The Bookshop Podcast
The Bookshop Podcast
Thad McIIroy: AI And The Future Of Books With
Start with the truth: technology keeps changing the book world, but the love of reading isn’t going anywhere. That’s our springboard for a rich, candid conversation with publishing analyst and author Thad McIlroy about what AI can actually do for writers, publishers, translators, and readers—and where it crosses lines that matter. We trace his arc from bookstore floors to indie publishing, investigative journalism, and desktop publishing, then into the fast-moving landscape of modern AI.
We get practical fast. Want a sharper pitch, a cleaner description, better metadata, and smarter keywords? Use AI as a drafting partner—then apply human voice and judgment. Thinking about acquisitions or submissions? Run a secure, opt-out upload and ask the model who the audience is, what comp titles make sense, and how to position the work. We talk tools—Claude for colloquial nuance, Gemini for reasoning, ChatGPT with training opt-outs—and how to protect manuscripts while you explore.
We also tackle the hard edges. Headlines predicting the “death of authors” are noisy, but real concerns remain: training on copyrighted books without compensation, contracts that quietly assign AI rights, and the limits of AI detection tools. Thad breaks down recent rulings vs. piracy, separates legal allowances from ethical responsibilities, and shares what to renegotiate now so creators aren’t boxed out later. On translation, we sketch a hybrid workflow—machine draft, human craft—that can open doors for books that would never otherwise travel, while preserving the nuance great translators bring.
Through it all, we return to the point of the industry: a reader choosing a book, a poet on a shelf, an indie bookseller who knows your taste. AI can help us market smarter, iterate faster, and reach farther—but it shouldn’t replace the human spark that makes literature worth saving. If that balance resonates with you, follow along, share this conversation with a friend, and leave a review to help more book lovers find the show.
Hi, my name is Mandy Jackson Beverly, and I'm a Bibliothes. Welcome to the Bookshop Podcast. Each week, I present interviews with authors, independent bookshop owners, and booksellers from around the globe and publishing professionals. To help us show its own people to share episodes with friends and family and on social media. And remember to subscribe and leave a review wherever you listen to this podcast. You're listening to episode 317. Well, happy new year to everyone listening. It's been quite an eventful couple of weeks for my family and I. I am now a grandmother, first-time grandma, and uh we have a beautiful, healthy little granddaughter. And uh mom and baby are doing well as is dad. That was pretty eventful, and I really needed to spend time with family. As I said on an earlier episode, family is my priority, and uh it was a true gift to be able to spend time with family over the holidays. I wanted to share with you again the dates and the office we have for the Lunch with an author literary series. Uh on January 14th in Ohio, Anne Hood will be my guest author, and she's also going to be doing a writing workshop. And on January 15th, Noel Stock, I will be in Santa Barbara and she'll be talking about her book, A Rom Calm Love Camera Act. And it's a really funny book. If you're interested in any of these events, you can go to my website, www.mandyjacksonbevely.com forward slash events, and you can register and pay for these events there. If you have questions, you can always email me at mandyjacksonbeverly at gmail.com. Okay, now let's get on with this week's interview. Pat McGilroy is a publishing technology analyst and author and principal of the future of publishing based in San Francisco. He's a contributing editor to Publishers Weekly covering artificial intelligence, digital innovation, and publishing startups. His latest book, The AI Revolution and Book Publishing, A Concise Guide to Navigating Artificial Intelligence for Writers and Publishers, was first published in July 2024 and was revised in 2025. McElroy has authored a dozen books and over 500 articles on digital publishing. He is co-author of the Industry Standard and Metadata Handbook, second edition, co-authored with Renee Register, and he served for five years as program director for Cybold Seminar, the publishing industry's Premier Technology Conference. He is a founding partner in publishing technology partners, a consultancy focused on a broad range of strategic technology issues in publishing. An expert on publishing metadata and online book marketing. He has taught the metadata for books course at PACE University New York in their Masters of Publishing program. In 2024, he joined the advisory board of John Hopkins University Press and became a visiting scholar at the Publishing Master of Professional Studies program at the George Washington University. He is a senior member of the Association for Computing Machinery. Hi, Thad, and welcome to the show. It's great to have you here.
SPEAKER_01:Thank you, Mandy. It's been a while since we first connected, so I'm very happy to be here.
SPEAKER_00:Yes, we had quite a few emails back and forth, but we're here now, so that's good. Okay, let's begin with learning about you and your life in Toronto, Canada, and what led you to found Virgo Press in 1977 and later your interest in journalism.
SPEAKER_01:I started uh out of high school for various reasons. I didn't want to go on to university. I wasn't ready. I'm not the only one who's who takes the year off. In my case, I took the year off and never came back. I went to work in a bookshop in Toronto where I'd grown up. And I was ambitious. And within a few years, I thought, well, I'll start my own publishing company, which I called Virgo Press because my partner was also a Virgo, and that was the first name that came to mind. And so the company was, I built it up over five years to, oh, we had 15 staff and we'd published about 40 books when the bank came knocking on the door one day and said, No, we want our money back. You're a money losing operation. Yep, we are. And that was the end of that. And so my father had been a journalist. Uh, he also had a couple books published, and um journalism, you know, it was always appealing, right? There's a there was a sexy side to journalism in a sense back then, you know, before Watergate and investigative journalism, and it it was very appealing. And so with the book publishing company gone, I cast about for another occupation and became a freelance journalist. And in fact, did a lot of investigative work as well.
SPEAKER_00:What kind of investigative journalism did you work on?
SPEAKER_01:My beat was the nuclear industry, the commercial nuclear reactor industry. And at that time, Canada was very involved in trying to sell reactors around the world. And I spent several years trying to track down where the money was coming from and where it was ending up.
SPEAKER_00:That's starting to sound like the synopsis for a political mystery thriller.
SPEAKER_01:Yes. Yeah.
SPEAKER_00:And when and why did you become interested in AI? Was that in the early 80s or was that recent?
SPEAKER_01:It was in the 80s. The transformation for me was the uh dawn of desktop publishing, you know, which now seems sort of quaint, but at the time, you know, it it was about 1985, the Macintosh was a new computer, you know, came out with this graphical monitor that was a first, because up to then everything was text. And then they came out with the laser printer. And now what you saw on the screen you could print out in your own office. You know, we had WYSIWEEB, what you see is what you get. And from a as someone who had published books using an old verityper typesetting machine that had this spinning disk and a text-based screen. And my gosh, it was, you know, it was painstaking to use this little typesetting machine. Suddenly, this desktop publishing was just a you know a revelation to me. And uh I it brought me back into book publishing at the intersection with technology, and uh I'd always been sort of, I don't know, half nerd, half interested in literature and writing. And so the nerd side kind of took over. And I began uh working as a desktop publishing specialist for a large company in Canada. I designed a piece of software that could automatically typeset a book in push of a button, one of the very early programs that could take a word manuscript and click and it was typeset. Um and that at that point I was thinking, well, maybe there's this. I'd heard about artificial intelligence and I started studying it to see if it could have any applicability to the automated production of book pages. And the answer was, nope, it can't. It has no value for that whatsoever at this time. And so I've I shelved my interest for the time being.
SPEAKER_00:And then when did AI come back into your life?
SPEAKER_01:Around 2015, there were uh a smattering of companies that were entering into the book publishing space with various business propositions that were ostensibly AI enabled. Some of them, you know, AI is a messy topic, and you know, we throw it into this one big bucket. And in fact, it's got, you know, there's this about it, that about it, it's actually a you know an assemblage of many different technologies, machine learning, natural language processing, blah, blah, blah. Um, so different startups were trying to use certain AI tools either for you know language analysis or to automate some of the back office functions using machine learning to automate some of the marketing. And these early attempts you know were brave and interesting, but didn't go anywhere. And AI once again fell off the radar around 2017-2018. You know, some of them sort of hang hung around for a bit, but their their little heyday was gone. Then came you know, 2022, ChatGPT, November 2020 were coming up on the third anniversary, and that was like again, lights going on, going, ah, okay, they've figured this one out, and this is going to make a difference.
SPEAKER_00:Would you agree with the idea that we've been using AI for for years uh with programs such as Grammarly?
SPEAKER_01:Yes and no. I mean, we're we're definitely using it now with Grammarly and with words, features for grammar checking and spell checking. Ten years ago, those features were not really AI enabled. It was it was a primitive sort of machine automation that wouldn't qualify as AI as we understand it now. But now, when I talk to authors, for example, who say, well, I would never use AI, the one responder, though you don't really want to push it in their face, is to say, but you are, you know, you doth protest too much. You are using it. And the we're at this kind of interim stage now where the tools appear, they're marketed as explicit AI tools, but over time they'll kind of disappear into the background in the same way that spelling and grammar checking is no longer something where we go, because there was a time when we bought spell checking software, right? It was a separate software program you bought for$99 and attached, you know, to Word. And then you know, over time these things just become part of the tool set, it's almost invisible.
SPEAKER_00:Let's talk about headlines because we live in a world where headlines in news articles can cause havoc. And in the world of publishing and writing, it would be something to the effect of authors being cheated by AI. So have you seen this in regard to publishing and writing about AI? Are these kinds of headlines relevant or are they clickbait?
SPEAKER_01:The other headline we see a lot is Amazon full of AI generated books, right? That's another big concern. And, you know, the yep, there's a a bunch of AI-generated books on Amazon, not all of them labeled as such. And so it's there, but you know, the headline makes it sound like the death of publishing or the death of authoring, or you know, it'll no longer be possible to make a living as an author. And, you know, that it's it's uh it's terrifically overblown. But at the same time, you know, I don't want to diminish it out of existence. You know, there's underneath it, there are real concerns that need to be discussed, surfaced, acknowledged.
SPEAKER_00:Would you say then that it's more the legalities of the different factions of AI that we need to be tending to and focused on at this time? I guess the headline would read the legal issues of AI.
SPEAKER_01:There's a there's the the big question. I mean, that as your listeners will know, right, that there's been a lot in the in the press about the recent lawsuit with Anthropic that's settled with a$1.5 billion settlement. Um there's been a lot about you know other forms of theft and fair use and these crummy books and author, you know, books that steal authors' names and you have lots of legal issues. There's also legal issues around copyright, where the copyright office has very emphatically said if a book is 100% AI generated, it cannot be copyrighted. A human didn't write it, a machine did. Copyright exists for humans, not for machines. So you know, so so that when you go down the legal road, lots of obstacles, lots of issues. The you know, I point out to some that the litigation will go on for years. And so if you're waiting for the courts to give clarity here, you'll have to be very patient. Because even you know, this small anthropic lawsuit was really only about one tiny issue. It wasn't about the larger issues that we're all concerned about in terms of the future of the use of copyrighted materials for the training of AI. That's this, you know, a ginormous issue that's really going to take a long time to go through. So short answer in the end, you know, the the legal issues, it's good to be aware of them, but don't spend too much time fretting about them because it's a you know, it's an endless road that you know it's like tune back into this channel three or four years from now and we will tell you what happened. You really don't want to know the whole story for the next three or four years.
SPEAKER_00:That's an interesting answer. Thank you. So, how do we decide what is fact and what is fiction? You mentioned earlier that books written by an AI program cannot be copyrighted. But this to me brings on another question for an e-book platform. How do they know if something has been written by AI?
SPEAKER_01:And that is also a very, very important question. A couple things. Um, you and I have talked before about images versus text. So there's very different uh issues around how images are generated by AI, how they can be detected, what the implications are for images. So I'll put that in a separate box for a moment and just stay with text and books. And you know, on the one hand, you I think all of us have become sort of our own AI detectors, and you know, most of us now who read stuff have encountered things where we go, that's that's gotta have been written by AI. That is so badly written and so pedantic and so full of you know subheads and bullet points. It's like that came from a machine, and it probably did. But there's also some very sophisticated abilities to generate text now, and you can go through multiple iterations, you know, with an AI engine, where if you don't like the first result, you can point to things that are problematic, or you can load manuscripts in your style and ask it to create text in your style. So it becomes quite challenging. There are some software programs that purport to be AI detectors, and um many of them do a half-decent job, none of them do a perfect job. And so, you know, the best of them do a very good job, but then you run the scenario on this. It's like that it's the false positive problem that it's going to get it wrong sometimes, and either A tell you that a piece of text is not AI generated when it was, or worse, that a text was generated by AI when it wasn't. And we're seeing this in the schools where you know students who have English as a second or third language end up writing in a way that sounds like a machine because they're not conversant in English and they get dinged. That was written by AI. No, no, I did it, but my English isn't very good. So we you know you have to watch that detection. There are tools out there, but it's not foolproof.
SPEAKER_00:So let's talk about AI and book marketing, which in your book is on page 54. Production descriptions, keywords, book titles, target audience, and that word that I call the little big word, metadata. Um, for anyone who has delved into self-publishing, hiring specialists in this area is costly, a cost that until now had to be added to one's marketing and publishing budget. Does the use of AI in these areas need to be declared? And if so, who to and where? And are the bigger publishers now using AI for marketing?
SPEAKER_01:Yes, indeed. You know, marketing we we thought of as the so-called low-hanging fruit because it's pretty easy for AI to do certain things that are marketing functions, such as writing a press release, such as writing a description of a book. You know, AI is very good at that because it's a you know, you're writing to a template essentially, and you know, it's it's it's straightforward text. AI does a good job at that. Um, so in that sense, you are creating some text. Should you disclose, you know, AI wrote this press release? I don't think so. But you know, perhaps ethically, you know, in in some jurisdictions, someone would expect that. I don't think that's going to be common practice. The the when you think about marketing more broadly, um, it's so many different tasks. And you know, and the marketing specialists that are out there, some of whom are quite expensive, charge lots of money for all of these different tasks. And I'm sure many of your listeners have had uneven results working with them. Some of them are great. There's a lot of, what to say, um, inexperienced marketers portraying themselves as, you know, I'm gonna make your book sell. I had someone approach me the other day to uh who said, you know, I can do marketing for your AI book. And I said, Well, are you willing to do it on contingency? You know, I'll give you a portion of the money that we make with all of your efforts. No, I'm not willing to do that. Unsurprisingly. So anyway, I digress. Back to AI and marketing. The the large publishers, yes, are mostly starting to look at it. Where I've been recommending it to publishers in particular, and then this is a slightly different use case for an author. I think I recommend to publishers now that if you have a book that you're considering publishing, even before you make an offer to the author or the agent or whoever, throw that into the AI engine. Parenthetical here. There's ways to make sure that the engine does not take the content and put it into its training database. There are, you know, there's options that you that even you know that ordinary users can check where you you maintain the security of what you upload to these AI engines, which is a big worry that people have and a legitimate one. So, yes, you have to use that option. But having said that, um, assuming you can uh maintain the security of that manuscript, which you can, upload that manuscript into ChatGBT, or it has several competitors, Claude, Gemini, and so on, and start a conversation with that AI engine saying read this book. Tell me what this book is about, tell me who would want to read it, tell me what markets I could sell it in, tell me whether it's going to be whether it's an online, you know, better to be marketed online or through bookshops. And start this whole marketing conversation with the AI engine around a manuscript that you're considering. And I think that's an incredibly powerful tool that publishers are very reluctant to use, understandably, again, but this is where you know the AI, we're at that juncture between fear and utility. So then for an author, you know, you'd say the same thing. Um, take your manuscript and before you submit it to a publisher, upload it to one of these engines. Now start talking to it. How could I improve it? You know, is it you know, do character, you know, have I developed the characters? Are there any weak characters? Do plots, you know, themes disappear halfway through the book, blah, blah, blah, blah, blah. And what are the marketing possibilities? And help me write the inquiry letter for the agent, etc. So there's you know, a ton of things that I think are very, very powerful around marketing if you're willing to work with the tools.
SPEAKER_00:And for writers who are using Chat GPT for marketing, how do they ensure that their work is not being used as an AI training tool?
SPEAKER_01:There's there's two parts of the answer. One is that you know, there's ChatGPT, it gets all the press, but there's actually better systems out there. Claude, C L A U D E. Is far more conversant with the English language, oddly enough, than ChatGPT. It tends to express itself in a more better colloquial fashion and is more responsive to prompts to get it to change its style. I think Google Gemini is smarter than ChatGPT, uh, so I often use that. Each of them have different options in terms of the security of your text. Uh, Claude, as I recall, um, by default, does not use it in the training. You don't even have to sort of click a switch. With ChatGBT, you have to go into the options. I forget what the phrasing is exactly, but it's basically, you know, don't use my material to train your engine. And Google, as I recall, has a similar thing, but each of them do have that option because they're very cognizant of people's concerns, of authors' concerns, of creative concerns. Yeah, of course, it's ironic in a sense because they already stole so much, which is what's before the courts at this point. Um, you know, so that which I call the original sin is always this big cloud over all of them. And, you know, I'll say to authors, when you can use that switch and turn it off, I don't trust them. It's like, well, okay, I understand that. I don't trust, you know, it's like, okay, well, end of conversation.
SPEAKER_00:Yes.
SPEAKER_01:Well, it's understandable again. You know, you do the, I don't dismiss it.
SPEAKER_00:Yeah, it's that fear of not knowing. That's a little scary for all of us who are still trying to understand AI. That is the one aspect of publishing where you see AI as a vital part of the industry moving forward. And what do you see as warning signs that AI is quote unquote going rogue? I was listening to a program the other day on NPR, I think, and they were talking about AI sometimes going rogue. And as an example, I guess you could ask AI something like, okay, how many children have gone to the moon? And you may get the answer back as five. That's something that they would say, oh, that's going rogue, because no children have obviously gone to the moon. So what do you think about that?
SPEAKER_01:The the AI is rogue. Uh, technology is a real concern, and it's one that I share. Uh, there's enough instances of um these AI tools showing what we call agency, i.e., you know, sort of initiative of their own. And of course, it's not true initiative, uh, in as much as you know, they're not in the back room thinking and scheming, but the way they're actually programmed is for their own survival, in a sense. And it's it's amazing how scary it is to remember and watch again 2001 a space odyssey with how the rogue computer is very true. You know, it it is that was truly press-in. Because the computers, the AI systems in the computers today are programmed similarly. It's not so much if you try and shut them down, they'll kill you, but you know, if you try to talk them out of existence, they'll fight back and say, no, you know, I want to continue this conversation, or you know, no, I'm I'm I'm going to continue working on that problem. And so we see this sort of glimmer of agency that's very, very disturbing. Uh and when that's then extended into thinking about AI being used by terrorists and you know by countries that engage in terrorist activities, it's terrifying. Um, so that's at the you know, the extreme. Five children on the moon is an you know, sort of an interesting small instance of how these things go wrong. And if someone was you know depending on that answer, or where we're seeing it more often is like with medical advice, if you get you know the wrong advice, there's there are real-world immediate implications of that. And so trusting the machine blindly is a mistake. But how do we create usable day-to-day guidelines that protect us and society? That's still very much open for consideration and debate.
SPEAKER_00:Would it be a fair enough comment to say if you were going to ask the question, how many children have gone to the moon? Instead, you could ask, could you please show me documentation on whether children have gone to the moon from a reputable newspaper or scientific journal?
SPEAKER_01:Well, very much so. Though some of your listeners will have also heard of cases. I've had this happen too, where you'll ask it, you know, what is this based on? It will make up story, you know, it'll make up articles, authors, it'll give you links and so on to things that don't exist. That's less and less common. That's one thing that most people understandably don't uh appreciate is how quickly this stuff is moving. And you know, I've been you know working with it now these three years, and I'm you know, stunned, astonished, blown away, even you know, based on what you thought might be possible two years ago, what you thought might be possible one year ago, what I thought might be possible six months ago, what I can do today, you know, is it's just so incredible the speed at which it's moving forward. So things like hallucinations, which were a really big problem, are a really small problem today. And as you say, you know, asking the machine to sort of to tell you why it thinks what it thinks, you can do that. And and even without asking it, it will often show you why it thinks what it thinks. And when you spot check that, it's good. You know, I'm I'm not finding these kind of glaring errors that were absolutely a part of it even a year ago. I'm not seeing that to anything like the same extent today, but still, I don't trust it 100%.
SPEAKER_00:That's good to know. Having spoken with translators and small publishers on the show who specialize in translating books to English, they are concerned about AI in this field. I remember listening to a conversation you had. It might have been with Joanna Penn, and you were saying that when you do a translation through AI, you translate the book from English to I don't say French, and then you actually get AI to translate the French version back into English. What I'm hearing from the small publishers who specialize in translated books, the first issue they see is in the emotive abilities of AI in translation. What are your thoughts?
SPEAKER_01:My book, you know, the AI and Revolution in Book Publishing came out first, gosh, it's only a year ago, July, is it 14 months, whatever. And I published it in 32 languages using this lean pub platform that I use. And it was automated. You know, we took the manuscript, it went through ChatGBT, and three hours later emerged in 31 additional languages, including the cover redesigned in that language and all the marketing texts rewritten in in that language, including many of the Asian languages, most of the European languages. And I spot checked those translations and they were adequate, not very good. This is part of how much things have changed. Three months ago, I guess it was in May, there was an instance where I had read about this book that's called Hypnocracy. It's written and published in Italy, and I read about it in Wired magazine. I thought it sounded fascinating. I went online, there was no English version available. So I went to the Italian publishers website, purchased legally for 10 euros the book, the ebook, um, uploaded that to Claude, asked for an English translation, which it produced 15 minutes later. I was going the next day to a meeting of the Association of Catholic Publishers, as it turned out, and was reading it on the plane to Chicago, and it was good. It was very, very good. And you know, to the extent, would a human translator have done better? No doubt. Um you know, was it perfect? I don't know. I'm not, I don't speak Italian. Um, was it fluid and fluent in seeming? Yes. There were were the parts where it was weird? Yes. You know, that if I'd had the energy, I would have spot checked some of it, and I'm sure I would have found some weirdness in it. You know, what I'm describing there is a practical use case where that book wasn't available in English, it still isn't available in English. I wanted to read it. The publisher got their 10 euros, the author got his royalty. Um, so was that a bad thing? Um, you know, the the issue around translation is a hugely sensitive one, and I respect how hugely sensitive it is and why it is. I've written a number of articles on AI translation that acknowledge, you know, translators can argue over a word. They can easily argue over a sentence and easily argue over a paragraph, and for good reason, right? And a machine doesn't understand those subtleties or nuances that will be lost in translation. And that's unfortunate if this becomes pervasive, where I think AI and translation has utility is on those books that would not be translated otherwise. And that's most books and most languages.
SPEAKER_00:And that's a sad thought on its own, isn't it? And that's why I admire the indie publishers who are specializing in uh translating books from other languages into English. But while we're speaking about translation, rather than an AI replacing human translators, do you see a way forward where AI supports human translators?
SPEAKER_01:Clearly there is. I mean, you know, it's it's not an answer for the livelihood of translators, but the obvious workflow for a translator now is to take this book that they've been asked to translate, let AI do the whole first run through, and then go in and improve that. And the task of translation would probably be the time required would be reduced by 90%. And because they would still be taking responsibility word by word by word, you know, it they would still be able to produce to their maximum ability, but they would save a ton of time. Um, but that's anathema, understandably, to translators. Um I think, you know, again, it's not about substituting for the work they're doing today. You know, it it's those books are books that are wonderful enough that uh, you know, another publisher says, I think I can find a market for this book in my country and I'm willing to spend money to do that. It's for all those books where they only sold 500 in the United States. There's no way that Italy is going to give you$10,000 to translate your book.
SPEAKER_00:Yeah, I've heard different themes on this issue. For example, an author is with a large publishing company and he's done a couple of books in a series. Um, the second one sold maybe 50,000, 60,000 copies, but the publisher has decided not to do the third book or you know, the next one in the series. And that's really heartbreaking for the author and for all the readers waiting for the next one in the series.
SPEAKER_01:And wouldn't why would they then not do it themselves using an AI tool and sell it on their website? Why would they not?
SPEAKER_00:Yes, good idea. The only thing I suggest for any authors listening is that that's another good reason to make sure a lawyer checks your contract with the publishing company because you want to make sure that if something like that happens, then you can actually go and publish it yourself in a different language. Uh yeah, that's a tough one.
SPEAKER_01:Yeah. The the publishing contracts are a real minefield in both in terms of the contracts that people have signed and the rights that they may have given away that they didn't realize at the time they were giving away because they weren't didn't seem terribly important. And moving forward, what what should be in a contract to acknowledge this changing AI world? A simple example is that most authors don't realize they gave their publishers 50% of any revenue that's earned from AI uses of their book, because it was in a sort of all-encompassing clause that said in any other digital utilization, blah, blah, blah, AI was never mentioned. And the author's guild, I think bravely and appropriately, although I'd be slammed for saying this by some publishers, the author's guild says, you know, that right for AI rights was never explicitly offered by an author or publisher. You can't just assume you got them because that's the language that you threw in that stands up in a court of law. It's not right. You know, you never, you didn't know about AI. You didn't know this was a revenue opportunity. And I feel very much with Authors Guild on that one, where I think that, you know, I understand that publishers invest in books and you know, subsidiary rights are very important and all of those things. But you never knew there was going to be this thing called AI licensing, and now you want to keep half of it.
SPEAKER_00:Yeah, I'd be interested to see a few contracts that have been written recently to see how they word this whole AI translating idea. In the chapter titled Concerns and Risks Surrounding AI, which is on page 72, you write about copyright infringement. Quote, was it legal to ingest all of this text to help build billion-dollar AI companies without any compensations to the authors? End quote. We actually spoke about this earlier, but I'd love it if you can speak to both sides of this argument.
SPEAKER_01:This is the one that's you know before the courts now. And the anthropic case that I mentioned earlier that, you know, would got the big headlines over the last couple of weeks. People are misinterpreting the outcome of that settlement, the 1.5 billion, as being a win for authors against AI. It's not. All that was decided by that court case was that pirating books, you know, because it was not about the training of the AI engine, it was the fact that they took the books from a pirate website. That was the illegality. It was the piracy, not the training. The judge in this particular case says that it's going to be okay if you if a publisher or if an AI company obtains a book legally, they can, without further compensation to the author or publisher, use that to train an AI engine. This was the first of what will be many decisions, so it's not the final word on that, but it's very concerning to authors who were hoping for revenue from licensing their work to AI companies. There's that still might happen because it might be a convenient way for AI companies to obtain books, but they're not going to pay a whole lot of money for it. I've been working with some publishers who've been off, you know, given some offers, and the prices have gone from 5,000 down to 50 bucks, kind of thing, as a result of these recent decisions. So, you know, so that's the legality of it. What I you know try and stress in the book is there's a difference between the ethics and the and the law. And even if you can get away with something legally, that doesn't mean necessarily that that's ethically correct. And that's where I feel there's there's no way to ethically endorse what the AI companies did. They took valuable content that you know millions of authors had created painstakingly over a long period of time and without requesting permission or offering compensation, use that to build these multi-billion dollar, you know, large large language models. That's not ethically correct. You know, any two-year-old can tell you that. The courts may decide it's legal because of fair use. And and, you know, in in favor of the AI companies, we needed that material. Yes, you did. We could only make these fantastic AI tools possible if we got hold of all those words. Yes, you did. And I think it's a great thing that you made these tools, but you were you have a lot of money. The authors deserve compensation.
SPEAKER_00:And just to be clear, a pirated book is a book or manuscript that is copied without permission from either the author or the publishing company, and that is illegal.
SPEAKER_01:I always say to authors, it's Tim O'Reilly had said this in a different way some years ago, but pirating is actually actually a compliment. You look at it as a compliment. It means that someone thinks your book is good enough that they want to steal it. And it's like, well, but then I lose all this revenue. Well, I think in the worst case scenario, 5% of your sales go out through pirates. And what you're seeing, you know, when you find your book on a pirate site, it means someone knows that there was enough opportunity because your book is good enough and doing well enough that they thought they could make a few bucks on the side, stealing it and selling it on the side. So don't, you know, don't lose sleep over that. I've talked to a lot of authors that you know, they'll they'll Google their book and there it is on three or four different pirate sites. And it's like, try downloading it. Half the sites put it up there and hope to sort of lure you into a subscription. The book doesn't even exist, it's not even there. And then on those few sites where it actually is there, it's a scam site, and you end up with a virus on your computer, and you know, it's just it's a whole mess. This it's overblown as a worry for authors. So I again you can see where it someone when they see that online and it's illegal. Yes, of course, that's upsetting.
SPEAKER_00:Yes, it is. Thad, in the subheading of your book, uh it says a consumption. Guide to navigating artificial intelligence for writers and publishers. Were they the main two groups that you wrote this for?
SPEAKER_01:I very much had in mind both authors and publishers. A lot of my work is with publishers directly, but I'm I I also work with authors. Some of my largest consulting clients are authors. I'm an author. I've written, you know, 18 books or something like that. I always lose track. And so, you know, my my mind is sort of this schizo, am I an author, am I a publisher? And I'm both. And so I always try to wear both caps when I'm writing and when I'm writing on my blog. When I'm working with publishers, I always say to them, you guys sometimes forget that if you're talking about the supply chain of what you do, they often sort of talk, well, acquisitions is the beginning of the publishing process. No, it isn't. The author is the beginning of the process. Somebody who got an idea and decided to devote, you know, many hundreds of hours to it, that's the beginning of the process. And too often, you know, you forget about that. So that by way of saying, you know, the two are together integrally. Uh, you can't talk about one without the other. An author without a way to get their book out there, call it a publisher or self-published, how whatever it is, is without their other half. And a publisher without authors is nowhere. So the two work together, though their concerns are not always identical.
SPEAKER_00:Is this book self-published? And do you publish it through Ingram?
SPEAKER_01:The interesting, I think, thing that I did with this book was use this platform Lean Pub, L-E-A-N-P-U-B, and it's the best platform for a book that needs to be constantly updated. When I thought to do a book, I knew there was no point in doing a proposal, getting to a publisher, you know, waiting three to six months, you know, to get the book accepted or not, a year, you know, and so on and so on. So it was like I'm gonna have to self-publish and I'm gonna have to be able to update it continually. I've updated the book oh, 14 times or so, but then comes the question of how do you get it out to the public in its updated version. If they buy it on Lean Pub, they always get the latest version because I it gets updated like that. For me to then get it out in paperback form, like the one you've got, I have to then use Ingram and Amazon. So at various points, I'll download the file off of Lean Pub, upload the file to Ingram and to Amazon, which used to seem like such an easy thing. You know, oh wow, I can publish a book just by uploading. Now in this age where books are changing like that, suddenly that becomes the most cumbersome part of the process, waiting for Amazon to do their review and da-da-da-da-da-da. So I end up not putting the totally updated latest versions on Amazon. But the one I sent you is my major update done in May of this year, a few months ago. I'm going to do another major update this fall, and I'll update all of the files. But people who want the, you know, sort of up to the minute version, they can get it on the Lean Pub platform as an ebook, not in print.
SPEAKER_00:I learned a lot reading this book. I wasn't scared of what I read either. I am always like a sponge, I want to learn as much as I can about new ideas and to understand them. And the way you've written this book makes that very clear and it's easy to read. I just learned so much from it. So thank you.
SPEAKER_01:Thank you, Mandy. I appreciate that. It was great to see the yellow tags and it warms the heart of an author.
SPEAKER_00:I think I'm the queen of sticky notes. The other thing I want to mention about this book is it is short. It's an easy-to-read short book that is going to give you a lot of information.
SPEAKER_01:I wrote it to be short because I knew people didn't want to read 100,000 words, or you know, it's it's only 150 or but it's only 20,000 words. It's spaced in a way that you have to make the book look big enough to be published, but it's only 20,000 words, which is not a heck of a lot. And then I wrote it very much for the beginner. You know, I I'd assume no knowledge other than AI means artificial intelligence. If you've got that part right, you can read the book. It everything is explained in there without again, you know, I didn't want to make it a children's version either. So I've written it for an intelligent but you know, not uh not conversion with AI type of reader.
SPEAKER_00:And if you choose to read it in the A-book format, you can actually just click on all the links.
SPEAKER_01:Yeah, I'm I'm a big hyperlink writer. I'm always concerned with a technical book about credibility. You make a statement and people, well, where is that based on? So I hyperlink it, and there are, you know, there's over, well, it's not 170 or something hyperlinks in the darn thing, which is overkill in a sense. But what you people find is I do my research, I do, you know, I do check these things, I do try and get the latest source. Uh, because I nothing is worse for a nonfiction author of a technical book than to have someone point out that you've gotten some basic fact totally wrong, undermining the credibility of the whole venture. Now tell me, Mandy, you have howl on the bookshelf behind you. What why have you got it there?
SPEAKER_00:Because I love it.
SPEAKER_01:I do too.
SPEAKER_00:And I found this copy when I was on a road trip from Los Angeles up to Colorado, and it was in a little dinky thrift store somewhere in the middle of nowhere. And uh it was in a box of books, and I found it, and I thought, oh my goodness. I think it was like 25 cents or something. So yeah, that one's prominent on my bookcase there.
SPEAKER_01:It's a beautiful edition. And you know, there's over a million copies in print now. You think of a single book of poetry with a million plus copies in print. Now that's something fantastic. And you reread how all it's it's just it's such a powerful poem. It's so powerful. It stuns me how much strength is still in this in those words.
SPEAKER_00:When I hear people say, Oh, nobody reads anymore. I just want to say, you know what? Yes, they do. You know, uh, have you been on social media? Or maybe you're scared of being on social media, but people are buying and reading books, or they're going to their libraries. Uh, you know, I think there's some misinformation there. Um maybe it's just that my community is a book reading tribe. I don't know, but I think people are still reading.
SPEAKER_01:They certainly are. Yeah, I'm in San Francisco, so our City Lights bookstore is downtown here, and and where that book was published, and they have the one of the best poetry sections, probably of any bookstore. Well, there's probably one in New York that's better, whatever, but it's a beautiful on the second floor, it's just the poetry room, you know, with thousands of poetry books and many that they publish. So it's fantastic.
SPEAKER_00:You've reminded me of something else I hear about reading, and that is people don't want to read anymore because they have short attention spans. And I want to say to them, have you thought about reading poetry or essays? This weekend I sat down and it took me a couple of hours, but I read this book and it's a micro memoir. What I love about this style of book is uh that a publisher has taken a risk on publishing the book for a start. That's number one, because uh it might have just a thought on one page, the author has put a thought down, then there might be a short essay, then there might be a a poem and another essay, another thought. I found it just wonderful to read. I think it's an exciting time for writing and publishing and for indie publishers.
SPEAKER_01:I do write in the book about the challenges that publishing faces and the industry as a whole, you know, is not growing. But you know, just because it's not growing doesn't mean it's disappearing, right? There's a you know, vast number of people who read a vast number of books, and AI is not gonna stop that, and video is not gonna stop that, and so on. You know, if it was a flawed format, but it's not, people love it as it is, and they love it in print, and they love audiobooks, and you know, it's like that's not threatened when people have that much passion for, you know, of course, we're gonna be fine.
SPEAKER_00:And indie bookshops are thriving too. I mean, there was a lot that closed during the pandemic, but my goodness, there are a lot that opened during the pandemic as well and have survived.
SPEAKER_01:Yes, indeed.
SPEAKER_00:Do you know Green Apple books up in San Francisco?
SPEAKER_01:Of course, yeah. I mean, I was on a panel with the owner, co-owner the other day. We had an event here in San Francisco, and Pete Mulvahill was on the panel, and he's such an interesting part player in the book publishing industry. You know, he gives the he sees the industry through a bookseller's eyes, and those are different than the on the panel. We had an agent, a publisher, me, the futurist, and Pete, the bookseller. And we were asked questions, and each of us tried to answer those questions, each through our own lens. And it's it's fascinating. The bookseller's lens is a really interesting way of looking at the world. And my first job was as a bookseller, and I argued in the panel a few weeks ago, I've often said, if you want to work in publishing, you've got to start in a bookstore. You've got to start your career in a bookstore because you have to have that tangible experience of someone coming up to the cash desk and saying, I'd like this book, please. I'm willing to give you money for this book. And that, you know, that's a magical moment that you can lose sight of when you're working in the back room at a publishing office and forgetting that out there there are readers who are going to be going into their local store and having, you know, making that connection, that magical connection. And that's what it's all about.
SPEAKER_00:Yes, absolutely. Thad, it's been great chatting with you, and I absolutely recommend your book, The AI Revolution in Book Publishing, a concise guide to navigating artificial intelligence for writers and publishers. I wish you all the best and thanks for being on the Bookshop Podcast.
SPEAKER_01:Thank you so much, Mandy.
SPEAKER_00:You've been listening to my conversation with Fad McKilroy, author of the AI Revolution Book Publishing, a concise guide to navigating artificial intelligence for writers and publishers. To help us overreach more people, please share episodes with friends and family and on social media. And remember to subscribe and leave a review wherever you listen to this podcast. To find out more about the Bookshop Podcast, go to the Bookshop Podcast.com. And make sure to subscribe and leave a review wherever you listen to the show. You can also follow me at Mandy JacksonBeverly on Instagram and Facebook, and on YouTube at the Bookshop Podcast. If you have a favorite indie bookshop that you'd like to suggest we have on the podcast, I'd love to hear from you via the contact form at thebookshoppodcast.com. The Bookshop Podcast is written and produced by me, Mandy Jackson Beverly. Theme music provided by Brian Beverly and my personal assistant is Kaylee DeGinger. Thanks for listening, and I'll see you next time.