks I knew in the tech industry, and then just started going down the path of solving the problem. And then pretty soon set up a company, raised funding, and here we are today. Who have you got on the platform so far from the creative side?

12:09We have a large number of great authors. some of the big ones we work with are like Walter Isaacson and Doug Preston. They've been two of the ones we work with most closely, but we've got a long list of folks you can see on our website. So Tripp, just explain to us how these different types of rights work all the way from just being able to train the models on some content all the way through to writing a poem in the style of Cameron McLean. That's a good style. Well, we should definitely license that style for sure. Yeah, so when we got started, we were really focused on training and solving the problem around training data.

12:42But we realized the idea is bigger than that. So we've developed a whole new system of AI rights. We have three categories of AI rights. The first is getting content for AI model training. The second is what we call reference rights, which is making content available for RAG. And then the third is what we call transformative rights. And this is for taking a copyrighted work and transforming it into something new. So this could be turning a book into a video using AI. It could be chatting with a character in a book. It could be writing a new book in the style of an author's voice. So we're creating all these different rights frameworks that haven't really been needed before.

13:18And we're creating them for this new AI world. And this will just kind of unlock a whole new range of possibilities, what can be done with copyright content. And just for listeners, RAG, you mentioned in the middle there. Yeah, RAG stands for Retrieval Augmented Generation. It's kind of a buzzword in Silicon Valley right now because a lot of people are using it. it's like imagine if you ask a question and the answer is on you know page 152 of a book rag will allow to kind of get that that answer directly versus having to train on it in advance so just it's a new it's a new technology and they they need you know ai companies need data to power rag and and we just are another data source for that got it and maybe if we compare it back to spotify and we were using that analogy earlier on i guess spotify had this incredibly positive on the music industry overall, where Natsda came along and was ripping off music, and the whole industry was declining.

14:09CD sales went through the floor, and there was kind of complete panic in the industry. Enormous boom since then because of Spotify and other streaming platforms, giving those rights. But I guess if you dig a little deeper, it's been brilliant for the mega artists, not so great for the smaller artists who struggled to monetize through Spotify as much. How do you see that playing out in the world of creative? So you've got some really big name authors you mentioned walter isaacson and so on will they get the lion's share of revenue and how will it work for kind of smaller creatives who might be more at risk from ai yeah it's a tricky problem i wouldn't say we have the answer but i we're you know we're dedicated to try to make this work for for all kinds of creators we want to make it work for for the big big authors i'm making for small authors eventually we want to make it work for you know anyone who writes a poem um so that's what we're trying to do i i wouldn't say we've like solved it yet but i mean that's a core part we want to do is to make this really you know widely available to creators of all kinds.

14:58And perhaps the way that creators engage with their audience will evolve. I mean, Spotify in terms of royalties hasn't been great for smaller creators, but it has unlocked a new sort of large avenues for touring in particular. So I think it's hard to kind of predict where this heads in a couple of years. I agree. And given that the alternative is piracy, I mean, it's just better to get a paid model going where creatives have some control versus just having this happen without control, without compensation. So we think really any baby steps in this direction is hugely valuable. And as an author, is there any sense of how much of their overall income this is going to make up?

15:32So you've got your book deal and you're getting paid for that. And maybe, as I said, there's some TV rights that might come. What percentage of your overall revenue as a creator is going to come from AI rights? Yeah, I mean, we really don't know, but we believe this is the next big model for the book industry. If you look at the history of the book industry, they're keeping new models that are new ways for authors to reach audiences. So first it was print books, and then it was eBooks, and then it was audio books, and there's been things like movie rights and podcast rights. We think AI rights are going to be like the next big thing.

16:01So it's going to take time to build it out, build the model, but we think long-term it's going to be the model of the future. And tell us a bit about that. I know Created by Humans has kind of a specific meaning, particularly kind of with how copyright can evolve. And how could you see this model and this technology being applied to other forms of creativity outside of books? Yeah, I think it being applied to everything really well. We just want to solve it for one vertical before we expand. And it's hard enough to solve this for one thing. So we're trying to focus on books, solve it for books.

16:29And then from there, we can expand this to film, to journalism, to music. I mean, really to any kind of human data. Let's talk a little bit about policy. You're here in London with us. And we were at an event with Peter Kyle, the Secretary of State for Science, Innovation and Technology here in the UK. And Elton John has been rather rude about our friend Peter. I called him a moron because they're so upset with the, I guess, the imbalance that the creative industry here in the UK sees between the creators and AI platforms and perceive the government as, you know, favoring the AI companies because the government really wants to supercharge AI.

17:03So there's a big battle going on there. Government sort of stuck in the middle. What do you think government should be doing, if anything, on updating copyright laws for the problem you're solving? Well, there's, yeah, there's a few approaches to solving this. One is on the policy side and other is by, you know, with these lawsuits and seeing where that resolves. I mean, we're focused on solving this on like the product side. I mean, we're just talking to customers and we're building products that solve the solution. I think that the government could certainly play a role in setting some policy here.

17:33We're putting together our kind of proposal. We're going to share with the world soon. And I don't think that the government needs to do a lot. I think mostly government just needs to let market forces play out here. But there's probably a few things government can do to sort of like unstick the situation and just kind of get this get this new forever i think you're looking at history i would say that government will not be the thing that moves the needle here um spotify we've referred to a few times is when you give both sides a great product that's a win-win situation that also works for consumers that's that's what we'll move it forward i think i mean that's what we believe and that's how we're trying to solve it and i mean yeah we're solving it by building a product and i mean i think there's probably some things that the government could do in terms of uh both protecting creators but also in terms of protecting air companies and making it easier for air companies so there's like a few, a few policies put in place.

18:21Um, but that's a whole long conversation about what, what they could or could not do. Every startup and founder needs to run through a few brick walls to kind of see their vision come to life. What are some of the walls that you're having to run through with created by humans to make this vision come to be? Well, I think we're running through brick walls constantly, right? Um, I mean, I think it starts out where you, where you go talk to authors and some authors are really enthusiastic about AI and some just don't even want to talk about AI. So we have those kinds of problems. Then you have AI companies.

18:54Some of them are into this idea. Some of them just want to let the lawsuits play out. So there's just a whole range of opinions here. And I think we just need to address each of these concerns one at a time. And I think by building a product that is there if you love AI, if you don't like AI, if you're in between, it just gives control to each of the different parties, given their different preferences. I think that's how we solve this. But I think it's just, I mean, that's what it feels like building a company is just one brick wall after another. And we just kind of clear one at a time and eventually it all comes together and things move forward.

19:25Trip, let's go back in your journey a little bit. This is not your first rodeo. You were the founder and CEO of Scribd, the digital publishing platform, created that from scratch in your apartment, right? In San Francisco. Tell us about the founding story of that. Yeah, that's right. We started in the Y scraper with all the other Y. You guys know the Y scraper? Tell the listeners about it. Tell the listeners, yeah. Well, in the early days of Y Combinator, a bunch of Y Combinator companies all moved into one apartment building and we built our companies out there. So it was like other companies that were like Dropbox and Weebly.

19:55I'd ask you guys, I mean, do you want to be in a world where every single piece of media you consume is created by robots? No. I don't think any of us want that world. I think it's great if robots are creating media, but I think we also want to have humans creating media. And that's what we're trying to figure out. We want to create that world. That was, you remember Justin.tv? Yes, of course. Before it became Twitch. Yeah, so they were in the, so Justin.tv was in the apartment. So basically, ever you go, there was Justin with a camera on. So there was like the whole thing was being live streamed.

20:24It was like a YC version of Big Brother. Everyone's on TV. It was a lot like that, yeah. Does that footage survive? Nothing dies on the internet, so it's somewhere. Yeah, I don't know. I mean, it was an interesting time. And yeah, I mean, there was always like venture capitalists coming by the building and just media coming by. It was, yeah, it was a really interesting time. How old were you at that time? Early 20s, just out of college. Amazing. Amazing. And so tell us about the idea and you had a conversation with your father. Is that right? Which was part of the inspiration? Yeah. We're at Harvard, me and my co-founder, Jared.

21:01Senior year at Harvard, we just started like playing with ideas for companies. We tried a few different ideas. We applied to YC back in 06 with a totally different idea. Yeah, we applied with one idea. Paul Graham told us that he would only invest if we did different ideas. We changed the idea, then we changed it again. We're actually one of the first companies to rapidly pivot a lot before we launch anything. We probably tried like 10 ideas. What was the first idea that you applied to YC with? The idea we applied to YC with was going to be this sort of like transportation search super app sort of thing.

21:37That's the most way from where you ended up. Yeah, it was a lot like it was actually we described Uber perfectly. But this was pre-iPhone, pre-easy payments. We were just way ahead of our time. And I think from that I learned like you can't be too early with an idea. Being early is as bad as just being totally wrong. So that was the original idea we applied to YC with. And then PG said, do you have any other ideas? And then we said, well, we're going to build Craigslist for colleges. And then PG said, well, that's a great idea. And then he funded that idea. And then we launched it at Harvard and we called it WhoList.