Episode Summary
Will book-writing AI make authors obsolete? Or can AI be a tool to help authors write more books, faster and maybe even better? As the technology improves and becomes more ubiquitous in our lives, these are just some of the questions authors are facing. We’re exploring Artificial Intelligence through the lens of the smart author.
Episode Notes
There’s no doubt that AI is sweeping across the digital landscape, changing the way just about every industry thinks and operates. It’s still early days for knowing how this will impact publishing, particularly for the independent, self-published author. But the technology is evolving so rapidly that “early days” may become ancient history in the blink of an eye.
In this episode of Self Publishing Insiders, D2D’s team of industry experts will take a look at how AI might play into the future of writing and publishing in this space.
//Draft2Digital is where you start your Indie Author Career//
Looking for your path to self-publishing success? Draft2Digital is the leading ebook publisher and distributor. We’ll convert your manuscript, distribute it online, and support you the whole way, and we won’t charge you a dime. We take a cut of royalties on each sale you make through us, so we only make money when you make money!
• Get started: https://Draft2Digital.com Get insider info on indie author success from our blog.
• Visit: https://Draft2Digital.com/blog Tune in to our monthly livestreams and ask us anything!
• D2D Live: https://D2DLive.com Promote your books with our Universal Book Links!
• Books2Read: https://books2read.com
//Get ahead of the Self-Publishing game with our Amazing Partners//
Findaway Voices || Find a narrator, produce your audiobook, and distribute it to retailers worldwide, including Audible.com and Apple Books.
• http://findawayvoices.com/d2d
Reedsy || Assemble your team of publishing professionals! Find editors, cover designers, marketing experts, ghostwriters and more.
BookBrush || Build graphics and video that help you market and promote your books.
• https://bookbrush.com/d2d-mockups/
//Join the D2D Community Online//
Facebook || https://facebook.com/draft2digital
Twitter || https://twitter.com/draft2digital
Transcript
SUMMARY KEYWORDS
ai, book, author, turing test, human, prompt, read, voice, gpt, question, point, intelligence, generate, world, kevin, people, advertising, talking, technology, keywords
Kevin Tumlinson 00:01
Well, hello, everybody, we are live. The computer machine has told me so. And that is ironic because the topic today is going to be AI and how it impacts the self-publishing industry and your career as a self-publishing author, and I can’t think of anyone better to help us chat through that than my AI friend, Nick Thacker.
Nick Thacker 00:27
Anybody? You can’t think of anybody? You need to know more people, sir. That’s your … No, I’m just kidding.
Kevin Tumlinson 00:37
Socially I’ve become something of a pariah. So I don’t have very many people to call on anymore as experts. So, you know, this is a … I’m gonna confess something. And you’ll back me up on this, probably by slamming me, really.
Nick Thacker 00:55
Challenge accepted.
Kevin Tumlinson 00:57
I have to say that I’ve kind of come around on the AI front after you and I had a whole session with an AI that you have access to. I mean, I still don’t think AI is anywhere near replacing writers. Right?
Nick Thacker 01:18
I don’t think so. I don’t think so.
Kevin Tumlinson 01:19
I know that China has actually released the first intellectual property created solely by an AI. They did that at some point over the past year. And so that might … I don’t know, that says something. I haven’t read the book or anything. But it says that the technology is kind of up there. But I remained unconvinced until you and I had our little session with the AI that you’re playing around with, which is, and I always get the letters mixed up. What is it called again?
Nick Thacker 01:55
GPT-3.
Kevin Tumlinson 01:56
GPT-3. And what does the GPT stand for?
Nick Thacker 01:58
I have no idea.
Kevin Tumlinson 01:59
Okay. That’s fair.
Nick Thacker 02:02
General public trial three. I don’t know, it’s …
Kevin Tumlinson 02:05
Why don’t we call it that? Yeah.
Nick Thacker 02:06
I’m sure there’s probably something for it, but …
Kevin Tumlinson 02:08
Don’t get mad at us. They may actually have GPT-3 monitoring us right now, as we speak, for all we know.
Nick Thacker 02:15
As soon as you engage its services, it follows you around the internet. But to be clear, to back up a little bit Kevin, what was it that you weren’t convinced about. regarding AI? I feel like I know the answer to this. But for anyone listening, like what’s the thing that you’re not convinced? That it’s gonna replace writers?
Kevin Tumlinson 02:33
You know, yeah. Well, that’s part of it, that it could ever fully pass the Turing test is what I always suspected. And if you’re not familiar with that phrase, and I’m thinking most authors probably are, but the Turing test is sort of a famous historical test for artificial intelligence. If a computer can fool a human, is basically what it boils down to, then it has passed the Turing test. And there have been a lot of, not quite AI attempts, there’s been a lot of sort of scripted service attempts that have come close to passing the Turing test over the past few decades. But it’s only been the past, you know, couple of years that suddenly we’re getting true artificial intelligence that can carry on a conversation with a human being and go undetected.
Nick Thacker 03:22
Yeah, I thought we had computers that could pass the Turing test. Again, I mean, it’s one of those things where, you know, somebody designs something. And it’s not out of the realm of possibility for software developers to develop software specifically to pass a Turing test. Which puts it in the realm of artificial narrow intelligence, right? This this very specific, like a chess-solving IBM computer, for example, is very good at chess and can beat Kasparov, right? But it’s not a general intelligence. It’s not something that you can ask, you know, how’s the weather? But Siri and Alexa, the things sitting on my counter, can do those things. But they couldn’t play chess with me, you know? So these are all examples of narrow intelligence. So I don’t think it’s out of the realm of possibility to say, we’ve got computers that can pass a Turing test. But the problem is, you know, when Alan … Oh, gosh, I’m blanking on his last name …. Turing. I guess it was Turing, wasn’t it? Yeah. When he came up with this idea of the test … I was thinking of someone else. I was thinking, you know …
Kevin Tumlinson 04:28
Forget artificial intelligence. Now we’re just challenging actual intelligence.
Nick Thacker 04:33
Yeah, I was thinking of Benjamin Cumberbutton. Because he, I think he played him in the movie. But, you know, the point is, I think, man, I think we’re there. I think as far as passing a very limited example of a test, we’re there. The thing is, everybody who looks at it then goes, but wait a minute, this isn’t … This isn’t Skynet. Like there’s obviously something missing. There’s a gap. So a computer that can solve the Turing test and, you know, a robot that’s gonna march down the street and decide to, you know, get sentient and go out on its own and kill people, we’ve got a pretty big gap there. And that’s kind of where we are right now, is finding and discovering algorithmic ways to produce intelligence in the box. Now, all of that to say, what’s fascinating to me is that the keyword there is in the box, you know, in a computer. The more we expand computing capabilities, the more we’re able to expand AI. Now, that sounds very obvious. But we’re going to get to a point very soon, I think, where, you know, hard drive space won’t double every 18 months. RAM speeds won’t … exactly, Moore’s Law … At some point, and I mean, this is somewhere on Wikipedia. I’m sure it shows the exact breakdown of when Moore’s Law stops working for us. We physically can’t, outside of using nano or quantum technology, cannot cram more chips onto a board. It just doesn’t work. Our teeny little fingers aren’t tiny enough, right? And so, I think that’s the point. But again, and you know, I mentioned two other keywords. We’re working on nano, we’re working on quantum, so we’re gonna get to this point where we don’t need to worry about silicone as much anymore, because we have these other technologies that we can use. And then again, the AI advancement will continue. But anyway, that’s my whole my invocation here to begin. I think I’m like you, I’m just imminently fascinated by what AI is capable of today, and I’m trying to put it through its paces to see if it’s useful to me as an author. And I think that gets into what people are going to be listening to us talk about, because there [inaudible].
Kevin Tumlinson 06:45
And we already have some questions coming in. And I want to take this moment to say thank you, first of all, to everyone who’s already asked questions. And we’re gonna get to them, I promise. And if you have not thought about it yet, this is your cue—go ahead and ask us anything you want in the comments wherever you’re watching this, on YouTube, Facebook, or elsewhere. And we’re going to try to answer everything that comes through. So today’s topic is about AI and how it impacts the self-publishing space and self-published author. So on that note, before we get to questions, what I want to do, Nick, is talk just a minute about the experience that you and I shared in playing around with the GPT-3 AI that you have access to. Because one of the things—basically, you accused me of being a curmudgeon. And I can forgive you for that, because it’s true.
Nick Thacker 07:43
I’m not sure I would change that perception at this point. I’m just kidding.
Kevin Tumlinson 07:46
But the sort of presentation was, you had kind of offered this up, and I came at it immediately skeptical about how useful it could be. But what we ended up doing was playing around with it and plugging in a few … You can ask this thing questions the same way you would any human being. You know, you can type in a question, and it treats it as a prompt and it spins from there. We did certain little things that generated things like, you know, I was writing a series on characterization for the blog, for the D2D blog—which by the way, if you are interested, you should go to draft2digital.com/blog, and you can find our blog with posts like this. But one of the things that we did was ask it to give us a blog post on the topic of characterization. I have not used that post or anything, so don’t go looking for it. But what it did was return a kind of listicle post. And it was, to me, as a professional copywriter, it was fairly well-written. It’s what I think a junior copywriter would have produced. So with some editing, it was pretty good.
Nick Thacker 08:56
Yeah, it was useful. I mean, it was something usable. You know, maybe not immediately, but with a little bit of minor editing. But the point is, you know, this isn’t … we’re not asking it to be a good copywriter. We’re asking it to essentially give us its opinion on what we were asking. That’s what that prompt is, right? So it’s not like this … We’ve seen this, anyone watching has seen these AI chatbots and things like that, where they’re basically designed to have canned responses with a little bit of, you know, creative flair that they can bring in. And we know that there’s some downsides to that—you end up with racist chatbots and things like that. GPT-3 is very different in scope, because it’s essentiallyvery untrained as far as that goes. It’s very much just, they fed it the internet, a big chunk of Reddit, things like that. And they said “Hey, tell us …” You know, it’s essentially a finish the sentence, but do it in multiple sentence format, you know? So finish the thought. And so that’s what the prompt is. So we asked it things like, we had it grab, we put a link in and said, “Hey, give us a synopsis of this web page that we could explain to a 12 year-old,” because Kevin and I are both essentially 12 year-olds. And it spit back, you know, it was a scientific paper, remember? Or something like that. It was like on animal husbandry or something very intricately involved in something that I’m not versed in. And it spit back a pretty good … Now this wasn’t the abstract, it didn’t just copy paste anything. It actually read the article and spit out …
Kevin Tumlinson 10:38
Yeah, almost like an executive statement of the article. Like, it gave a very nice summary of the ideas, that’s what was impressive to me. It wasn’t just that it, you know, grabbed all the copy and boiled it down to something, it actually expressed the ideas of the article, which is profound.
Nick Thacker 10:59
And so I’ve used things like, I’ve done things with it, where I want a list. So I will, I’ve done this before, and actually I’ll just pull it up here, I can just read some of these to you guys.
Kevin Tumlinson 11:13
Real quick, this is from Leroy, who says that GPT stands for Generative Pre-trained Transformer. Because Leroy is clearly an artificial intelligence himself, is what that means.
Nick Thacker 11:27
That’s right. Clearly, Leroy. Yeah.
Kevin Tumlinson 11:29
Thanks, Leroy. We really appreciate it.
Nick Thacker 11:32
You know, Kevin, and I were on there, we did a call, we’re just kind of playing with it. And you know, we’ve started this podcast on the side that is about just the kind of things that we research for fun as authors that we throw into our fiction, called Stuff That’s Real. And so the idea with the podcast is, it’s things that are real, that also we think are pretty cool that we’re researching. Anyway, the point is, we dumped into the GPT-3 playground, is what they call it. And we’ve said, I don’t remember the exact prompt I gave it, but it was effectively, you know, I knew I wanted to list like, you know, 1, 2, 3, 4, 5 down the page, of ideas of things that are either scientifically, historically, technology, like they’re real things that exist that would make good podcast subjects. And that was effectively the one sentence prompt. And then I gave it two examples that we’d already used on the show: Geronimo, the Native American man, the Fermi paradox, and then I think I threw in the third one, the faces in pyramids in Antarctica that you’ve talked about Kevin. And then I just put four, period. So I’m telling it, I want a list, you know, then leaving it hanging. And it filled in 18. Now I have a list of 18. So it filled in 14 more prompts for us. In the fourth one, for example, it says the giant stone heads on Easter Island, which may have taken hundreds of years to build. Number five, a giant elaborately decorated oasis in the middle of the Arabian Desert. The stone heads of [inaudible] in Siberia. These are real things. Like, it’s not taking keywords and smashing them together in a way that sounds cool, but doesn’t actually exist. It’s got the Antikythera mechanism, the ancient Greek scientific instrument that some believe was used to predict the eclipse. So it’s grabbing this stuff, and it’s not plagiarizing, either, you know, it’s not just ripping from Wikipedia or whatever. It’s understanding what we want, and it’s giving us an actual usable list. I mean, this is something, you know, obviously, we’re not going to publish this list, we’re going to use it as prompts. So we can use it immediately.
Kevin Tumlinson 13:30
It’s essentially performing the function of like a virtual research assistant, which I would love. That’s why immediately after we did our session, I went and applied to see if I can get into the, like, beta program for this thing, because that alone would be very useful to me and cut down on a lot of my research time, so that I can churn out more novels faster. That’s very impressive. Um, we have a question. I think you may know more about this one than I do, but I have looked into it. So Mira, I believe that’s how I pronounce your name, Mira. If it’s not, I apologize. She’s coming to us from YouTube. “What do you think about AI vs. the interactive stories I’m seeing come out about like Chattables, via the ECHO with Amazon?” And Chattables, I think it actually is AI-driven. But it’s kind of like an interactive, almost like, choose your own adventure kind of AI story. Right? Have you heard of it?
Nick Thacker 14:34
Yeah, I played with them a little bit. Yeah, they’re a bit AI-driven. Obviously these, because it’s an Amazon brand type thing, going through the ECHO and all that, there’s obviously some limitations and boundaries put on this AI. So it’s highly in the realm of a narrow AI. You know, if you think about it in another way, another sense, you can do this without AI. You can have a long list of story prompts, and a long list of story pieces that fit together and that are interactive. And we already know that the ECHOs, the Alexa devices, can hear people’s voices and interpret and parse English at least, and you know, other languages. And so we don’t really need an AI to put this together. The fact that they’re using an AI for some of this stuff is really neat. But this to me is just again, a very much an artificial narrow … I don’t think I talked about that. But you have three basically different gradations of artificial intelligence, generally—ironic, pun intended—that most of us are using to describe AI. And the first one, that kind of the baby version, is artificial narrow intelligence, like we’ve talked about. And then you move into more of an artificial general intelligence, which is where GPT-3 is trying to go. And it’s still a bit limited, it’s a bit like talking to a four-year-old, sometimes. You may get some really interesting things out of it, but it might not be the highly academic conversation that you think you might want to have. But once we get to that point, where you don’t really know you’re talking to this AI thing, you’re just having a conversation, and it’s understanding, just like Kevin, and I know, okay, it’s his turn to talk, and then I’m going to talk, and then there’s a question, we’re gonna, that’s where the artificial narrow or the artificial general intelligence is. I think humans are pretty much considered artificial general intelligence. Maybe not artificial. And then you move to the Skynets, you know, to the Ender’s Game kind of intelligences. And these are the artificial super intelligences, where they are basically able to learn on their own. They’re able to, if you connect it, this is the sci-fi theme, right? If you plug the computer into the wall, and give it access to the internet, what we call the singularity will happen, where it will be learning so quickly and expanding on itself and improving itself so fast that humans cannot possibly keep up. And it will suck up all the computing resources and become Skynet. And humans will just be meat for it’s … so that’s where obviously we get into the cyber realm and …
Kevin Tumlinson 17:10
Or become batteries for its flying robot fleet.
Nick Thacker 17:15
Right. And so that’s, those are the three different layers here that we’re kind of talking about. We’re right at the cusp of artificial general, depending on who you ask. That’s my opinion. But we’re not there yet, and we’re far away from the singularity.
Kevin Tumlinson 17:25
And I kind of believe that’s where it will stay for quite a while. I don’t think that …
Nick Thacker 17:30
I think so too. Because again, you’re talking about computing, right? You’re talking about computing power. Even if an artificial intelligence could be plugged into the wall and suck up all the information in the internet, how can it possibly process that with a single CPU, or even you know, if some Bitcoin miners got a bunch of GPUs sitting next to, even that is not enough. Even the supercomputers that China is building, that we’re building, that … you know, it’s just not enough yet.
Kevin Tumlinson 17:54
So swinging this back around to how this relates to authors in particular, this is a good question actually. Shawn on YouTube asks, “How about AI for audiobooks?” And I have to say, this was a thing, and this is something that is being explored. But recently, there have been developments that have shifted the game a little. And what I think is going to happen is, we’re seeing … Recently there was an article, and I don’t remember where it appeared, maybe Elyssa or somebody can track—I’m not gonna put that on Elyssa, never mind. But it was an article about how in the gaming world, they’re using AI audio to generate character dialogue. So basically, you get to a point where you’re interacting with these characters on screen, and all of their dialogue is being generated dynamically by the AI. Where I think that’s going to be important is in generating, it’ll work the same way … That technology will eventually come back to … Actually, no, I’ve framed this entirely the wrong way. That is one aspect of it. The other aspect of it was, they can provide the script, and the AI voice could read it naturally, which was much more relevant to this question. So that technology already exists. There was kind of a little farcical parody site that had Jordan Peterson’s voice, and they would plug in all kinds of weird stuff for him to say. You know, the technology to mimic someone’s voice and make it sound very natural is already here. So I think it’s, right now, we’re at a point where the barrier to that being in full use really comes down to the legality of certain things, the intellectual property and that sort of thing. But also the competition with like the Screen Actors Guild and stuff like that. So …
Nick Thacker 19:50
Yeah, two things on that. And Elyssa already just dropped the link in to both the things, the thing you just mentioned, VentureBeat’s article, and then the thing I was going to talk about, which I’ve been playing with a little bit. Amazon web cloud, their web services division, has a service called Amazon Polly. And this is an AI neural intelligence text to voice thing. And it’s really, really, really good. It’s not there yet, you know, it’s not quite ready yet for primetime, but it’s really, really good. It’s got some built-in voices, just like, you know, if anyone’s got a Mac, you know, you can have Alex read stuff. Windows has the same version, I don’t know what it’s called. But this is like a much better version of that, where you can actually go in and tell it, not just what to say … And this is where it needs to improve a little bit, right? It can’t automatically—I mean, it can, but it’s not great yet—it’s not quite ready to completely interpret the tone of dialogue. For example, it doesn’t understand sarcasm out of the box yet, you know? And so you can go in and program some of these inflection points. But if you’re going to be spending time doing that for an entire audio book that’s 10 hours long or 14 hours long, you know, you’d be better off just paying somebody to read the audiobook, paying a human. Now, the second thing I was gonna say is, I have been playing with a service, and I can link to it, I don’t remember the name of the company. And they weren’t very good yet, so I don’t want to mention it, you know, but they’re on the right track. And this company is doing the same thing as Amazon Polly, what I just described, is doing. But they’re also, this is what I was looking for. They are also allowing you to go in and I, so I’ve done this, I’ve sat down and you read 50 paragraphs of text that they show up on the screen. And it automatically records and stops the recording. And then it processes your own voice, your actual human voice, and tries to make one of these neural voices that’s based on your voice. And so I was really pumped about it, because this is exactly the question I wanted to get. Because this is where I want to see audiobook stuff go. For me as an author, personally. Now, I love narrators. I’m not saying I don’t like you, narrators. But for me as an author, how cool would it be to just finish my manuscript and have an audiobook popped out?
Kevin Tumlinson 22:04
And what that does is, it’s the same thing that most indie publishers sort of fall back on. It lets you own that process without necessarily having to hire outside. So I, as a narrator, you know, I like my voice, and I do some narration and then, you know, I could do that. But it’s so time-consuming to do an entire book. Where I was hopeful was that one of the services would allow me to do the book, and then I can listen to that production. And then because it’s my voice, when I come to a part that they’ve flubbed or whatever, I can just record that part myself and that would cut down the time of recording. Right.
Nick Thacker 22:43
Right. And I mean, again, because this is AI, and the way the sky this works technologically speaking, is, it’s similar to quantum computing in a very small sense, which is, everything that would happen in the box with this AI, with this neural net is, it’s going to be given like a grade of quality, a quality score, if you will, like a margin of error. Like, oh, well, this is a 95% margin of error, which means this sentence probably isn’t gonna sound as good. And so just by stripping away some of the some of the UI or the UX there, we’ll be able to get potentially get in and see, okay, well, this is one section of the book, just like Kevin’s talking about, where I need to re-record this paragraph or this chapter, because it’s flagged as potentially not sounding very good. So it can save us even more time by just saying, okay, well, all this stuff that’s not flagged at all is going to sound exactly like me, because it’s my voice, all that, and I don’t have to waste time going through that. You know, so I mean, yeah, this is this is really exciting to me as an author because I think we’re already getting to the point where we’ve got some artificial narrow intelligence doing editing, by way of things like Grammarly, Pro Writing Aid, you know, there’s other AI things on the market that are coming out that’s going to help us process our manuscripts and get better at the language part. And then we are very, very close to being able to have, even if not our own voice—which is, again, I tried this service out, and I’ll just be, it was just really bad. It just didn’t sound anything like a human being at all. Certainly didn’t sound like me. It was cool that they’re working on it, though, but it’s not good.
Kevin Tumlinson 24:12
I’ve heard some that are pretty good.
Nick Thacker 24:14
But they do have other voices, yeah. That’s what I was gonna say is, they’ve got, you know, their own branded voice that, you know, you’re free to use, and you got to pay them to use the—sorry, you’re not free to use, you’re free to use it for an audiobook, for example, if you wanted to go that route. You would just pay for the time spent recording and things like that.
Kevin Tumlinson 24:35
I think it’s only a matter of time.
Nick Thacker 24:36
It is, that’s my point is, it’s only a matter of time, right? I mean, we’re getting there very fast.
Kevin Tumlinson 24:41
So, uh, Leroy had a question that I think is interesting on the list of topics, meaning when we had it generate our list of blog topics, or podcast topics, right? “Did it give you an idea of the number of hits you would likely get from the subjects?” And it did not, but we didn’t ask it for that.
Nick Thacker 25:00
Correct. So if he’s going in, like an SEO, from an SEO standpoint, search engine optimization, that’s what I’m getting. No. But I, you know, I haven’t tested anything like that. But because it’s a very general intelligence, it’s something that it’s just an open prompt, I don’t see why I wouldn’t be able to craft a prompt and get it to zero in on some of that.
Kevin Tumlinson 25:18
That’s what I’m getting at. I think you could potentially ask it to identify topics and then rank them. I think you could actually do that, have them rank those topics in order of what … In that, relating that back to the author world, you could do the same thing for keyword research. You could actually have the AI go and find all the top sellers in the category of books that you’re producing, and have it produce for you a list of the top ranking keywords.
Nick Thacker 25:50
Okay, so there’s a caveat though, with all of this. We’re talking about, you know, let’s say, like Amazon keywords, if we’re just going to have an AI spit out a list of ranked keywords by order of how much they’re searched for. Don’t forget, you know, in using something like this, it’s got to have access to that data, which means it’s got to be public knowledge, where it’s already been fed that knowledge in this sense. Because they don’t want to create Skynet over here, open AI. So you know, and we’re not in any danger of getting there right now. But they’re feeding it a selected body of information. And that, you know, like I said, I mentioned Reddit, there’s other websites they’ve given it, Wikipedia, that kind of stuff. That information, it may not actually be there, it may not be in its database. And you know, keyword tools—like, you know Dave Chesson is a friend of ours, Kevin, you know. He’s got Kindlepreneur, he’s got a keyword tool, that’s proprietary software that this AI would not have access to. So Dave’s data over there is great, but it’s not like this AI can just go steal that and then, so it’s gotta have the data source somewhere out there.
Kevin Tumlinson 26:54
Right. And it’s not going to necessarily replace software like that. You’re going to be better off using that, because that software is fine-tuned specifically for that purpose.
Nick Thacker 27:05
I just mean, if you had to give it a prompt, it may not know where to find that information. It may give us a list that’s in order, but it could just be a guess you know?
Kevin Tumlinson 27:10
What it could do is give you a list of keywords that you could then plug in and try out on Publisher Rocket or a tool like it, and then that puts you quite a bit ahead. Because part of the process with Publisher Rocket is for you to bring in some keywords of your own to test out. So.
Nick Thacker 27:31
And I have used this thing for coming up with ideas for books, you know? Kevin and I write, you know, we write in the same genre, this action adventure. I wrote one sentence. And it picked up from that one, and I’m trying to find it now, I’ll dig it up. But in that one sentence, it picked up that I was writing a book description. And then it continued with an actual synopsis that didn’t exist. This isn’t something that it plagiarized.
Kevin Tumlinson 27:54
Yeah, it didn’t grab it from anywhere.
Nick Thacker 27:56
Yeah, didn’t grab it from anywhere. And it gave me an idea for a book. And …
Kevin Tumlinson 28:00
I read that and thought, you know, this is not bad. This is actually really good. There were some things I would change. One of the things I noticed about it was that it goes … Like, if you say, write a book description up to 200 words, it’ll get up to around 200 words, but it doesn’t necessarily finish. Like it doesn’t know how to close the loop, basically. So it just stops. So you couldn’t necessarily depend on this thing to just generate that stuff for you whole form. But as a sort of assistive tool, I mean, that’s profoundly useful. And the ideas were, the idea was kind of good. Like, I would just develop it a little. Um, you look for that. We have a question. This one comes in from D2D’s own Elyssa. “Weren’t people mad when they got duped by Google’s AI voice when it was booking appointments for people?” And if you don’t know what she’s referring to, when Google released its … I forget what they call it. Google, I don’t remember what the name of the service is supposed to be. But they did a sort of test when they announced it live, and they had it do things like call a hair salon and book an appointment. And the voice was so perfect. And the responses, I mean, right down to like the little audible cues you get from people when you’re talking on the phone. Like, you know, “heh,” you know, little things like that and it was so natural-sounding that it was kind of eerie. It was sort of uncanny divide territory, right? So, I’m sure people may have gotten mad about that. I don’t remember hearing like a …
Nick Thacker 29:43
“May.” Come on, let’s not pretend like people don’t get mad about literally whatever they want to get mad about. People are gonna get mad about stuff. But technology is going to advance, and …
Kevin Tumlinson 29:48
It’s true. And I think they cobbled the software a little after that, with like things like, “Hi, I’m a robot and I’m going to ask you for an appointment,” and things like that. Which I find kind of irritating, because to me, I don’t think we should have to call that out. Who cares if you’re talking to an AI?
Nick Thacker 30:08
People do. I found the synopsis real quick, if you wanted to hear it? It’s good. So okay everyone, I fed into the prompt—and, you have to just trust me here, I’m reading verbatim what it spit back out. All I fed into the prompt was National Treasure meets Indiana Jones. Now that’s a kind of a, I don’t know what to call it, but a dichotomy, I guess, that I’ve used for things like Facebook ads or whatever to describe my books in the past. So I just thought that’d be fun. Let’s see if it knows that I’m describing a book or a movie or whatever. And this is how it finished it. National Treasuremeets Indiana Jones in this thrilling new series for young readers. High-speed chases, cryptic clues, mystical adventure. The most valuable book in the world has been stolen from the Library of Alexandria. What was the book known only as the Codex, and why was it so important? 2000 years later, the answers lie hidden in the ancient city of Alexandria, and the only people who might be able to retrieve the Codex and solve its secrets are three teens from the year 2016. Kate, Michael and Sarah—it gives us names for these teens—arrive in Alexandria, Egypt, and are soon in over their heads. They discover that the Codex holds the key to all of the world’s ancient wisdom, and that everyone from the Italian mob to the CIA are after it. To survive, they’ll have to pass ancient trials and unlock the secrets of the Codex. But time is running out and the stakes are high: the fate of human knowledge and the fate of the world. The first book in a thrilling new trilogy—now it’s got me writing a trilogy, by the way—from the author of the best-selling …
Kevin Tumlinson 31:36
I don’t need you to commit me to a trilogy, okay guy? I’m already committed.
Nick Thacker 31:40
From the author of the best-selling Guardians of the Flame series. So that may be a real book. I looked up, I tried to copy-paste different versions of this to see if it was just getting a book series. And then just, so it didn’t plagiarize, just kind of re-wording it in on its own. But as far as I can tell, it’s not. Now, somebody out there who reads young adult may, that may actually be a book out there. I don’t know. But I thought that was utterly stunning.
Kevin Tumlinson 32:08
Did you say you Googled that? Guardians of the Flame?
Nick Thacker 32:10
I did. The Guardians of the Flame one might be, I think that was a book, but it’s not this book. So it just sort of grabs different pieces and then puts it together. But this is a full and finished synopsis. And, assuming it doesn’t exist already, it could be written, just from that.
Kevin Tumlinson 32:27
Yeah, yeah. I do like how it committed you to not just a trilogy, but an existing trilogy. Now you have to go back and write …
Nick Thacker 32:34
Now I have to go find it.
Kevin Tumlinson 32:36
Well, the thing, what I found interesting … So for an author’s standpoint, so we’ve already covered that it could be a sort of virtual research assistant, which I think is probably one of its more profound uses. The idea of AI voice technology allowing you to program your voice in, or the licensed voice of a contributor, and be able to knock out your audiobook, all under your own steam, that is intriguing from the perspective of cost for the author. Because, you know, presumably that would cost a lot less than hiring someone to do the work. I don’t want put narrators out of business, but I mean, come on. That’s hard to resist, you know?
Nick Thacker 33:23
Exactly. And we can’t look at it that way. We can’t look at it as, we have to fight against this because it’s going to put people out of business. That is admirable, but we have to recognize that it’s going to happen whether or not we fight against it.
Kevin Tumlinson 33:34
How do we leverage this as a tool rather than seeing it as something to be afraid of, or an enemy?
Nick Thacker 33:38
Yeah. To fight that battle really looks like, how do we get narrators to do something different that AI can’t quite cover yet? That’s sort of the, because this is this is happening. This is just gonna happen. It’s gonna continue happening.
Kevin Tumlinson 33:49
Yeah. And it’s going to evolve. I think, one of the purposes I would like to try to put this thing to is, if I can … I don’t know, you may know the answer to this. Can you feed this thing material for it to draw from, so if …
Nick Thacker 34:04
Not yet. Not the playground version. But they have an API and they’re making it more and more open. It’s called Open AI, is the name of the company.
Kevin Tumlinson 34:11
So if the material were on the internet already, it could do it?
Nick Thacker 34:15
Well, yeah. So there’s no mechanism they’ve allowed for us to do that in the playground. They’ve limited it. So essentially it’s, you’re just …
Kevin Tumlinson 34:24
But you gave it a URL and it went and summarized an article.
Nick Thacker 34:28
Correct. So I suppose that if you said, here’s the book in HTML format on this webpage, completely written out, you know, go read that and then … I guess what I was kind of interpreting your question as— and this is absolutely possible, they just don’t give it to us yet—to literally feed it our books. You know, to say, here’s the files, here’s all the books that I’ve got. I need you to build me a Nick Thacker-sounding prose library. And then I want you to spit out this synopsis as I would have written it.
Kevin Tumlinson 34:56
As cool as … Yes, that is one of the things I was thinking. The other direction I was going with it was, you know, I for one, like I’ve got 50 some-odd books out there. There’s about 14 books in the Kotler series alone. And what I could really use is somebody to go through and find the character arcs, and give me a summary of the book, the character, build the character history, you know? If I could feed that stuff into that AI and get that, that’s … And I think part of what we’re trying to do in this episode is speculate about how, what the future is going to look like and how AI will assist the author, instead of the very feared replace the author. How do we get into sort of a hybrid relationship with AI to help enhance the books? And that is one way that I see that happening. What do you, can you think of anything along those lines?
Nick Thacker 35:53
Oh, man, absolutely. Yeah, I’ve got a whole list of things that I’ve been keeping over the years, of ways that AI is gonna help me. I’m already committed. I’m like, it’s not if, it’s when they’re gonna be ready to help me this this stuff. I will say this. I, and maybe this is wishful thinking, but I think human creativity is sort of the last bastion of intelligence. You know, and that you can argue with that, I mean, there’s definitely some things … Obviously I’m being pretty vague. So there’s maybe some more specific ways of saying it. But I’m not worried about the AI all of a sudden being able to just write all our books for us, and then authors don’t have a job. I do think that’s getting closer to the realm of artificial superintelligence. And we might get there. And that might be something, you know, we can talk about 20 years from now, or 10 years from now. But the point is, right now as it exists, and as the direction we’re going, I think I’m still safe as the creative. And I can come alongside something like this AI, once it’s a little more developed, and have it become a working partner. And some ways I see that happening are things like advertising. And I don’t just mean, you know, Facebook ads, Amazon ads … Sorry. Gotta love dogs. Um, I think that means something like, you know, we’re in a world now, a very connected world, I think anyone would say. And in modern society at least, you’re getting bombarded with advertising. Nobody likes ads, right? We all say we hate ads. But I think the truth is, we hate ads that aren’t targeted very well to us. And so I imagine a world where we sit down on the subway to get to work, and the subway ad that’s in front of us interfaces with us somehow. And the AI that it’s running, the ad network, knows that … It knows who we are, it recognizes us, right? We’re already there in places like China. It knows who we are, it knows what we’ve done. It knows our habits, it knows where we’re going, it knows that we’ve just finished a project at work, and we need some, you know, some reading material, and then it’s going to advertise a book to us. And you know, we’re going to, I don’t know, click our tongue or something it’s going to download into our brains and we’ll read it, right? Which is better advertising than …
Kevin Tumlinson 37:57
Which is much better than the rectal path I was afraid they would take.
Nick Thacker 38:01
I’m not limiting them, you know? It’s, the AI can do whatever the AI wants to do. But ways of connecting, the ways we’re receiving marketing information is going to vastly improve. I always make the joke that like, when I’m watching TV, it’s been years since I’ve watched cable television with commercials and all that. But I always made the joke that, like, I don’t care about commercials, I just wish they weren’t advertising things to me that I’ll never buy. I’m not really into women’s razors, you know? I’m not going to buy that stuff. But you show me a commercial about an adult Nerf gun, and I’m on Amazon immediately, right? So if you can figure out who I am as a person, and then personalize the advertising to me. And the only way to do that, in my opinion, is with an algorithm. It’s with AI, it’s with a neural net. It’s not something that individual humans can do. Here’s one real quick.
Kevin Tumlinson 38:50
Yeah, you’re kind of talking about the idea of taking the existing sort of online model, where they use cookies and things like that, and tailor advertising to you based on your history. And then applying that to advertising worldwide, IRL, I guess, would be the idea.
Nick Thacker 39:09
Here’s one real quick, I know I’ve been talking a lot …
Kevin Tumlinson 39:12
And we’ve got some more questions I want to pop up too.
Nick Thacker 39:14
Cool, this will be short. I’m just really excited about this. Anyone who knows me knows that I started as a musician. I’m still a musician, I do a lot of music stuff for church and all that. But I have a music degree and so I’ve been following the music industry. And I have said this before, and I’m going to preach about it until I die. The author industry, the especially the indie community, follows the music industry about 10 or 15 years behind. We are all excited that Kindle was a thing in 2010, you know, whenever it started becoming a big thing. That was 10 years after the iTunes Apple Store kind of became a thing right? So in a very limited, shortened way, what’s happening right now in the music industry, SoundCloud is actually changing the way they’re paying artists, from a pool system where all of the money that SoundCloud gets, gets split up among all of the streams, and whoever has the most streams gets the most money, to a user-centric model, where if I am a SoundCloud user, I pay $10 a month to SoundCloud. But then if I only listen to my favorite band, Need to Breathe—shout out to Seth and those guys—then all of my $10 goes to Need to Breathe. This is incredible, because this is what we all want as musicians, this is what we all want as listeners. But the reason we can’t do it, you know, Spotify has come out and said, we can’t do it. It’s too hard for accounting, for computing power. It’s too hard, we can’t hire enough human beings to be able to do this, right? Ai can do that. AI can handle that situation, right? We write some software, we let the neural net do its thing, and it spits out how much money everybody’s paid. And it does the transactions for us. I’m predicting that Kindle, and potentially maybe it’s another store, but I’m using Kindle as kind of the 800-pound gorilla in the room, is going to start paying authors in a reader-centric model. So if you’re reading Kindle Unlimited $10 a month, and all you read is Kevin Tumlinson, then that $10 goes to Kevin Tumlinson. Rather than what it is now, which is the pool model where all of your $10 goes into their big pool. And then their millions of dollars get split up among whoever has, you know, the most page reads and they get the bonuses and all that. So let that percolate in everyone’s minds.
Kevin Tumlinson 41:29
I do like the part where the $10 went to Kevin Tumlinson.
Nick Thacker 41:33
I want the millions of dollars to go to me.
Kevin Tumlinson 41:34
We have this comment. It’s a question from Tom on YouTube, and this is very meta, this is a meta question. “So I can get AI to summarize my book about AI merging with humans to present to sellers’ AI to recommend to humans who are trusting AI suggestions? I’m in my element.” That’s true, though. As convoluted as that that question was, this is the sort of bag that this is going to open. Like, there are all these things that AI is going to be able to do for the author, we’re only sort of scratching the surface of it. And right now everything I can think of is almost all marketing driven. Or, like we said, doing the book description or generating the idea or something like that. I mean, you know, for a lot of authors, I think just generating a book description of the book before you write it would probably benefit them a great deal. You can also generate an outline, you know?
Nick Thacker 42:40
Or the blurbs or the book description on Amazon. You know, this is really, like you said it’s very meta, but think of it this way. Like, we’re designing, or we’re asking AI to design things for us, for human consumption. We’re also designing human things for AI consumption. And obviously, in the past, humans have designed for human consumption. This is talking about AI designing for AI consumption. This is talking about non-player characters entertaining each other in the book world, right? So this AI wrote this book series, and this AI is reading this book series. And we as humans are watching going, wait, okay, that’s creepy. But think about it this way, like Pixar is all computer-generated graphics. Why couldn’t an AI design and write an entire movie that is then generated for human consumption? And Open AI is digging into this, if you guys want to search around a little bit, their website has—they don’t allow access to this yet, but they’ve got examples of their AI, again, this is GPT-3, but putting it in a different box over here—doing some creative things like art, doing some creative things like music, and as a musician, I’m fascinated. This is, it’s good stuff. I’ll just go ahead and say, it’s good music. It understands form and structure and how to put sounds together.
Kevin Tumlinson 43:56
Well, and okay, so here’s where I think, though, that it’s going. So now it can generate that sort of thing. It could generate music, it could generate deep fake films, animated films. You know, that’s in the realm of possibility now. Where I think it’s going is, as AI becomes more intelligent, and our use of it becomes more sort of natural, we’ll be able to do things like, you know, I’m sitting at home and I want to watch a film on a certain theme. I can say, you know, “Hey, Siri, generate …” Oh, I’m sorry, Siri, I should know better than to do that. “Create create a movie in the style ofSteven Spielberg that’s about this topic.” And then partway through the movie, I can say, “Look, let’s change this, you know, add a new character,” and you know, it becomes more like a holodeck experience.
Nick Thacker 44:55
I want more John Williams music, not Hans Zimmer music.
Kevin Tumlinson 44:59
Exactly. Or the flip side of that is the film production community itself has AI more advanced than that, and they’re directing it as they go. So Spielberg is saying, “Stop, add another dinosaur, this time with wings,” and then go. And then the AI unfolds the story. And the human is basically just directing, I see that as being a very real possibility, which means books would be generated the same way. So there’s still the human-created …
Nick Thacker 45:32
I think books would be first, because it’s simpler. It’s gonna be books first, right? It’s gonna be the ultimate choose your own adventure. You’re going to have an author who is partnering with an AI. And I’ll get to the obvious question here in a second. But, you know, an author who’s partnering with an AI, who comes up with a framework, right? Instead of a template or a blueprint, it’s more of a framework that is loose, but has some boundaries. And those are set by the author. The fantasy writer, you know, gets to create the world and gets to explain the rules of magic in that universe. And then the AI gets to create the book. But eventually, readers will be able to interact with that book in different ways. Well, they’ll say, you know what, I don’t actually like this character. Can we rewrite this so that this character dies instead of this character? And let’s see how that spins it. The obvious question that we’re getting to is, will we get to the point where AI is the owner of intellectual copyright and intellectual properties, right? And so will I own that book as the “author”? Or will the AI who actually wrote the book based on my framework own that, right? Or will I own the AI, and we get to a [inaudible] situation?
Kevin Tumlinson 46:37
That’s the can of worms that we’re currently opening up and you know, there aren’t any answers, because this is the first time in history we’ve had an opportunity like this. So we’re at the end, so we’re gonna go and wrap up. But just in summary, the ways in which AI can affect the publishing industry, and authors in particular, are innumerable and profound. We’re not going to know all of the ways that this can go until we start really digging in and experimenting with it. But just as it is right now, I could see it as a very useful tool. Do you want to throw in any last tidbit of knowledge or wisdom before we wrap up?
Nick Thacker 47:16
Um, you know, I feel like we’re talking about a lot of things that are potentially scary things, Because, you know, it’s fear of the unknown and humans are, if anything, we’re good at being afraid of things we don’t understand. Guys, don’t be scared. This is something that is inevitable. So it’s one of those stoic, you know, hey, it’s gonna happen. So don’t worry, don’t be scared about it. Don’t have to be on the cusp of anything here. You don’t have to, like, you know, I’m not telling everybody to just go be on the bleeding edge of AI. But pay attention, because there’s some ways that this can really benefit our author careers, you know? And by pushing against it, or by not wanting to have an AI read our audiobook, isn’t going to prevent the AI from being able to do that. So pushing against it is, we’re being curmudgeons by doing that. Right? So yeah, embrace it. Understand it.
Kevin Tumlinson 48:06
Be a Nick. Don’t be a Kevin.
Nick Thacker 48:09
Yeah. Be a Nick, don’t be a Kevin. But yeah, this this stuff is exciting. It’s definitely scary because of the things it’s capable of. But just remember, I think creativity is the last bastion of human intelligence. That’s going to be sacred to us for a long time, I think.
Kevin Tumlinson 48:25
Yeah, I agree.
Nick Thacker 48:28
So rest easy.
Kevin Tumlinson 48:29
All right, everybody. Thank you so much for tuning in. If you haven’t already, make sure you subscribe to us on YouTube at youtube.com/draft2digital. You can also subscribe to us here on Facebook—I say here on Facebook, you’re either watching on YouTube or Facebook. You can subscribe to us on both. Facebook.com/draft2digital. And make sure that you are sliding on over to selfpublishinginsiders.com where you can follow, you can find all the previously recorded episodes of this show, the Self-Publishing Insiders podcast, complete with transcripts, and a whole bunch of other goodies as well. And we’re furiously building behind the scenes so that we can make that an even more amazing experience for you. So thank you so much for tuning in. And I think that’s gonna wrap us up for this time. Thank you, Nick, for sitting in and chatting with us about AI. You are a very convincing AI yourself, sir.
Nick Thacker 49:31
Thank you. Thank you. That’s good that my deep fake worked. Just two Kevins talking to you today.
Kevin Tumlinson 49:35
Two Kevins talking. All right. Take care everybody. We’ll see you all next time.