Subscribe
Share
Share
Embed
AI is the biggest technology shift of our lifetime. This show is about how to profit from it together. Each week I talk with the founders and CEOs closest to AI and Content, the ones figuring this out in real time. I’m also building an AI content business myself and share the lessons I learn along the way.
WHAT WE COVER
THE TITANS: How companies like OpenAI, Anthropic, Meta, and xAI are moving, and why their decisions matter.
THE INCUMBENTS: How content giants like Disney, News Corp, Universal Music Group, and Reddit are responding to AI, and what it means for creators and publishers.
THE PLAYBOOK: Real lessons on AI business models, content strategy, IP licensing, distribution, and getting paid.
ABOUT YOUR HOST: Rob Kelly has interviewed Steve Jobs and Bill Gates, helped pioneer early web content licensing, and built multiple companies with more than $100 million in total sales. His work has appeared on CNBC, CNN, TIME, and Entrepreneur.
Beyond business, every episode explores what AI means for jobs, creativity, families, and the next generation.
If you want clear thinking based on real experience in AI and media, Media and the Machine is your guide
Thanks! -Rob
I'm Rob Kelly, this is Media and the Machine, a show about the biggest technology shift of our lifetime and how to profit from it. Each week, I talk with the founders and CEOs closest to AI and content, the ones figuring this out in real time. I'm also building an AI content business myself and share lessons of what I learned along the way. You know, life's funny. I began my career lucky enough to interview leaders like Steve Jobs and Bill Gates.
Rob Kelly:Then I went on to be a three time founder and CEO, driving a $100,000,000 plus in revenue and some failures too. And now I'm back at the table, interviewing this new world's current and future leaders. This isn't only a business story, it's a human one. So every episode ends with me asking my guest what AI means for our jobs, our families, and the next generation. We'll figure this out together from the inside.
Rob Kelly:Welcome to Media and the Machine. My guest today is Thad McElroy, author of the AI revolution publishing and the most strategic thinker I know on where AI and books collide. He's an insider and knows the big publishers, but he also tracks more than 1,800 startups in the space, including 300 in AI. In this conversation, Thad breaks down the quote original sin of AI in publishing, why it's driving so much fear and anger across the industry. We talk about why publishers are in his words constipated by copyright, and how that same mindset crippled them during the past tech waves from Amazon to Kindle to the Google Books project.
Rob Kelly:He shares a wild story of an author who made over a $100,000 by creating 200 books using AI. He also lays out his simple 15% framework for publishers, calls out one common approach from publishers as pathetic. He explains John Grisham's AI pushback, he's banning it as a bad move for the big five publishers. And for those of you building startups in this space, he lays out what he calls the manifest destiny of publishing, and how AI could finally make that a reality. Make sure you stick around for the final question.
Rob Kelly:It turns personal in a way I didn't expect. Special thanks to Zach Stewart at the Knessa Art Gallery on Montgomery Street in San Francisco, right at the heart of AI for connecting Thad and me. Now please enjoy my conversation with Thad McElroy. I know you're excited, bullish about AI for book publishing, but can you play devil's advocate for a moment? Can we just air out some dirty laundry about why book publishers and authors are so fearful of AI, so scared, freaking out in many cases, suing AI?
Thad Mcllroy:Absolutely. The concerns mostly have to do with content theft. You know, all of the material that was hoovered up into Chatuchi BT and into COD and into Gemini. You know, they they could not have made those engines work if they'd not obtained many hundreds of thousands, perhaps millions of books. And they did so without compensating the authors.
Thad Mcllroy:And so there's this, what I call the original sin of the AI companies And that's, it's a big sin. You know, it may not be illegal, but it's a big sin from a creator's point of view. And then the, you know, the obverse of that from a creative perspective is not only did it steal all my stuff, now it has created a technology which can replicate all my stuff. And we're seeing more and more instances. There's a big article in the New York Times on Sunday about AI generated stuff that's actually selling.
Thad Mcllroy:You know, we were saying until recently when the AI stuff has stopped, nobody will buy it. It's getting so much better, people are buying it.
Rob Kelly:Absolutely. I bought a couple of books for my son already and I bought them just because they were the only ones available on a niche and they were clearly AI generated. So Mhmm. My son was happy.
Thad Mcllroy:Doesn't care.
Rob Kelly:And can you give a little bit of a primer on the LLMs and how they used Books? Obviously, they're not admitting to use of pirated books. Maybe Anthropic did in some way in their settlement of their lawsuit. But can you sort of just give a primer on what's known about the types of books that LMs use to train AI?
Thad Mcllroy:One of the things they love about books, the embodied AI, what they love about books is that books make logical sequential arguments. A statement is made, question is raised, an answer is posited, exactly the kind of thing that the AI engines want to be able to replicate. The books also in many cases contain information that the AI engines then abstract in some of their responses. And the fiction, you know, brings in a kind of linguistic fluency that you don't find exactly that kind of linguistic fluency in any other format. And so that has enriched the models a great deal in their obsession with language.
Rob Kelly:Can you share just roughly what the scale of the volume of pirated books that AI used?
Thad Mcllroy:That it was so fascinating and horrifying and wonderful and all you know, the best of times, the worst of times as Dickens would have it, to find out that Anthropic had downloaded 7,000,000 books. 7,000,000 books. Right? But they were stealing books from, you know, over decades. So on the one hand, you know, it's like 7,000,000.
Thad Mcllroy:And then when the courts delved deeper, it turned out that only half a million of those books are part of the settlement. And that's due to an abstraction in the copyright law that if if an author or a publisher does not register the copyright of a book with the copyright office, it's not eligible for legal remedy. And so on the one hand a lot of books got in there where no compensation is going be paid. On the other hand there's actually 1,500,000,000.0 that's going to be paid out to authors and publishers and that's a nifty thing.
Rob Kelly:But out of those 7,000,000, only 500,000 of them were copyrighted works that got payment as part of the settlement.
Thad Mcllroy:That will get paid. Will get paid now. Yeah. Yeah. I spoke to a publisher the other day and the deadline for filing claims passed at the January.
Thad Mcllroy:But apparently, you know, there's just so much complexity in establishing, you know, what about if a book that had four authors and only three of the authors sign up? Now what? You know, what about a book where the publisher went out of business? What about a book where the author died? And so all of these sort of complications are gumming up the works and it may be some time to that 1,500,000,000 gets released.
Rob Kelly:And the mechanics of it, can you talk about how an AI company like Anthropic would even find these 7,000,000 books?
Thad Mcllroy:There are several online pirate sites. Anna's is a really big one, there's a few others. I think they used Anna's in this particular case for most of them. And these pirate sites, they're very interesting in a way, you know, I've downloaded a few books and they're complete. Somebody goes to the trouble of either scanning a book or getting the e book and taking off the DRM from that.
Thad Mcllroy:And you know, this all of these books exist, many more than 7,000,000 exist in Anna's archive. And there's another archive, Sci Hub, which has all of the scholarly journals, has millions of scholarly journals that can be illegally downloaded. So there's a whole pirate world of folks who say they believe that information should be free. But if you want a license to their free information, that's actually quite expensive.
Rob Kelly:That's ironic. So you have to pay them to get their feed.
Thad Mcllroy:To get that scale. Yeah. If you wanna download a scale, it's a substantial fee that they charge.
Rob Kelly:Yeah. I guess hypocritical is a better word than ironic in that case. Can you take us down memory lane a little bit? Some kind of other reference points in technology's history and lessons learned from those milestones. Yeah.
Rob Kelly:So the first is Amazon. So 1995, they launched selling books. Seems like that's obviously the first big book tech event in Internet history. What did we learn from that that would help us think about the new era of AI?
Thad Mcllroy:It was catastrophic from a book publishing centric perspective. It was pre Amazon, of course, the only way to get books was to go into a bookstore or order them by mail order. Someone, you know, that you could have them shipped to you, it took weeks, you know, for that to happen. Into this mix comes Amazon which destroyed at that time the independent bookstore sector and destroyed much of the chain bookstore sector as well. The upside of course was that, you know, many more books were instantly available at better prices no matter where you lived.
Thad Mcllroy:And that's an undeniable advantage. And later on there's also the advantage of self publishing that we can talk about. But in the short term it was such a disruption to the retail structure.
Rob Kelly:Okay. So if I had to sum it up, really devastating for retailers and that world, but a net positive for consumers in terms of pricing.
Thad Mcllroy:Certainly was. And you have to acknowledge the consumer, right? I mean, you know, we're very sorry your books are going under business. On the other hand, I can now choose between 10,000,000 books at really good prices.
Rob Kelly:Yep. Yep. Hard to argue with that. And I just want to remind folks that FAD is bullish and optimistic about upside in AI, but we're walking through memory lane here. So let's fast forward a bit around seven years later.
Rob Kelly:So the Google Book project, can you also describe what that was a little bit and how that might inform our view of the next generation here in AI?
Thad Mcllroy:The Google Book project is enigmatic in my thinking of it, you know, from a historic perspective. On the one hand, as I perceive it, the impetus for the project at Google was part of their desire to, you know, be a repository of all the world's information, of all the published content in the world. And so, you know, they wanted every website, they wanted every journal, they wanted every book. And so they arranged to start digitizing many millions of books. And of course the publishers being publishers immediately said, well you can't do that.
Thad Mcllroy:You don't have the right to do that. You have to give us money if you want to do that. And they couldn't see the advantages, they couldn't see the upside, what they saw was theft. And that's perspective also informs the AI era, it informs every technology change in the publishing industry.
Rob Kelly:Yeah. When you and I last chatted, you you were calling the Google Book Project the great tragedy. And I at first assumed you meant a tragedy for book publishing in general. But what did you mean by the great tragedy? It's a little different than that.
Thad Mcllroy:Indeed. The tragedy is for the public, for us, for for researchers, for readers, for people seeking knowledge. You know, this could have been an absolutely extraordinary, you know, Library of Alexandria, you know, the much mythologized Egyptian library. Imagine, you know, if we could have every book online accessible in a moment. Instead we have the Amazon model where not all the books are available, they're all at a certain price, they all have digital rights management locking them in.
Thad Mcllroy:It's a very different kind of business model. And so I feel that was a tragedy.
Rob Kelly:And how'd you describe, you had a funny phrase for the publishers and the attorneys for publishers were constipated by?
Thad Mcllroy:Yeah. They won't like me for saying it, but I do feel they're constipated by copyright.
Rob Kelly:Cop constipated by copyright. Okay. Got it.
Thad Mcllroy:Yep. Yep. It's understandable, you know, they say, well the industry is built on copyright. Yes, it is built on copyright, but it's also built on the exchange of information, the sharing of information, the sharing of inspiration. And the publishers are constantly, you know, slapping that with a truck coming down the road saying copyright, copyright, copyright.
Thad Mcllroy:And the two don't mix comfortably in many many scenarios.
Rob Kelly:Alright. Next milestone. So we fast forward from the Google Book project next was ebooks. I'm kind of calling Kindle the in circa 2007 a key milestone. What do we learn from that?
Rob Kelly:From ebooks in general?
Thad Mcllroy:E books to me are, you know, one of the great innovations in publishing. And I'll talk a couple of ways in which that's the case. Self publishing per se predates e books substantially. There's there was a guy in the Bay Area, Dan Pointer, his name was. He's an almost legendary character in self publishing because he was out there in the I think it's starting in the seventies and certainly in the eighties evangelizing the possibility that you could, you know, arrange to print your own book and sell them directly.
Thad Mcllroy:So we had the sort of original conception of directly from author to reader rather than having that intermediated by a publisher. Good. But with print books that was still very, very cumbersome. Jeff Bezos, bless his heart, in 2,007 put the energy behind releasing not only a format but a device that could consume the format, the Kindle reader, and blasted open the possibility for individual authors to get their work out there. And created a massive and underappreciated industry of self publishing that now is probably, know, want to say it's about a $4,000,000,000 industry of individual authors publishing their own work and reaching out directly to readers.
Thad Mcllroy:And that to me is both marvelous and wonderful.
Rob Kelly:Net positive then, sounds like.
Thad Mcllroy:Very much so. Of course, you know, publishers, anything that threatens their business model is not something they approve of. And self publishing to them is still something they look down on, very much look down on, until there's a book that they realize they could offer to publish on behalf of the self published author and make a billion dollars. And then they love self publishing.
Rob Kelly:Okay. So ebooks and and Kindle really ended up being an extra distribution outlet for publishers. I mean, that's a new way for consumers to consume books.
Thad Mcllroy:Well, for self published it did, of course, you know, it was also something that the commercial publishers reluctantly took advantage of. It's another instance where the technology of version of book publishing reared its ugly head. Book publishers looked at e books and they saw a threat, not an opportunity. They saw something that was going to undermine their ability to sell print books, which was their business model. E books, as Bezos originally priced them, he had them at $9.99 maximum price.
Thad Mcllroy:And of course the paperbacks and hardcovers were higher priced and that was going to undercut it. And so that led to litigation and led to a way in which publishers were able to charge, you know, $12.95, $14.95 for a digital file that was, you know, two megabytes. And that's still going on. They still have this business model that does not support e books, but accepts them reluctantly.
Rob Kelly:So then shortly after Kindle and eBooks, I wanted to ask about audiobooks because shortly after Kindle came out, maybe in just a year later, 2008, Audible was bought by Amazon and that really changed things too. Was that a net positive for book publishing?
Thad Mcllroy:That one was. The publishers have been pleased with that. They never hated audiobooks the way they hated ebooks. The audiobook industry, I'm sure you'll remember, some people probably don't remember now, but it was originally Books on Tape. And there was a company called Books on Tape.
Thad Mcllroy:And then there was Books on CD. And you go into Barnes and Noble and there's a whole section in there of audio books, but they were on CD because we didn't have a digital method to distribute them. And Audible was one of the first companies to think through how to make audiobooks available for streaming rather than something that you had to encase in a piece of plastic. And Amazon, again, you know, it's extraordinary how prescient they were to make those kind of aggressive moves at the right time. And they took over the audiobook industry and still are by far the dominant player in audiobooks despite Spotify's entry and so on.
Thad Mcllroy:And that too, you know, has revolutionized the way publishing is done. Audiobooks are now about 15% of most publishers' sales. And they're accessible to self published authors as well. Audible introduced methodology and a business system called ACX, Audiobook Exchange, which allows self published authors to create audiobooks as well. And so that was a a liberation that Amazon enabled.
Rob Kelly:If you asked a book publisher today on those four events, so you got Amazon selling books, still around today. The Google Book Project ended up being a way to discover books through Google, ebooks, the Kindle, audiobooks. If you told a publisher today that you had to take one or more of those away, would they be happy? They'd
Thad Mcllroy:be delighted.
Rob Kelly:They would be delighted. They'd like to go back to the old way.
Thad Mcllroy:Of course they would. And in many ways they're still in the old way. They've never taken full advantage of the transformation. They dislike, sometimes hate Amazon to this moment. Ebooks, they disdain.
Thad Mcllroy:The Google Book project was a smear on the, you know, on the across the heart of the industry. Audiobooks, they love them because they're kind of artistic, you get a fancy, you know, Hollywood actor to narrate them. Well, that kind of cultural exchange is very much in their wheelhouse.
Rob Kelly:And again, I know you're bullish about AI for for books. Can you describe the main sort of pain and problems that book publishers and authors are feeling right now? How would you describe it at a primal level? Is it fear?
Thad Mcllroy:Oh, fear, anger would be the two primary
Rob Kelly:emotions. Uncertainty, maybe two?
Thad Mcllroy:Yeah, but which creates, you know, fear. There's a lot of anxiety about, you know, what the change how extensive will the change be? To what extent are they, you know, going to have to respond to the change? Is it an existential threat or is it an incremental technology? They'd love to think it's just an easier way of doing kinds of things as previous technologies have been.
Thad Mcllroy:The idea that it's going to radically change the industry is not something they want to countenance, but I believe it will.
Rob Kelly:So when when John Grisham, who was part of the lawsuit, George R. R. Martin, Sarah Silverman, the comedian and recent author, when they all joined together and chose to fight Anthropic, it was all out of a combination of both angry about the theft and fear that it's gonna hurt their business.
Thad Mcllroy:The lawsuit per se is the angry part.
Rob Kelly:Okay.
Thad Mcllroy:You know, you can't litigate fear. But you know, they could go at AI companies for not being compensated for the use of their work in the training sets. I believe it was fair use under the legal definition of fair use. But I'm not a lawyer, I've read a lot about it, but you know it's always folly for non lawyers to venture concrete legal opinions. But you know, everything you read about how fair use is educated, you think, yeah, this is essentially what that law has to do with where it's a transformative use of a creative product.
Thad Mcllroy:But that's not the way the authors see it, they see it as straight out theft.
Rob Kelly:Right. And and by fair use, just to be clear, you're saying the training of AI using books is so transformative because it's teaching this new AI how to do things, both facts and how to write, that that's transformative, looks like fair use, but then you have the actual results. You're not saying that the AI should spit out excerpts from the Game of Thrones for George RR Martin or John Grisham's The Firm. They shouldn't be spitting out entire books or chapters.
Thad Mcllroy:Of course not, because that's not a transformation. Right? That's just a theft. Yeah. And sadly for the AI companies, there's a bug in the software that you would have seen in news reports.
Thad Mcllroy:Harry Potter, you know, you can grab all of the first volume of the Harry Potter series or could, you know, out of one of the LLMs, they were caught with their pants down there because that's not the way the LLMs are intended to work. They're meant to try and abstract the language that they ingest, not to store the language. But in some cases, people were able to effectively prompt large chunks of text from the LLMs. That looked bad.
Rob Kelly:So in the end, Anthropix offering to pay 1,500,000,000.0 to these authors, only 500,000 of whom of the 7,000,000 books will get a piece of that, works out to $3,000 per book title. How do you feel about that amount of money per book, or how do authors feel?
Thad Mcllroy:First distinction to make there is, you know, that the 1,500,000,000.0 is strictly for theft. It has nothing to do with the fair use.
Rob Kelly:Okay.
Thad Mcllroy:And people looking at that lawsuit say we won, you know, we beat the AI company and they stole our books. With Nuna that's not the point at all. What the point is that the anthropic data scientists wanting to get a whole chunk of books found that the easiest, most expeditious way was to download them from a pirate site. The point that I feel that is being missed by many in the discussion of the settlement is that the settlement had nothing to do with fair use. And in fact the judge in his opinion around the case made clear that he believed it was in fact fair use.
Thad Mcllroy:The SIN was downloading from the pirate site. That's something that quite apart from AI, you and I, no one is allowed to, you know, download books illegally that are in copyright. That's theft. And that theft is what Anthropic is paying for. And yeah, it adds up to about 3,000 a book.
Thad Mcllroy:Is that fair? Who knows? You know, it's some people are saying no, it should be, you know, a $100,000 a book. Well, that seems excessive. Other people would say $50 a book.
Thad Mcllroy:Well, that seems too little. 3,000? Okay, not bad.
Rob Kelly:Why do you think they ended up settling that part, anthropic?
Thad Mcllroy:Because it could have been, you know, if it had gone to a jury, it could well have been 50,000 a book, 30,000 a book. And so, you know, at that price that they could afford it. It's incredible, you know, that a company can afford 1,500,000,000.0 and still carry on their merry way. I think they raised another 10,000,000,000 that week. But yeah, they could afford it.
Thad Mcllroy:And they couldn't have afforded a judgment of 30,000,000,000 or 50,000,000,000, which technically they were liable for.
Rob Kelly:Yeah. Those investors don't like seeing litigation there in the business plan. I should just point out that Anthropic of the major LLMs is actually the only LLM that I know of who has not paid for any content through AI content licensing, which we'll get into in a moment. But OpenAI has done AI content licensing deals as has Gemini, Meta. But Anthropic has been a holdout in that regard.
Thad Mcllroy:It's curious. Right? There's there's I've never heard publicly a thesis as to why they haven't joined the licensing party. On the other hand, the licensing party is also, I think, over celebrated when you actually break down the licensing deals. It amounts to a smidgen of content.
Thad Mcllroy:I view the licenses as largely sort of ceremonial signaling a respect for the value of content. Yeah, we stole it, but we do believe it has value, but we didn't, you know, it's kind of confusing. If you believe it has value, why did you steal it, etcetera etcetera. But given what has already happened, it's virtue signaling, I believe, most of the licensing.
Rob Kelly:Okay. We've established you got these book publishers and authors feeling angry, fearful, but you're bullish on AI. Can you share why they should be excited about AI?
Thad Mcllroy:Lots of reasons. What's most exciting to me, I have the fundamental techno optimism. You know, I've looked at technology over the decades and seen each innovation as something that enhances publishing's core mission, which is to get more creative content out to more consumers, readers, students, libraries. The mission of publishing is the dissemination and technology that enables dissemination. Learning is something that I feel bullish about.
Rob Kelly:Have you ever seen a technology innovation that did not help book publishers and authors in the long run?
Thad Mcllroy:You think about something like e books, did that help? It helped readers. Did it help the publishers? Maybe, maybe not. But we're not in business to help the publishers.
Thad Mcllroy:Right? We're in business to create a connection between writers and readers. Years ago there was a publishing company called Writers and Readers and Glenn Thompson was the founder of that company. Glenn would say to me, you know, that the whole point is writers and readers. And you know, the publisher gets in the middle of that ostensibly as an enabler of the process.
Thad Mcllroy:But to me just as often they become a barrier.
Rob Kelly:I love that. Yeah. In my household, I mean, I've got my wife who winds up each evening in front of her Kindle. And then I've got me reading a print book every night. My son is now reading a lot of stories generated by AI, and I wanna ask you about that in a bit.
Rob Kelly:So we cover a few different bases in our household.
Thad Mcllroy:One of the things when you raise that the the few different bases, one of publishing's great failures is not to see the opportunity that each new format presents to be augmentative rather than reductive. And by that I mean that you bring in e books, you bring in audio books, and the way the publisher should be thinking, I would argue, is wow, maybe I can sell to Rob all three. Maybe because each one is so good in its affordance, Rob's going to say, yeah, I want the print book because I love print and I want to show my children the value of print and indeed print has its own affordances as they say. E books have convenience, searchability, the ability to annotate, all kinds of things that should be there but are very clumsily implemented in the Kindle reader. Not because of Amazon, but because of publishers not really getting behind the format.
Thad Mcllroy:Audio books have been encapsulated as a separate instance of a book where wouldn't it be great, you know, if the average reader said, you know, when I'm encountering this book, when I'm at home, I use the print version. When I'm traveling on the train, use the e book. And when I'm driving, I listen to the audio book. And that would be a great thing for publishing. Instead, they sell one format because of the way they bifurcated the functionality or trifurcated, I suppose.
Rob Kelly:Amen. I forgot to mention that the fourth platform in our household is audio when we're on a road trip. It's right to Audible. Mhmm. Let's talk about publishers and authors being fearful that AI is going to maybe put them out of business.
Rob Kelly:It's a valid concern that's gonna dent their business, change it for sure. Right? What are some ways that they can monetize through AI?
Thad Mcllroy:We should talk separately about authors and publishers with that question because authors have, you know, one methodology and one set of opportunities and publishers have a very different set of
Rob Kelly:opportunities. Yeah, let's start with authors.
Thad Mcllroy:Author's opportunities around AI are are pretty much offset by some real concerns that that they legitimately have. So on the one hand, AI gives them an opportunity if they're fiction writers to brainstorm, to imagine new characters, to think about ways to make the plot more compelling, to spot check, make sure they've not made major grammatical errors. If it's a non fiction book, for research, for, again, throwing ideas against the machine, does this consistent, does it make logical sense? Lots of ways that authors can use these tools to do better work, to do it more quickly, sure. But to do better work, you want to be thinking, how will I use these tools to be better at what I do?
Thad Mcllroy:On the other hand, you know, there's a big article on Sunday last in the New York Times of a romance writer who's using exclusively AI. She produced, you know, some large, like 200 different books using AI, uploaded them all to Amazon, sold very few copies per book. But because she'd uploaded so many, she ended up earning over a $100,000. And this has got the industry in a tizzy because this was the imagined threat was that the machine would take over from the creative.
Rob Kelly:Were the books print or electronic or?
Thad Mcllroy:Electronic. All electronic on Amazon.
Rob Kelly:Okay. Electronic only, so very scalable. Yep. Fascinating.
Thad Mcllroy:Yep. It's the first sort of well documented in a major publication where people are seeing a true instance of AI taking their jobs as writers. And that's that's that's troubling and terrifying and I certainly sympathize with their concerns.
Rob Kelly:Okay. But in that case, monetary upside for that particular author doing it the new AI way?
Thad Mcllroy:Yeah, she was fine for her. And you know, that then points to the transformative aspects of AI, that we have to rethink what our role is as authors and of course publishers the same, being willing to sort of fundamentally rediscover their purpose in an AI era, but that takes a big leap of imagination.
Rob Kelly:So if you were a book author starting from scratch, never written anything before but excited to do better work as you put it, How would you go about it using AI?
Thad Mcllroy:I'm a believer that when push comes to shove, humans are better creators than machines. On the other hand, machines are incredibly valuable enablers of human creation. And so as an author I would be thinking, what is my unique perspective? What is my unique voice? And how can I use these tools to create my work faster, better, perhaps more affordably, distributed more widely, using AI also to help me in the marketing of my work?
Thad Mcllroy:Lots of good things that way.
Rob Kelly:Do you expect that someone starting from scratch can make the same sort of living as an author as someone pre AI?
Thad Mcllroy:No. It's they can't. It's it's a business that's been troubled for a long time by the simple fact that there's far too many books competing for people's attention. Not only are there roughly two and a half million books published every year, on top of that there's at least 50,000,000 books you can purchase tomorrow, choose from on Amazon. So you've got that many books all competing for an individual's attention and okay, they're all in a whole series of topics, but if you look at something, you name the topic, there's, you know, a thousand or 50,000 books on Amazon on that topic, and a bunch new being published every year.
Thad Mcllroy:With AI now we can up the output even further so that exacerbates a problem that already exists, but is going to make it notably worse.
Rob Kelly:Alright. So to all the authors who just got totally bummed out that they're not going to make as much money, what do you say to them?
Thad Mcllroy:Find another way to make a living. And that indeed, you know, do some work with the there's a large association of self published authors in The UK called Ally, a l l I, and I work with them on their semi annual income survey. And we just finished the 2025 survey and we'll be releasing that next month in London. And what we're seeing from the self published authors is that the book is just one part of their income strategy. And the book for many of those authors is a third of what they make.
Thad Mcllroy:And they make the rest of it, it really varies by the type of book. For someone like myself, know, my books bring in 5% of my income, but yet they are responsible for my, or partially responsible for my ability to be a consultant to talk to you about what's going on without my book as my kind of marker that says Thad knows about this stuff because he wrote the book about it, you know, you're able to establish an authority. And it's wonderful at this time that, you know, where books are so threatened that the book itself is still quite powerful. It's a powerful talisman.
Rob Kelly:Social proof. Yes. So I've been in the information business for pretty much all my career. Is it really just come down to now a book author needs to make sure they have a back end as we call it in the industry Yep. Whether it's consulting, speakers, fees.
Rob Kelly:What else?
Thad Mcllroy:For the creative authors, you know, there can be appearances, there can be book clubs, special editions, they can do workshops. For the non fiction authors, it's a different game which has more to do with consulting, training. In my case I'm trying to build a whole AI training course that I'll sell separately from the book.
Rob Kelly:But for for writers of fiction, trying for the big leagues of getting your story licensed by a Netflix or a Disney or movie companies or TV shows as well.
Thad Mcllroy:Yeah. Big payoff if you do that, then people have talked about the opportunity of making a Hollywood sale, but you look at the number of books that actually get sold to Hollywood, you know, it's just a fraction of a fraction of a fraction. Big payoff, very prestigious, but that doesn't happen very often. With the creative authors, one of the things that really worked for them, which is ironic in a sense, is actually teaching other authors how to write. And there's an anecdote I remember from several decades ago now where a novelist was telling me that he went to a publisher with his second novel and the publisher said, I can't sell this book, but if you can write a book about how difficult it was to write your second novel, that book I can sell.
Rob Kelly:What about on the book publishing side? How do you make money if you are a book publisher these days in this era of AI?
Thad Mcllroy:The publishers that I'm talking to, I'm giving them a really simple challenge so that it doesn't overwhelm them in the immediate term. I say think about these tools. If you could drive out 15% of your cost and up your sales by 15%, specifically using these tools, you would go from being a marginally profitable company to a significantly profitable company just with that alone. Still doing what you do today, still a lot of pushback before we even get to the more transformative possibilities.
Rob Kelly:What's going to change in copyright related to books, in your opinion?
Thad Mcllroy:There's so much that needs to change. I mean, the traditions of copyright are many centuries old at this point. The one thing that needs to be transformed is the copyright ability of AI generated work. And, you know, the AI, the Copyright Office rather, takes the position that copyright can only be conferred upon a human, not upon work created by a machine. And, yeah, okay, fine.
Thad Mcllroy:Let's go with that argument for a minute. Now what happens if that human outputs a chapter of their book with a machine, but then goes in and deletes a sentence there, changes a word there, adds a paragraph of their own, deletes that page. Now who created that work? Well, the machine? No.
Thad Mcllroy:The human? Not totally. Is it copyrightable? It's gotta be. It has to be.
Thad Mcllroy:And the Copyright Office is leaning in that direction because it recognizes that, you know, it's an impossible position to hold on to, that that stuff's uncopyrightable, but it still is in that kind of never never land at this point where it's unclear that you, you know, will be able to, as a publisher, take a work from an author and assert copyright in that work. It becomes more complicated with images than with books whereby the copyright office has said, well, the image can't be copyrighted, but maybe the prompt can.
Rob Kelly:So if I had to take your last answers and then also a couple of what some of what we talked about earlier. So it sounds to me like if you're a betting man, you're saying training AI through books is gonna be okay. It'll be transformational. The inference or rag, which are fancy terms for sort of like, let's call it just answers coming out of AI, will be maybe less changed. It's not gonna be okay to return for ChatGPT or Gemini to generate the Harry Potter book series for its users.
Rob Kelly:Sounds correct? Maybe that part stays intact.
Thad Mcllroy:The rag is is its own beast and really complex for publishers to get their minds around. You know, on the one hand superficially we describe rag as if it was kind of at the library, right? The idea now is the machine, you know, goes to your database full of information. This, you know, rag is mostly assumed to be information, I. E.
Thad Mcllroy:Non fiction rather than fiction. And, you know, the machine consults your library, finds, oh, there's the answer that that person over there who asked me as an AI engine. So I'm going go over there, I'm going to grab that little piece of information, I'm going to pay a penny to the library for, you know, borrowing that piece of information, and I'm going to spit it out to the user on the other side who I've monetized how? Advertising, you know, subscription. Well, we'll figure that part out.
Thad Mcllroy:So all of that to me is fanciful and it's based on an older model of the library, of the idea that you know information exists in a repository which is then consulted and upon each use monetized. And I don't believe that that's the way it's going to work in AI as it moves towards AGI.
Rob Kelly:But but to be more specific, if you go to a Chats UBT or a Gemini and ask a question that is related to a book, there'll be some line drawn about how much can be returned. Right? It used to be you could have a little summary, a snippet. Right? Probably not even a full page of a book.
Rob Kelly:Right? But you see that changing? Will you be able to return, you know, a full Harry Potter book with no compensation to JK Rowling?
Thad Mcllroy:No. Because that's not transformative. But what you could return to the query is, I'll tell you about that book. I'll tell you about the characters. I'll tell you about A summary.
Thad Mcllroy:The action in the book. You know, I'll give you a summary. I'll give you an analysis. I'll give you a critique. I'll give you a chance to write your own version of it.
Thad Mcllroy:I'll answer your specific questions or however you want to do it. That's transformative. And so that's something they can do assuming fair use is accepted by the courts.
Rob Kelly:And then if the user wants to buy the book, they're probably going be able to do that right at the AI point of use and JK Rowling and company get compensated.
Thad Mcllroy:Exactly. And that that's a virtuous circle. One that again the copyright folks say, no, you know, you're stealing. And it's like, no, it's a virtuous circle in which the reader becomes more intimate with the text. And they're always, in most or many cases, they'll say, that sounds so interesting.
Thad Mcllroy:I'd like to have the original.
Rob Kelly:Absolutely. As David Ogilvy used to say, you know, the more you tell, the more you sell. And certainly if someone reads, and that's why Amazon, because they're very smart about this, they let you look inside the book because they know that that's more likely to hook the person than just showing them the title.
Thad Mcllroy:Absolutely. I'm always trying to push publishers to be better informers of the potential readership. You go to a publisher's website, you can't even find the table of contents. You find this, you know, crummy little description. How about about the author, three sentences?
Thad Mcllroy:You know, it's pathetic that they don't create a rich web experience for each of these books that they've spent two years developing.
Rob Kelly:Crazy. And then the third point is what you're saying is the creative works coming out of AI, that's gonna be up in the air because it's gonna be hard to tell how much a human used AI for some creative work and it'll of course be different in books versus images and video, different rules, but that's TBD.
Thad Mcllroy:Yes. Very much TBD.
Rob Kelly:In terms of AI's distribution, you know, do you believe that in this new world, people will start more often to begin their journey interacting with book type content via one of the big platforms like ChatGPT, Gemini, Perplexity? Will that end up being how they find in the same way they used Google and Amazon to find books in the past? Will AI be the way they start to find their books?
Thad Mcllroy:It's what's happening now. So it's and it's increasing, you know, dramatically and rapidly. And so short answer, yes. I mean, that's that's what they're doing. It works better than other methodologies for finding answers.
Thad Mcllroy:Yes, it's got all kinds of problems, factuality and hallucinations, blah blah blah. But it's actually, it's tremendously efficient. And so as a transformative model from the previous Google search towards the AGI future, this is a really nice stepping stone. I don't believe that's what we'll be doing five years from now. I think we'll be doing something quite different.
Thad Mcllroy:Don't ask me what, I'd be rich if I knew. But it doesn't feel it feels like the metaphor is a metaphor from the nineteen seventies or eighties, not a metaphor from 2050.
Rob Kelly:Yeah. I've heard also Wiley, I think you brought up earlier, Wiley is now allowing Perplexity users to if you're already a Wiley customer, you can use Perplexity and just access the academic journals and books from Wiley through Perplexity because customers starting with places like Perplexity these days.
Thad Mcllroy:Yeah. That that model to me again is quite flawed. Okay. I can explain why to illustrate the Wiley licensing model. A company told me about an instance where they're licensing their content to a startup for veterinarians called VetGenie.
Thad Mcllroy:And Wiley has a large corpus of journals and books that support veterinary science. Great. And I asked ChatGPT what percentage of the broader published corpus of information for veterinarians is represented by the Wiley corpus, and it estimated about 40%. So if I'm a veterinarian using this startup site and I ask you for questions, you can only give me Wiley answers. And Wiley doesn't have all the answers.
Thad Mcllroy:And so at least theoretically, this startup has to license from all of the information sources if it wishes to be what it proclaims to be, the place where veterinarians can go to get answers for medical conditions of animals. And I don't see how they can get around that dilemma. Again, the current licensing models are various, you know, one off kinds of models. And I don't see that as the way we're going to be doing things in the future.
Rob Kelly:I want to get your hot takes on some different parts of the book publishing industry. How about Penguin Random House?
Thad Mcllroy:Penguin Random House is a beacon because they're the largest in the world. And they're a very sophisticated company. And they do have AI scientists on board. They have a whole sort of think tank within Penguin Rainer House in The United States and they're part of a far larger international operation, Bertelsmann based in Germany. Clearly internally they're saying, oh, we need to find out how big this transformation is going to be while running their internal think tanks to find out exactly how they're going to be able to take advantage of it.
Thad Mcllroy:It's really difficult because the authors are coming to them now and the prominent authors, you know, John Grisham will say in my next contract with his publisher, it's, I don't know if it's Simon Schuster, Penguin Random House, he's with one of the big fives. He'll put it into his contract, you cannot use AI in the editing, the cover design, the marketing of my book. I will only sign a contract if you agree to not use AI tools with my book. And I say to the publishers, that's a mistake. I mean, it's hard to say no to John Grisham, I understand.
Thad Mcllroy:But you're going to end up with two workflows. You're going to be a publishing company that has the one division that doesn't use any AI tools and the other division that uses them, you know, extensively.
Rob Kelly:What would John Grisham not want AI to be used for in his book? Is it the cover art? Is it the actual editing? Which part?
Thad Mcllroy:Both, all. The tools for editing are coming along at tremendous pace just using, you know, ChantiPT, Gemini, and Claude in particular as editing tools for manuscript analysis, for manuscript enhancement, for proofreading, for copyediting. The tools are proving, I've used them. They're very, very powerful. And again, it's not just about cheaper, faster, it's about better.
Thad Mcllroy:It's about producing a better quality manuscript with the use of these incredibly powerful tools. But John will say, you can't do that. I just want a good old fashioned editor to do talking to these editors, they still print out the whole damn thing and carry that manuscript home with them and with their pens at night and their yellow stickies. They go through and make corrections and slash little bits and suggest a different word here or there. That's still the way publishing takes place.
Rob Kelly:By the way, I don't know who published it, but I was reading a Percy Jackson related story this morning to my son. And over the last three days, three typos. Like, actual real typos, you know, lacking the word it or the. And I gotta say, AI wouldn't make that mistake.
Thad Mcllroy:It could, but it probably, you know, it it
Rob Kelly:It would do better than humans in that case.
Thad Mcllroy:It would do better. Yes.
Rob Kelly:So how about on with the book publishing companies? How about HarperCollins? How how are they doing in this age of AI?
Thad Mcllroy:Dow Jones, you know, which is the technical owner, has a lot of, you know, has done licensing, you know, is is publicly very enthusiastic about the technology. Harper Collins did that one deal with Microsoft. A very weird deal where Microsoft agreed to pay them 5,000 per standard non fiction book for licensing to Microsoft for training. And that was a little payday for 10,000, they took 10,000 books. And so HarperCollins split that fiftyfifty with the authors.
Thad Mcllroy:My brother-in-law, who has two very modest books about beer, pocketed 5,000 as a result. So that was an interesting little payday, but it it was a one off.
Rob Kelly:Was he happy?
Thad Mcllroy:Delighted.
Rob Kelly:So he got 5 k just from that one deal because it was sort of found money. It was out of the blue.
Thad Mcllroy:It totally found money. The royalties. These books are 15 years old or older, you know, and his royalty checks for $27 a year off them, and then in comes a a payoff of $5,000.
Rob Kelly:And these are books about beer? Yes. So literally literally beer money.
Thad Mcllroy:Yeah. Yes. Very and he is indeed a beer enthusiast. This is gonna cover him for a while.
Rob Kelly:When you look at each new wave of tech, it's an opportunity for a new company in any industry to take a big chunk out of it. So the Internet enabled Amazon. Right? You had audiobooks enabled Audible, now part of Amazon. It's still its own brand.
Rob Kelly:What new brand or type of company will AI enable in the book industry?
Thad Mcllroy:The manifest destiny of publishing is one to one. That each book, it might be a video, it might be whatever, but each content instance will be dynamically created for that individual in real time at the moment that they wish to consume it. And that's always been the manifest destiny because that's the best. Let's stay with non fiction for a moment. I want the book that knows what I don't know and knows what I want to know and knows what I learned yesterday so it doesn't repeat it to me today.
Thad Mcllroy:That's the book I want. I don't want the textbook that was written five years ago, revised three years ago and comes in a $70 hardcover. So then who's going to deliver the one to one book? No one's set up to do that. Publishing as it exists today, it's the one to many.
Thad Mcllroy:And people talk about that business model as if that's the right business model or the one that, you know, we should have. But in fact, the only reason that business model exists is because the printing press couldn't do one to one. The printing press had to print thousands of books in order to be economical and so we got an industry of one to many. But now we have a technology that enables true one to one creative interchange and that's a marvelous thing that no one yet has come up with a compelling startup that delivers on that promise.
Rob Kelly:And I should say, by the way, that you have how many startups in your database that you track related to
Thad Mcllroy:the 1,800.
Rob Kelly:1,800 startups related just to the book publishing industry.
Thad Mcllroy:To book publishing and authoring, 350 of them are AI. And of those 350, 325 have started in the last three years.
Rob Kelly:But none of them are going after the one to one. There's a couple
Thad Mcllroy:that claim, you know, they're ones where they go talk to the book type startups, but they're very banal and the technology is not sophisticated or, you know, it doesn't transcend the chat interface. And so I haven't seen anything that begins to make my lights go on and say, they're going in the right direction.
Rob Kelly:Was it possible it's already game over that I mean, in our household, my son is using Gemini and ChatGPT for one to one stories. That's what he's using it for. I don't think this is an edge case. Maybe we're just kinda ahead of the curve because daddy's got, you know, the $20 per month AI solutions for all these tools because it's what daddy does. But, you know, he will get turned on by, say, mythology overall or Star Wars or Percy Jackson or Atlantis, and we watched Aquaman the other day.
Rob Kelly:But for him to then take the next level, you know, there's only so many movies or only so many books on a topic, and they're not all immediately at our fingertips. But, you know, he's 10 years old when he wants something. It's now Gemini or ChatGPT, and he's making the story. He's saying, what if Luke Skywalker and Anakin Skywalker were both at their prime, who would win in a battle? Stuff like that.
Rob Kelly:But what ifs and outpops, basically stories, mini little books. And I'm pretty sure without knowing everything that goes on in the black box of AI that they are one to one. They're unique for that moment because his query is very customized.
Thad Mcllroy:Yeah. That's great to hear.
Rob Kelly:Much longer than I even gave. His queries can go on and on. I'll I'll leave him in the bathtub and he's still talking voice to text. And then I come back and he's finally done.
Thad Mcllroy:Share my congratulations with him because he's setting the pace for for the rest of us in in trying to explore how this how this works. One of the things I argue, I've been working on an article all year and all of last year and into this year. The title of the article is that AI is a publishing medium. And the argument I want to try and explore and convince people of is that AI itself as it exists today, the large language model AI, is a publishing medium. It's like the Gutenberg printing press in that sense.
Thad Mcllroy:It's a way in which information is assembled, transformed, delivered. That's publishing. AI is a publishing medium. And if we use the publishing lens to view AI, it gives us a whole other perspective that's different than the coder's perspective or the AGI perspective. And that's what I want to spin out into 2,000 words rather than the three sentences that I've spun out so far.
Rob Kelly:Is it then likely that no startup could come around other than an OpenAI and Gemini, someone at that scale who can afford the LLM to do the one to one.
Thad Mcllroy:No. I think it there's lots of other people who can do it.
Rob Kelly:I want to ask you if you were a story writer, whether it's fiction or just a writer non fiction right now, would you just make sure that all the LLMs had everything you wrote and not keep it to yourself? Is the benefit and the upside from starting from scratch and feeding the LLM going to help get you that back end in terms of value and business? Does it just make sense to make sure AI is grabbing your content?
Thad Mcllroy:That question is one that haunts authors and publishers right now. And the dilemma there, and it's a huge dilemma, is if the query to the AI answer engine involves an answer that's contained uniquely within your book, surely you want the Answer Engine to know about that, provided there can be some kind of attribution. But the system is not felt in that way. It's clumsy. It's awkward.
Thad Mcllroy:You don't know if the answer's gonna come from your work. You don't know if the attribution's going to be there. And so thinking it through as that kind of a value proposition I think is the wrong way for authors to think. The way to think about it, in my view, is the age of copyright has changed. You want your creative work to be out there in the world in a belief that the appreciation and the dynamism of the creator economy will ensure you compensation.
Thad Mcllroy:You've got to believe. Well, what's the compensation model? I don't know. But, you know, it's continuing to evolve. People will always appreciate, you know, the human instance of creativity regardless of what tools enabled it.
Thad Mcllroy:And they will be willing to exchange cash for that.
Rob Kelly:One of the coolest innovations in book publishing that I've heard about recently was from you, which is you've got your book up on Leanpub. Can you just share how Leanpub works and sort of the magic of it?
Thad Mcllroy:Leanpub is magical. It's not well known at all. I always mention in my webinars with publishing people and they're like, Leanpub, know, they never heard of it. Leanpub, as it sounds, l e a n p u b, and it's a self publishing platform. I use it as the writing engine for my book.
Thad Mcllroy:I can literally republish my book every time I change a word. Because it's as easy as that and there's no cost to republish it on the platform. So of course I don't do it every time I change a word, but I often will go in and correct, you know, a company that's gone out of business or a new engine that's now available or a new technique that I heard about because Ethan Mollick put it on his blog, and I'll update my book and then I'll republish it by, just by saving it, we can republish it to everyone who subscribes to the book on that platform. At the same time, I'm able to export the book and make it available through Amazon, which I do. So on Amazon you can get the, you know, the e book, you can get the paperback, you can get the hardcover, you can get the audiobook.
Thad Mcllroy:And through Google you can get the French version of the audiobook. And on the Leanpub platform I have 31 other translations which were created on the platform using one of the big three AI LLMs. And so it's it's a cornucopia in terms of the publishing paradigm today in which you want to make your work available in every possible format, Leanpub is a panacea.
Rob Kelly:Amazing. I've got some humanitarian questions that I want to wrap up with. Just kind of speed round, just give me, you know, as short or long an answer as you'd like. Do you consider yourself an AI optimist, pessimist, or some other descriptor on the spectrum?
Thad Mcllroy:I'm a great AI optimist of the use of the tools. I'm terribly apprehensive about the larger issues, not copyright, but the larger issues for the world at large, be it economic issues, environmental issues, all of the panoply of problems that we see. I'm very concerned about many of those. And in that sense I'm apprehensive and perhaps a little pessimistic.
Rob Kelly:What are you most worried about related to AI?
Thad Mcllroy:It's so powerful. Right? It really is the atomic bomb of the information age. And we've we've never seen this kind of power. And then you release it into a world full of very smart people, some of whom are quite evil in their intention and in their nature and you create a very dangerous situation.
Rob Kelly:When do you think AI can replace your role?
Thad Mcllroy:Today. Today. You don't really need me. The answers are available in the engine. Thank God no one's figured that out yet, but the reality is you don't really need that.
Rob Kelly:What do you think of universal basic income, UBI?
Thad Mcllroy:I'm absolutely in favor of that. I grew up in Canada and I'm a dual citizen, but you know, my heart belongs to the Maple Leaf. And so I come out of a different social perspective on how society should support the individual. So I'm very much in favor.
Rob Kelly:And forgive me, I should know this asking the question. Does Canada have its version of UBI?
Thad Mcllroy:Not at all. No. But it'll be one of the first to do so because it's it's in the fabric of the social contract in Canada that should that be necessary, it will be implemented.
Rob Kelly:Wow. What are you telling kids and younger folks in your life about what changes to make in this new world of AI?
Thad Mcllroy:It's very very concerning for young people, very very concerning. It's hard to go to them these days and say, hey kids, you know, look at all this fantastic opportunity, when they're having to grapple with the fact that the existing institutions have betrayed them, largely the educational institution has betrayed them and not delivered what it promised. Particularly, you know, the further up you go in the educational institution. I think kindergarten still works quite well. But that university is collapsing in its value proposition and too much was promised to too many that's no longer being delivered.
Thad Mcllroy:And it's unclear what the replacement of that will be and what that means to the careers in the future for young people. I'm I'm so worried for them.
Rob Kelly:Will you create an AI avatar for your family, friends, or business so that those close to you will be able to have conversations with you while you're alive or even after you've passed away?
Thad Mcllroy:The Well Alive avatar has never had any appeal to me. I'm intrigued by the after I'm gone avatar. I can't imagine many people would want to interact with it. But it would delight me if I could have an avatar of my father who died when I was quite young. And it's been an eternal regret that I never was able to talk to him about the big questions.
Thad Mcllroy:And so it was intriguing to think what that would mean in the current era when I could have a Kim McElroy avatar.
Rob Kelly:What would you have asked him?
Thad Mcllroy:Oh, there were so many things to ask. Primarily starting with, do you love me? Wow.
Rob Kelly:You didn't didn't know while he was alive?
Thad Mcllroy:I didn't know. You wouldn't say.
Rob Kelly:One of those dads that didn't say it. Yep.
Thad Mcllroy:Yeah. Never said it. Never once.
Rob Kelly:Wow. Thanks so much for sharing, Thad. Really appreciate this conversation.
Thad Mcllroy:Thanks for having me on your your podcast, Rob. I very much appreciate it. I'm delighted to get to know you.
Rob Kelly:Well, this is Media and the Machine. A few things about you and me. If you wanna hear about the next new episode, make sure you hit follow on the show in your podcast app. If you wanna go a little deeper, head to mediaandthemachine.com and subscribe. When you share your email with me, can see handcrafted transcripts, read the essays in my newsletter, and be the first to hear about who the guest is on the next show.
Rob Kelly:You can also email me directly from there. Maybe you wanna recommend a guest. I'll give you a shout out if you do. I love paying it forward. From time to time, I also open up office hours and host small meetups for subscribers just to meet, talk, and build things together.
Rob Kelly:If you're creating something of your own or thinking about it, I'd love to help. Maybe you've got a podcast in you. Finally, I don't have a marketing budget for this show. So if it's finding you and others, it's because someone like you passed it along. I'm genuinely grateful.
Rob Kelly:And if you have a moment, an honest rating helps me make this better for you. You can just go to the show page and click one of the stars. And I'd rather you give me a low rating than no rating at all. I mean it. It pushes me to get better.
Rob Kelly:Thanks again, and see you next time.