18 - The AI Revolution in Publishing with Thad McIlroy from The Future of Publishing

In this episode of Brave New Bookshelf, we sit down with Thad McIlroy from The Future of Publishing to discuss how artificial intelligence is transforming the publishing industry. From continuously updated books and global translations to AI-powered marketing tools, Thad shares his insights on how authors and publishers can leverage AI to streamline workflows and expand their reach. We also dive into the challenges of effective prompting, the nuances of AI-driven translation, and exciting new tools like Veristage and Shimmr. Visit our website https://bravenewbookshelf.com to view the full episode notes, links and apps mentioned in the episode, and the full transcript.

[00:00:00] Welcome to Brave New Bookshelf, a podcast that explores the fascinating intersection of AI and authorship. Join hosts, Steph Pajonas and Danica Favorite, as they dive into thought provoking discussions, debunk myths, and highlight the transformative role of AI in the publishing industry. 

**Steph Pajonas:** Hello, everyone. Welcome back to the Brave New Bookshelf. I'm Steph Pajonas, CTO and COO of the Future Fiction Academy, where we teach authors how to use AI in any part of their business. We've definitely been doing a lot of that lately, which I'm really, really grateful for. This fall is looking to be a lot of fun, especially with all the new tools that are out.

Lots of cool new video and audio stuff going on, music. I mean, You can't go wrong with any of that kind of stuff. So we're looking forward to playing with them, right? And I'm here with my lovely co host Danica Favorite. How are you doing Danica?

**Danica Favorite:** I'm Danica Favorite. I am the community manager at [00:01:00] PublishDrive and we also have a busy and exciting fall. We launched our metadata generator over the summer, which has been fantastic for authors who don't want to do all the prompting and all of that stuff. They can just pop their book in and get amazing metadata and book descriptions for their books. And we've just launched a free plan. For anyone to use pretty much our entire ecosystem for free, which is also pretty amazing. So even distribution now, you've got the opportunity to enter into that whole wide ecosystem. So, That is also pretty super cool. Very excited about that. 

And today, I am even more excited because we have Thad McIlroy for our guest. I've been following Thad online for quite some time now, actually, because he's got some really great perspectives on AI and publishing. When Steph and I sat down and made up our dream guest list, I have to say, Thad was on the list. 

We read his stuff. When we talk [00:02:00] about following AI trends and what's happening in AI and all of the important business y stuff, we really rely a lot on Thad and his information and he's got a new book out, so I was like, okay, this is perfect timing to see if we could get him on. I heard him speak at the IBPA PubU conference where I loved a lot of what he had to say just about coming into the balanced idea where Steph and I are like, Hey, you don't have to do it all. You don't have to jump in. With your flags waving your balloons and cheering and you also don't need to have your pitchforks out either. 

And so that is definitely why I was like, let's hear about him his new book AI stuff. And fortunately, he was gracious enough to accept our invitation. So without further ado, I'm going to introduce that McIlroy. 

**Thad McIlroy:** Thank you Danica. Thank you Steph for having me on your show. Much appreciated. Always thrilled to talk with anyone in and around authoring and publishing about [00:03:00] technology, AI, sure, and lots of other things. I've been working for some years now at the intersection of publishing and technology. A lot of the work I did during the last decade had to do with metadata.

And I co authored a book, The Metadata Handbook, with René Register. And that was, that got me very deep in what, as you folks know, is a deep, deep rabbit hole. There's a superficial metadata, and then there's all that other stuff underneath it. So I spent a lot of time trying to understand what's underneath metadata.

I came into all of this through publishing and through authoring. My very first job was in a bookstore in Toronto where I grew up. And I started my own publishing company when I was still in my 20s and managed to bankrupt that several years later and then made a segue into journalism, into authoring and had my first books come out in the 80s, in fact. And then got back in the late 80s when desktop publishing came out, that's when computer [00:04:00] based technology, as friendly as it was now intersected with publishing, which had been a very non tech industry.

And so I carved out a space there at that intersection between tech and traditional publishing that my roots go deep in the traditional, but my fascination is around the technologies. And that's what, as, as I've moved forward, that's brought me to this intersection with AI and publishing, which I actually started studying in the eighties and started looking at this potential applicability in the late eighties, and it didn't have any at that time.

It's only really been the last two years that it's suddenly, aha, now we've got something here we can use. 

**Danica Favorite:** Yeah, I love that. And I love that you were looking at it. Even back in the late 80s. I have a friend who was in the tech industry. Nothing to do with publishing, but he, he was telling me, Oh yeah, we were looking at the late eighties, blah, blah, blah. Oh yeah. I started working on LLMs in the early nineties and I'm like, wait, what? I think it's really interesting how long this tech [00:05:00] has been around because like you said, we've only recently found the applications for authors, but it's certainly been there and certainly been something that's been here and people have been working on for years. So, I love seeing how people can now use it for publishing.

I'm not nearly as techy as Steph. Like Steph is super awesome in the tech stuff. And I've just been, I don't understand this, but I'm fascinated by it. And 

**Steph Pajonas:** was going to say that way back then we were using Quark Express, to like layout stuff for,

**Thad McIlroy:** Yeah.

**Steph Pajonas:** for print, right? I don't even know if that's even around anymore. I think InDesign has totally usurped it in the business, but yeah, 

**Thad McIlroy:** The InDesign has usurped it. Quark is actually still out there but it's used at, at corporate level, not at the kind of more creative design level, though it's still an extant company. And as part of that tradition, by the very first book I typeset on a Macintosh, I used Microsoft Word version [00:06:00] 1. 05 on a computer that didn't have a hard drive.

We had to use floppy disks to capture content as we went along. Had to manually have a hyphenate and very deliberately place the running headers and so on. So that's come along 

**Steph Pajonas:** a long 

**Danica Favorite:** really has. I, I remember I was, I worked on our school magazine and There were things we had to still typeset by hand, picking out the little letters and putting it in.

Wow, we have just come such a long way. It's really amazing.

So let's talk about a little bit about your background and you have a new book about AI and publishing out. And I'd love to know a little bit about your book and how this AI and publishing book came about, particularly because one of the things I noticed is you could actually purchase it directly from you. And get all the updates to the book as you update it, which what an amazing use of technology there. So tell us about your book and tell us a little bit about the technology and [00:07:00] how that all came about. Because you do some really super cool things with that. And I know our audience is going to be really curious about all of it.

**Thad McIlroy:** Yeah, it's interesting to me that, the book is two things. It's a book about AI and writing and publishing. It's also its own story, its own process as a story in itself that I think is as interesting in some ways more interesting than the book itself. When I decided that I would do a book on AI, I mean, it became so obvious to me instantly that, looking at a traditional publishing route, even a traditional self publishing route was potentially absurd.

I needed to have a way to constantly update the book. Because at any point, it's so obvious at the point that the book is printed it's out of date, and they say that about any kind of any a lot of books people say that I don't think until the advent of AI it's ever been so true. Just in the last week I was hearing on the call yesterday. There's a new feature in ChatGPT that I [00:08:00] wasn't aware of. I can't remember what the feature was. I have to check my notes on that, which I have many, but you guys are familiar, right? This stuff is changing at a breathtaking pace.

And so, the book that I have printed that's, on Amazon right now, I'm already, in the midst of creating a new update. So how do I create the new update? You can't buy it from me per se. I went to a platform called LeanPub. And I think more self published authors would be delighted to know more about LeanPub.

Their motto is publish early, publish often, and the idea is it's specifically optimized for books that are a work in progress, a constant work in progress. And so I first offered the book back when I saw you at IBPA, it was 75 percent complete. And so I put the book up on LeanPub, making it very clear, there's a work in progress and I made it free at that point because it seemed fatuous to charge money for something that [00:09:00] was still a work in progress.

And I was doing that also because I could get feedback from early readers. So that was fun. Then I went to a 90 percent version, still made it free. Then in July, a month ago, I went to the 100 percent version. The funny thing there, it's not the 100 percent version, it's the 99 percent version, because the book is still being updated. I've done six or seven updates to it since it became so called 100%.

The challenge was if you say, I'm going to sell you this book that's not complete, even at 99%, you're signaling Not Complete. And so LeanPub said, you got to say it's a hundred and then explain in the fine print that if you buy it on LeanPub, you will be entitled for life to the updates as they're published. So it's a nifty platform that way. You hit the publish button, with Abandoned because it takes, no time at all.

Where with Amazon or Ingram, it's not like it takes a long time. But two or three days to get the book back into the channel, and [00:10:00] LeanPub you can be constantly updating. You know with Amazon I can make I guess the Kindle updated someone can download the latest version, but it's a cumbersome way to put a constantly updated book out there.

So that's part of it. And then I'll stop blithering on here for in a moment. But the other real big aspect of working with LeanPub is they've introduced this translation feature, global translation. And for 500 dollars, the book appeared more or less instantly in 31 other languages. 31 languages beyond the English version. And all the key languages, all of the European languages, all of the Asian languages. And they populate the site with all 31 languages. And so when you go to the English page, you can see that it's available, if you're anywhere from around the world, you can say, Oh, it's also available in Romanian.

That's great. Same price, 9. 99. And if you click that, you go to the Romanian page where the [00:11:00] description of the book is in Romanian. And you find out that if you want to get both the English and the Romanian version, it's 14. 99, not 19. 99. So you get it for one and a half times cost, because I felt that, and this is another complex issue, but it strikes me that the only people who are going to come to the translations at this point are people who have English as a second language, would prefer to read the book in their first language, but might want to double check aspects of it in the English edition.

**Steph Pajonas:** Wow. I love this because you just press a button and you got all of those translations, right? And I'm going to guess that they're using an AI translator on the backend, whether that is a LLM or it's 

**Thad McIlroy:** It's 

**Steph Pajonas:** or

whatever it might be, chat

**Thad McIlroy:** is ChatGPT. 

**Steph Pajonas:** Excellent. Excellent. So that's great because now you've just opened up your market to worldwide and given people options . Do they offer any audio?

**Thad McIlroy:** Well, that's good you asked that. So, what you see here is the result of a [00:12:00] startling realization. The audio I, I'm using Eleven Labs and I'm using Google. To do automated foreign language audio 

**Steph Pajonas:** Hmm. 

**Thad McIlroy:** 11 labs can actually work with the English and do the translation. But Google, you take the translated EPUB that was prepared by LeanPub, publish that to Google, and then they'll do for free a French language. They have about four languages that they can do on Google, four or five languages. 

And so I tested that with the French language and got that out there getting reasonable feedback on that. I did the German language through 11 labs. I'm getting some interesting feedback on that. What I realized is something that, I'm sure many of the people in your audience who've been doing audio book versions of their self published book.

Well, it's not as simple as just tossing the, the EPUB into the automated audio engine. All these little tags are places where Oh, yeah, you're going to have to [00:13:00] explain what it actually shows on the page for people to understand what's going on in the audio book version. And so now I'm going through and writing a new script for the audio book, which I will use 11 labs to do an automated generated audio book from it.

But it's then going to become the master script that is appropriate for an audio book audience.

**Steph Pajonas:** Yeah, correct. That's great. Yeah, because Well, we're here in video. We also distribute it via just audio, obviously through the podcast apps, but then if we're holding up things or whatever, people can't see that. So we, we do have to like narrate it a little bit, 

**Thad McIlroy:** Oh,

**Steph Pajonas:** same same for books, especially nonfiction books. So 

you know, we got to do 

**Thad McIlroy:** holding up 

**Steph Pajonas:** so great. Thank you.

**Thad McIlroy:** a book full of little yellow tabs for those of you on the audio audience. 

**Steph Pajonas:** putting, you're keeping Post it notes in business is what you're doing.

**Danica Favorite:** Absolutely.

I feel like, as [00:14:00] much as we have technology I'm looking at my desk right now and it is literally covered in post it notes. So I'm like, this is the, the thing that will never go away. I don't think.

**Thad McIlroy:** No, no.

**Danica Favorite:** Oh, that's fantastic. And I, I love hearing that you're doing this AI book about AI and you're using so much AI. with it because I think that is really good for people to realize that these are amazing tools. I love the idea of someone who is not a native speaker of English who can have the book in English, but then when they want to check something in their own language, they can go back and forth, which is really nice.

And I say that as somebody at PublishDrive, we have a very international team. And so sometimes we'll talk about things and it just doesn't land with the rest of the team members because of the language barrier. And so being able to go back and forth between the languages is so helpful.

**Thad McIlroy:** AI and translation is one of the most fascinating topics [00:15:00] in AI today. I do this series of webinars for the book industry, study group free webinars. We did one in June that's worth looking up. It's on YouTube, a first look at the issues around AI and book translation. The issues are, as with so many things, far more profound underneath the surface than the obvious, click a button, book, translate it.

There's so much nuance going on here and there's so much history of pre the current generation of AI, there was a very well established machine translation, MT industry with many, many hundreds of thousands of practitioners around the world. And it's, this industry is going through its own very radical transformation into LLMs as the basis for their translation.

Is it good enough? In what cases, of idiomatic expression, as you're suggesting, Danica, does that translate with a push of a button? I think ChatGPT is actually excellent at idiomatic expression. [00:16:00] It's not a fact based engine. It's a vernacular. And so for much of the, free flowing languages around the world, Chat GPT is actually a pretty good tool.

**Danica Favorite:** Yeah, I like that. And I like the point that you're saying about chat GPT and vernacular, because as Steph and I have talked about a lot, and some of our guests have mentioned, different models are important for different things. And so, when you're like, okay, what am I trying to get? The AI to do you really do have to think about, okay, what are its capabilities? What is it good at?

Because there are some things if you throw into chat GPT, you're gonna get terrible answer And if you throw it into Claude, you're gonna get a better answer and it just really depends on which tool you're using. So I think it's good for readers to know. Readers, I would say readers because I'm an author and I always default to that language but for our listeners to know is that if you're looking at translations, Chat GPT is probably one of the best tools because it gives you [00:17:00] that more specific in the languages.

So I like that a lot. I appreciate that distinction.

**Steph Pajonas:** There was a time maybe 20 years ago where I was in tech and we were looking at localizations of software. Not just having your software in English, but having it in Russian and Chinese and Japanese and whatnot. And that, business of localization for softwares was so fraught with, was fraught with problems because there weren't always people there to translate those things and then implement them.

And so there would be large swaths of the earth that were left out because they couldn't have technology in their language. Now I feel like we're finally able to reach so many people in their native language because of AI. And I'm really, I'm really excited about that. Really excited. Especially as I love to learn language. I'm learning six, six, seven, maybe languages right now at a time, [00:18:00] which is why Duolingo is always on my desk over here.

He's wearing some sunglasses today. But yeah, I just really, really love the fact that AI is giving us a chance to expand on language and expand markets as well. So it's exciting.

**Thad McIlroy:** And what you're talking about there with the software localization, I think points to the Achilles heel of Chat GPT, with a software manual or a manual for your stove, if it gets, push this button, wrong button, That's a problem. 

If I say, stoves are full of controls, some of which are now digital controls, many of which are still, analog controls, making a sweeping statement like that, ChatGPT is probably going to get the sense of that pretty accurately.

But if it says, turn this button in that direction, and that turns out to be high, not low. That can be, that has implications. And so, the limitation of ChatGPT as a fact engine and as a fact translation engine are, are pretty serious or pretty real important. But, for [00:19:00] writers of fiction, I think it's there today.

I really do think it's there today. Unless you're Tolstoy in which case, you're at that level of sort of literary precision. There's still going to be a lot of pushback on that, but, if I'd written, I don't know what, I don't write fiction, I only write nonfiction, but if I was not that worried about every single word, I thought I would be really thrilled to use these engines today.

**Danica Favorite:** Yeah, I think, and I think it's so funny that you said, oh, yeah, it's there today because yeah, I remember a year ago when Steph and I were just friends, chatting about stuff and AI technology. We're both like, yeah, it's, it's there, but not really. And as we keep seeing, and which I think is great about you constantly updating the book, it's like, okay, no, all the stuff that I said it couldn't do a year ago now it can do.

And it's, it's pretty amazing how fast it's changing.

**Thad McIlroy:** What, what do you tell your listeners about ChatGPT's facility as a grammar tool?

**Steph Pajonas:** Well, [00:20:00] I tell people, students, people I know that it's not a good proofreader. I mean, if you've tried, I don't know, I've run it through its paces, been like, I'll take out a comma here, I'll use a homonym here that's wrong, and I'll basically feed it a document with lots of errors that I know are wrong, right?

And it sometimes gets like, 60 percent of them. And then I'll say, look again. I'm sure you can find more. And often it just doesn't, it's not very good at it. So I find that I tell people, Chat GPT, any one of these LLMs are good for text generation. Like they're going to help you get from this idea to your next idea.

They're going to help you flesh out your ideas. They're going to keep you writing. But then when it comes time to sit down and work on, grammar and syntax and those sorts of things, that's a great time to move over to a specialized tool like Grammarly or ProWritingAid, Autocrit, any one of these that they have [00:21:00] engines on the back end that are used to going through syntax and understanding, the difference between break B R E A K and B R A K E, you know what I mean?

Like the all the homonyms. I get all the homophones mixed up all the time. So I'm really grateful for those tools because they find those things and I wouldn't trust ChatGPT to do any of that right now, at least. Mm

**Thad McIlroy:** I'm with you on that a hundred percent.

**Danica Favorite:** Yeah. Let's get into our questions. In terms of AI, I know we've talked a little bit about it, but how are you approaching AI and publishing? Do you have a specific approach or how do you see AI and how it works within the publishing industry?

**Thad McIlroy:** A great question. The people I try and talk to are both authors and publishers, and I think, your audience is going to well know that. The publishing industry doesn't embrace them as enthusiastically as they might like sometimes. And sometimes, you'd almost think that publishing had nothing to do with authoring.

Where [00:22:00] authors know that their work has everything to do with publishing as well as authoring. Publishers lose sight of authors as, as, integral to the overall supply chain, if you put it that way. So when I connect with publishers on this, the first thing I try and make clear to them is you have to see this as a continuous workflow from the author at the ideation stage through to you at the marketing and distribution stage.

It's part of a long, long process, and to just, pick it up at some arbitrary point is inappropriate. That being said, publishers think of their work in a workflow metaphor. And so, it is a series of steps, manuscript acquisition reader's report then developmental editing, et cetera, et cetera, et cetera. Cover design, printing, and, and, and, and so the easiest way for me to communicate with publishers about this sometimes is to talk on very specific use cases.

And so, as we're [00:23:00] describing, don't use it for copy editing, do use it for developmental editing. It's an excellent developmental editing tool. And I try and focus on those areas where it has the strongest promise or delivery at this point. One of the things I talk about in the book is, for every organization, where there's almost always still a degree of skepticism, look for an early win.

And, look for something where people go, wow, oh, I, That's, that's amazing. Metadata generation, for example, boy, that's so much, easier, faster, more accurate than I could have done those kinds of things are, writing some of the marketing copy, all the different things where it's like, wow, it does that really well.

So I do advocate for early wins and then begin to take a more holistic view of the entire workflow. 

**Danica Favorite:** Yeah, I love that. I love that you bring in that holistic view and the idea of a workflow because you're right, you guys were talking about the copy editing. I use mine for developmental editing [00:24:00] a lot because especially now I've, I've joked about this a few times where things that I used to say all the time in my writing are now known as AI isms and I'm like, ah, well they became an AI ism because we use them so much.

And so now I'm like, okay, I have this thing that I just said that I know as an AI ism. Give me some ideas on how to rephrase it and developmentally that works so well or to say, help me plot, help me find these holes in the plot. AI can do that. ChatGPT's really good at that one. So I like looking at that as a workflow and saying, okay, where am I using it?

What am I using it for? Where can I, as you said, get that win to make it easier to use AI for your publishing process so that you can get some of that time back. So since we're talking about workflow, this gives me a really great segue into my next question, which is, what does your AI workflow look like?

**Thad McIlroy:** I was afraid I'd have [00:25:00] to answer that chaotic and cacophonous would be the description. I guess it's not so much as I think of it is not AI's fault that my AI workflow is so damn poor. It's that my workflow is so damn poor. And then you layer on technology onto a bad workflow, and the technology won't save a bad workflow.

I think a lot of people, don't understand that's sort of a base rule of technology. That if you try and, bring in a new tech that's magical and promising, and layer it on a mess, you're not going to find redemption there. And, it's, it's to my continual embarrassment and shame that I don't have a more professional work, given 

**Danica Favorite:** I love that. You said that I super love that. I think this was the conversation Steph and I had last night where she was talking about all of her automation systems. And I'm like, ah, crap. I'm so bad at that. But I like how you disclaim, [00:26:00] how you described it for yourself, because that is me.

Like I am chaotic and a mess. I'm trying to get my mess in order and I'm like, Oh yeah, AI is going to save me. And until I figure out how to think through those processes, AI is not going to save me. It's just going to add to that mess a little bit. So I love that. We have another chaotic workflow person in, in the group because I, sometimes I just, I'm like, Steph, can I just follow?

I actually, this is my dream. Steph, you don't know this. I'm going to share this right now. My dream is to just come and spend a few days with you, like regular days, not. On a retreat or something, just so I can see how you do all of your workflow. And I'm like, I need to do, I need to like channel Steph here.

Like I think when I'm struggling with workflow, that's what I say to myself. What would Steph Pajonas do?

**Thad McIlroy:** The reality of publishing, is that the workflow is I mean, I'd love to see your workflow too, Steph, because my belief is that publishing is almost inherently [00:27:00] unmanageable. There's a constant struggle to tame that, which is full of so many exceptions, I'm looking at the book here, the UPC code is different on Amazon than on Ingram. You find out at the last minute that they each handle that differently. And it's such an example of the kind of a little nitty gritty things that, you can build that into your workflow with awareness. On Ingram, they had trouble with the spine width because the paper stock they're using is slightly different than Amazon uses.

So I had to do a one sixteenth of an inch adjustment because they wouldn't let me through, approval without that. With those kinds of things going on, I think self published authors need to be kind to themselves when they're feeling like klutzes because it is tough. This is a business of exceptions and the exceptions are always important.

It's all of them are make or break. And so, yeah, the workflows are hard to harness. 

**Danica Favorite:** Yeah, I think that has to be one of our poll quote, Steph. We need to be kind to ourselves when we're being [00:28:00] clutzy. 

**Steph Pajonas:** We do. Yeah, you guys are flattering me for sure, but, it's like, this was one of those things where I had publishing down pretty pat after, like, I had published 40 books at the point where I'd started taking up AI. So I knew my process. I I had my steps that I went through for everything.

There would always be, there would always be something that would, that would go wrong. Always without fail. It would be something like Amazon. I remember this one time I published my book and Amazon said it wasn't available. And I was like, what do you mean it's not available? It was a glitch that was happening to certain books.

It was these, there are lots of things that are just unforeseen that get in the way. So we try to get our process to a point where we're 95 percent there. And then there's that 5 percent of possibly things could go wrong. Right. But I feel like my life, since I've added AI to it, where it has streamlined some things, other things have been [00:29:00] super chaotic, just because the industry itself is super chaotic.

There's, there's new LLMs coming up out all the time. And then. Suddenly it's like the prompt that you used yesterday that was working like a charm is real crap today and it's just not producing anything. So it can be, it can be very dizzying, almost like one day you feel great and you're ready to go. And the next day your head spinning. 

So I feel you, I could give you a hug right now, like I give you both hugs and understand that you're not alone. You're not alone whatsoever. That everything about this business in general. Publishing and AI and then you put the two together is definitely chaotic. So, just do your best.

Just do your best. You'll be all right.

**Thad McIlroy:** the hiccup that I faced when I uploaded the book to Ingram?

**Steph Pajonas:** I did. I'd love for you to talk about that actually. 

**Thad McIlroy:** It was extraordinary because I, as always, again, with publishing, we're working on deadlines. I've got to get this book out. Well, suddenly it's [00:30:00] Friday, and I must've uploaded it earlier in the week because it was on Friday morning that I get a notice from Ingram saying your book has failed to pass our, system and will not go on sale and has been rejected, not merely delayed, rejected click here if you wish to appeal.

So of course you're, click instantly. You write, I don't know what it is you're talking about, because they give you a list of nine different reasons why it might have failed. One of which is, the book may contain AI generated content. And I look through that list of nine things. And the only thing that I could even imagine that the book might have sinned for was the AI generated content because you go into a part of the book where I'm explaining how to prompt AI and I quote, specifically how the AI responded in that particular moment. So that was picking up on that and failing the book. 

But when I clicked my appeal, you get another automated thing back saying, This will take [00:31:00] up to 14 days. You then Google, Ingram appeal rejection process, and you find on Reddit and elsewhere, Oh, it took me two months to get that cleared up with Ingram.

And you think, well, that's ludicrous. And I can't have my book locked into the twilight zone for two months. So I went public with it on LinkedIn and elsewhere. And, fortunately Ingram responded to that. They came back to me on the Monday saying it wasn't because of AI. It was for something else.

We, we can tell you what that was, but only off the record. So you can't tell anyone what it is we're going to tell you, which the reason your book was rejected but it wasn't, it's just like, well, if I can't describe that to other people, we don't want you to tell other people because bad actors could take advantage of that if they knew.

And it's like, I don't really think so, but say what you will. But the book went into distribution and just the part that really struck me was the irony that [00:32:00] if indeed it was rejected because it contained AI by a system that would be AI based anyway, right? Their systems are all machine learning and so AI is rejecting a book that has AI that's about AI because of that and so on.

It was mind, mind blowing at the time. 

**Steph Pajonas:** It's recursive. You're stuck in the matrix.

Right? So I'm glad they cleared it up and got you through the process because there's nothing worse than sitting there and feeling like you just, you can't do anything. Right. You can't do anything. 

**Danica Favorite:** And I think that's a good story because I, think that a lot of authors have that frustration where their books get rejected for really unknown reasons and they're like, wait, what? How is this possible? This happens and we find ways of dealing with it. And so to everyone who's had their book rejected for a really bizarre reason, you are not alone.

**Thad McIlroy:** not [00:33:00] at all. And these just, these institutions are so powerful now that, we have to deal with it. And we're powerless. They have all the power. They act like people with all the power and treat us with the kind of contempt that the powerful treat the powerless. So it's very, it's demeaning, humiliating, angering. And inappropriate. 

Aside from that, I think it's a good system.

**Steph Pajonas:** Aside from all of those things, it's a pretty good system. I'm grateful. I'm grateful for all of the, all of the opportunities we've had now as independent publishers. We certainly did not have these opportunities, in the past and I'm, I'm grateful for them now. I just, I wish, I wish they weren't sending me more gray hairs every single day.

**Thad McIlroy:** Yeah. 

**Danica Favorite:** for sure. For sure. And, and hopefully, like the hope. is that eventually the AI would help be more of an equalizer and be able to do that. I know at PublishDrive we have a book review process as well. And some of it [00:34:00] goes, again, through AI that automated thing looking for the errors. And then we have a human review it at the end. 

I found one where I was just talking today with the team and I was like, Hey, like. Our instructions to the AI are bad, we need to correct this instruction in this place, and they were like, I didn't even think about that. So it is a learning process where the machine learning and this happens with any AI you're using, where you take the machine learning and you're are constantly training it and talking to it.

Like Steph said, the prompts we use the other day. Aren't working today. And why? Well, they, they made a tweak in the AI. And so then you have to go back and say, okay, AI, this is the prompts that we use. Here's the result we got. This isn't the right result. How do we tweak the prompt? And then we'll come back with a tweaked prompt and then we're off and running again. And some days it says, yeah, no, I'm taking a break.

**Thad McIlroy:** Do you guys advise your listeners about the challenge of [00:35:00] prompting, of getting it right? I'll say my perspective is that it's still hellish for the average user. It's worth getting it right. That is your incentivized to really work hard at understanding prompting and trying to get good at it, but it's not easy at all.

And so for the average user, it's arcane, complex, doesn't make sense. At the detailed level, it makes sense to say, well, you got to ask it a question. Okay. Got that. But the question actually has to be three pages long. If you want the result to be what you're expecting, how do you advise people around that?

**Steph Pajonas:** At the at the Future Fiction Academy, we teach people how to actually prompt the AI, how to layer their prompts with details in order to get the kind of results they want. You can't just ask for five science fiction romance ideas, you have to ask for, I would like five science fiction romance ideas that involve a secondary planet and a kidnapping and you [00:36:00] have to give it all of these ideas that you may have in order to get a better output. 

I often tell people that what they need to do is they need to think of the AI as a junior writer. Somebody that you're going to be giving them an entire brief about what they're going to be writing. The style of the writing that they're gonna do, the characters, the setting, and everything that's going to be involved.

You cannot leave this poor junior writer flapping in the wind, not really understanding what they're going to be writing. So prompting is very much the same. We have to be explicit about the kinds of things that we want. And you have to learn writing craft in order to have the vocabulary to prompt the AI.

Because if you don't understand things like 3 Act structure or emotional cues, visceral feelings. These are all words that the AI actually knows, but if you're not using those words in your prompt, you can't get a good output back. So we [00:37:00] try to teach them these techniques and get them rolling and we'll see how they're prompting and we'll get feedback on those prompts to show them that, Well, we weren't detailed enough here, or you didn't put your instructions all the way at the bottom because that's important too. All these little details. 

And then we just, we try to do this through education as much as possible. It's not, if somebody doesn't want to learn how to prompt, then they're just not going to learn how to prompt, right? You got to be motivated in order to learn these things. It's a whole new thing to be educated on.

And this is where I feel that we were going to make the biggest splash in this business is through education. It's not just giving an author an easy button to press, To get output. It's to teach them how to ask for the things that they want from these tools. So that is where we're going with that.

We're just trying to teach one person at a time and hope that that person teaches somebody else.

**Danica Favorite:** Yeah, and I think that's important to that, because the FFA just launched a [00:38:00] really cool free tool called Raptor Write. It's fun. It's wonderful. And I'm listening to all the people who are saying, Oh, this doesn't work. This doesn't work. And always it goes back to, well, you don't know how to prompt and you need to learn how to prompt in order for the tool to work properly.

And, one of the big ones that I just saw was someone. Mad because they were using negative prompt and we've learned negative prompting doesn't work with the LLM. They don't understand the word. No They don't understand if you say no, you can't have ice cream All they're gonna think about is ice cream and they're gonna give you look look at this ice cream. Look at this ice cream. Oh, no, wait, I have better ice cream for you. And it just doesn't understand that.

And so I think it's really important to educate people on what the LLMs are capable of doing what they understand and how they think because we know they don't actually think. It's not the same kind of thought process that we go through so we have to tell them how we want them to think and how [00:39:00] we want them to act. I really like depending on which LLM you're using a lot of times if you look at the documentation for the LLM, they give you suggested ways to prompt the LLM. So that's always a great way to start. 

But also the other thing I like to do is just have a conversation with it. To say, okay, here's the results I'm trying to get. I'm not getting it. How do I get it? And really like, okay, what am I missing in the prompt? And I spoke in a previous episode about my struggles with mid journey. I asked when I use, Midjourney and I get good results. A lot of times, it's because I've asked Chat GPT or Claude or Gemini, I've chatted with them and said, okay, I'm trying to make a Mid journey image. This is what I'm looking for. How do I do it? And nine times out of 10, they give me a prompt that I then put in Mid journey and boom, it works! So, 

**Steph Pajonas:** Helping another.

**Danica Favorite:** Yeah. Yeah. 

**Thad McIlroy:** We also point to, I mean, you're saying that, AI [00:40:00] doesn't think the way we do what we seem to see is that you're treating it as if it does. In fact, as you're saying, Steph, that thinking of it as a writer's companion and in publishing, we say thinking of it as an intern is the way to do it and talk to it like an intern, talk to it like a junior associate.

And ask it for help and be explicit and even be emotional in your, reaction. I really need this answer and I need it right away. My job depends on it will elicit a different response than not including that kind of emotional appeal. But there's lots of things to learn and not a one button push on this stuff.

**Steph Pajonas:** Wasn't it Amy like last week or so that was telling us that she tells her AI that if it doesn't answer the question properly, she's going to take away its cat videos because it watches cat videos on YouTube. I was like, Oh God, that's a great idea.

**Danica Favorite:** Yeah, that's where I was going with that. Where, like, I love that Thad's like, do the emotional appeal. I'm going to lose my job. And [00:41:00] Amy's like, I'm going to take away your cat videos. And so threats work.

**Thad McIlroy:** don't know which is worse.

**Steph Pajonas:** I don't want to be remembered like that in the robot apocalypse, though, I wanted to remember that I was said please and thank you and I was very nice to you.

**Danica Favorite:** Yes. Yes. This is the nice lady who said, once you get the job done, I'll give you cat videos.

**Steph Pajonas:** I'll give you, I'll give you cat videos. It's all good.

**Danica Favorite:** Yeah. Yeah. Oh, that's awesome. So we've talked about a lot of tools. And so our last question, do you have a favorite AI tool? I know we've talked about different AI tools, do different jobs, but is, is there a favorite or something you really love? 

**Thad McIlroy:** I, there is in fact, but it's more publishing specific per se, although it's absolutely applicable to an author that really interesting tool to look at is from a company called Veristage, V E R I S T A G E. And they have a tool called Insight. And it's their attempt to put a front end onto AI with a publishing specific focus.

And [00:42:00] so, you start off with Insight, where you upload the raw manuscript, and then it takes you through a series of steps from there. And so, when we're talking about workflow, Veristage is trying to formalize the workflow of publishing, of editing and publishing. Through an AI lens is really, really interesting.

The other one that I like is Shimmer, which is a AI tool for marketing, where you prompt, it creates advertising that can go on Google or Facebook based on your book. It's optimized for fiction, and it delivers those ads to its assessment of the audience that would be most receptive to the advertising and the publishers who are using it so far are seeing really dramatic ROI on it. It's as applicable to a self published author as it would be to a publishing company. I don't know. They're set up as yet deal to work with individual [00:43:00] authors. It's really an intriguing use of A I that goes beyond the obvious.

**Steph Pajonas:** That's great. I actually heard of the Veristage. VeriStage, right? VeriStage tool through LinkedIn, maybe a couple months ago, saw somebody talking about it. So I was like, Oh, I'm gonna check on that later. But I haven't heard of Shimmer. That's interesting.

**Danica Favorite:** Yeah. Yeah. I actually have heard of Shimmer. Part of it. They were on the panel with that, but we'd heard of it through some other means as well. And it's really interesting that you're saying people are getting good results. I may have to take a second look at it. Because what I was seeing when I was playing with it a little bit for some of the demo stuff they had, I, I wasn't impressed.

And so now I'm like, oh wait, if people are getting good results, I think ultimately that's how you have to measure it is to say, okay, who's getting the results? And even if I look at it and say, Ooh, that video is icky, I would never click on that if that video's getting clicks. I think sometimes we have to take our own personal opinion out of the equation say, okay, I may not like it, but if it gets [00:44:00] clicks, who cares? So, I'm 

**Thad McIlroy:** They do a really fast, they create a psychological profile, both of the book itself and of the reader. And try to, interpret through their own psychological assessment of, sort of the interest points and the sensitivity points, what would be most appealing. And so they have a very sophisticated view of it. And as far as I've heard so far from customers, they're happy with 

**Danica Favorite:** cool. That's really good to know. I'm glad to hear that. Because like I said, I, I saw it and I was like, and I'm like, okay, it's worth a second. Look, if people are getting results because that ultimately, it is what I'm always after is, okay, but do you get the results? So very, very good. I am glad to hear that.

**Steph Pajonas:** Excellent. We've definitely covered a lot here and I'm, I'm excited to find out about some of these other tools that you mentioned and I'm excited to read your book as well. Well, let's tell the audience about where they can find out more about you.

**Thad McIlroy:** Well, my website is called the future of [00:45:00] publishing. com where I've always been a little bit embarrassed by the URL. I claimed it back in 2005 when it was sort of one of those ones where it's like, what's the URL going to be for my new consultancy? The future of publishing, I guess. And then it's like, Oh, so you think you're the future of publishing, do you? Think, well, I got a couple of ideas about it, but no, I don't. There's lots of other people with even way better ideas than I do. But nonetheless, that's my URL, the future of publishing. com. The book is The AI Revolution in Book Publishing. You can subscribe to it in a sense on LeanPub or for the same price.

It's the same price everywhere. On Amazon and all the other retailers. It's 9. 99 as an e book and 14. 95 as a print book.

**Steph Pajonas:** Fantastic. Fantastic. Well, I think that you might've been a little bit prescient there with your domain name of choice. I think that was a good choice of yours, so I'm all for it, especially now since you've come into the AI and publishing as well, which I happen to think is the future, and I think Danica agrees with me [00:46:00] on that

one, don't you? Excellent.

**Thad McIlroy:** on the same page there.

**Steph Pajonas:** Well, thank you so much for coming. It was great talking to you today. For any of our listeners, we're going to, of course, produce a blog post about this particular episode. We'll include any of the links or companies , Thad talked about today so that you guys can go check them out for yourself as well. And in the meantime, we're going to say goodbye and we'll see you guys in the next episode. Thank you so much for coming, Thad, and we'll say goodbye now. Bye.

Bye.

Thanks for joining us on the Brave New Bookshelf. Be sure to like and subscribe to us on YouTube and your favorite podcast app. You can also visit us at BraveNewBookshelf. com, sign up for our newsletter, and get all the show notes.