AI Finally Surpasses Human Book Editors

HOST: It's good to have you back, Bo. Last time we got into logical fallacies and positive psychology, and I've been thinking about that line you used in your first episode: "It's not difficult to publish a book—it's difficult to publish a profitable book." Today, I want to test whether BookEditor.io is actually closing that gap. The site says multiple passes cover grammar, consistency, style, and structure. But "better than a human editor" is a bold line. What's the specific mechanism here—what does a human line editor systematically miss or get wrong that AI catches, and how do you know? GUEST: Those are a lot of good questions. Let's start with AI technology and where it is today. Prior to, oh jeez, I don't know, maybe even a year ago, it was pretty basic and AI did a good job at checking for grammar, basic grammar and spell check. But a lot of the consistency, style and structure part just wasn't quite there yet. But with the release of newer models, especially in the Anthropic family and in the OpenAI, we can now do this. And what's amazing is, compared to like a human editor, a human editor, first of all, would have to read the entire book, which can take weeks for a substantial book. But when you do, you don't quite remember everything that that you read, at least not the specifics. So it it's kind of difficult for the human mind to be able to uh, edit a book in the same way that artificial intelligence can take in the entire book and keep everything in memory at once, and then look for those consistencies and fix them and and change things that a human editor would just miss. And being in the industry for as long as I've been in, and going through so many books myself and and using so many different editors, no matter how many times you use an editor, and no matter how good the editor is, something is always missed. Several things are usually missed because the editor is only human, and that's what happens. Usually, the large production firms have a team of editors, and uh, individual authors just can't compete with that. You You don't have the budget to hire a whole editing team or department to go through your book for months. You just can't do that. But now with AI, with what we've built, we're finally putting authors on the same level and that that whole competition um, problem with being able to compete with the large publishers, with the quality and the style, that's that's a thing of the past with with AI. So it's very good for independent publishers. HOST: So, essentially, you're arguing that AI isn't just faster or cheaper—it's structurally better because it can hold the entire manuscript in active memory at once. If that's the claim, where does the 5% problem come in? In your interview on cover design, you famously said AI handles 95% of projects really well, leaving 5% that still need a human. Does that same ratio hold for editing, and what does that 5% look like? Is it literary fiction, highly technical non-fiction, or something else? GUEST: Yeah, I would say it's probably closer to 99% when we're talking about the book editing, but there is a there will be some unknown percentage of people who need a specific editor. And I think you nailed it with the highly technical non-fiction. There are certain types of editors that focus and specialize in certain books. And these are these are pretty high priced and we typically don't deal with them in our industry in the self-publishing part. But there are editors that will like let's say that they're they're doctors, they're medical doctors. So they'll be able to edit in medical journals journals. Um, AI can do that, but AI doesn't have a PhD or an MD specific in in that area. So it it will be lacking uh with that highly specialized knowledge. But besides that, I can't see any other reason why users would choose a human editor over something like what we've built. HOST: That makes sense—you're saying that for hyper-specialized domains, like medical journals, that human expertise still matters. Now, looking at the four control modes you've built in—everything from "Just Fix It" to "Suggestions Only"—that's an unusually granular approach. What are you seeing in user behavior? Which mode are authors gravitating to most, and has that surprised you? GUEST: Well, we put that in there because some authors don't like their words changed. They they write a book, they have a message and they want to get it out, and that's what's most important, even if their language isn't perfect. And that's perfectly understandable. Other authors realize that they're not the best at grammar and structuring sentences, and they say, you know, just fix it for me. Uh, it's okay if you change my words, as long as the overall message stays, that's what's most important. I want it to be very readable, I want it to be enjoyable. So, just go ahead and fix it. So, those are the the two basic modes that we have to meet virtually any author, uh, and their preference of editing style. What user behavior am I seeing? Mostly, the just fix it. I think that, uh, most people are very happy with the way AI will change what their exact words are in the structure of the words. They realize that, yeah, AI could do a better job than I can at this. As long as they're my, it's my idea that's still being communicated. That's what's most important. HOST: Got it—so, for most, the draw is definitely that frictionless solution. You also mentioned you can generate a "story bible" at the highest tier—what does that look like and who's benefiting from it? Are these series authors or just looking for consistency in a single book? GUEST: Yeah, it's one of the um one of the major many benefits of the higher package. It is something that uh some people have been asking for, something uh with a little bit more detail and explanation and ideas to enhance the the book where they don't actually have to accept everything right away. So, most authors will go into the editing process with the idea that their book is ready and it just needs to be looked at and anything that's caught, you know, like uh inconsistencies or structure or grammar errors, whatever. Uh but other people want an editor to say, okay, read this and tell me what you think. Where can I make improvements? What What's lacking? What could make this better? And that's where the story Bible really shines at the highest tier. HOST: So, it's not just about grammar or word choices at that level—you're getting structural suggestions on the narrative itself. Now, the site just added a direct pull from SelfPublishing.pro into BookEditor.io, which closes a loop we've talked about over several episodes—from writing in BookBud to editing here and then distribution. Is BookEditor.io the piece that was missing to make that full pipeline actually work end-to-end? GUEST: It wasn't the piece, but it was definitely one of the pieces. Many of the authors that come to us usually ask about editing to begin with. So, we've tied that in. It's completely integrated. So, it just makes the process a whole lot easier and more enjoyable. HOST: It definitely sounds like a smoother workflow for authors. Before we go, for listeners who want to follow up on what we covered, where can they find you and the work you're doing? GUEST: You could find more about the website we're referring to at bookeditor.io, or you could find out about all our companies at archyboy.com. HOST: Great. Good to have you on again, Bo, and thanks for breaking down how BookEditor io works and why you see it as a step above traditional human editors for copy work. And thank you for spending part of your day with us. Let's get you back on soon—there's plenty more we can cover. Until next time—that's a wrap.