By Use Cases
In this episode, James Timberlake, executive consultant at J. Galt Financial Suite joins host Tara Shankar to explore the fascinating world of generative AI in independent publishing and ghostwriting, and discuss how AI can streamline processes, enhance content creation, and revolutionize storytelling. Listen to gain insights into the benefits and limitations of AI, the importance of human input, and the ethical considerations surrounding its use.
TJ – 00:00:03:
Generative AI takes the center stage. But is your enterprise still watching from the sidelines? Come on in, let’s fix that. This is Not Another Bot, the generative AI show, where we unpack and help you understand the rapidly evolving space of conversational experiences and the technology behind it all. Here is your host, TJ.
Hello and welcome to Not Another Bot, the generative AI show. I’m your host, TJ. With us today is James Timberlake. He’s a multifaceted professional with an impressive career spanning business brokerage, transmedia production, and most notably, AI as an independent publishing. James is the co-founder of Noob Studio and MetaSkys Inc, as well as the founder of the Story Sherpas, an AI ghostwriting service that has reshaped the independent publishing’s landscape. With a rich academic background in philosophy and artificial intelligence from revered institutions like MIT and Yale, James brings a unique perspective to the intersection of AI, creativity, and publishing. A prolific author, AI enthusiast, and forward-thinking innovator, James isn’t just riding the wave of AI evolution in publishing, It’s actually shaping its course. Welcome, James, to the show.
James Timberlake – 00:01:18:
Thank you so much.
TJ – 00:01:19:
So let’s get started and the way we do this is, James, we would love to know more about you. So the first question is always about you, so could you actually walk us through your journey all the way from academic studies in philosophy and AI at prestigious institutions to becoming a thought leader in the field of AI in independent publishing, especially as a founder of the story, Sherpa, and an author?
James Timberlake – 00:01:42:
Like most things in life I imagine it kind of happened by accident, right? So yeah, I started, graduated seminary and thought that was going to be my life, but had some interesting interactions with some of my fellow classmates and it kind of left me a little disillusioned in some of the practical sides of ministry, if you will, people that saw it more as a a job than a calling and so that kind of changed the direction a little bit. And that’s what led to me getting a second degree in business and philosophy. And so started off in retail, got robbed at gun points I was like, I won’t do that anymore. So that’s when into like inside sales, which led to marketing and project management and yeah, it was crazy, so it’s like, it kind of just progressed from there and eventually got to the point where, you know, I got to work on some really fun projects that helped me kind of expand my horizons on what I was exposed to and eventually led to, as you know, it was just a ghost writing shop. You know, we did everything manually, we just helped people, professionals, celebrities, high net worth individuals tell their stories either to become a thought leader or they wanted to leave legacy or something of that nature. And what we started doing is just slowly integrating more and more automated systems. So first we automated the transcriptions, right? And then the transcriptions, we were like, well, now that we have those, what can we automate some of the editing? And so breaking it up into sections, well, can we start automating some of the interviews, right? And start doing the video interviews. So we just kind of started slowly adding, you know, different aspects of what our workflow was we just kept adding different systems or expanding the systems we had until we could automate more and more of it. And, I eventually sold that company and it went on and has a new life now, which is great and that freed me up to do a lot of my own kind of personal experimentation with writing and things like that, I actually wrote a book with 100% AI about three years ago. I think technically it might’ve been the first, but it was published, but it never did anything with it as far as like submitting it for AI or whatever, so nobody even knew that it was an AI book and that was kind of the point. But more recently I’ve been, I had a lot of, uh, exciting conversations, so MetaSkys, as you mentioned, you know, that’s my kind of one of my current companies, a co-founder of, and we’re working with what we call, uh, personalized generative entertainment. Right. And so one of the companies that you might be familiar with right now, that’s just doing really groundbreaking work with this is a fable simulation and if you’re not familiar with the name or if your audience isn’t, they might be familiar with their case study that they’re about to pull down actually. But right now you can still see it where they have a platform where you can create your own South park episode, right? And you go on there, you tell them exactly who you want in it, you want the scenario to be, you know, I want this guest star or I want this kind of thing, and it actually builds a thing. And so I’ve been talking about that concept for the last two and a half years pretty often, and you know, most people didn’t, it’s never going to happen. Then all of a sudden, you know, chat GPT came out and mid journey came out and, you know, uh, Dolly and all these and all of a sudden everybody’s like, oh, maybe, but it’ll be like five or 10 years. I’m like, no, you don’t understand. And now people are starting to see it and they’re like, Oh, that’s so anyway.
TJ – 00:05:20:
That’s what you were talking about. I know, totally. Awesome. No, this is brilliant.
James Timberlake – 00:05:25:
Yeah, but it’s exciting to see because I think, right now, especially with the writer strike that’s in Hollywood, and now the actor strike as well. But these are not new conversations especially the writers have been underpaid for years, they’ve been fighting this for years. But now the actors are starting to see that, oh, well, I’m in trouble too because now you can take my license, my image, and you can license it and so now, take Marvel as an easy example. I no longer have to pay Chris Hemsworth increasing amounts of money every year every time I want a Thor movie and also have to worry about the fact that he’s supposed to be a god that doesn’t age, but he ages every year. So now we’re spending money to de-age them back, it’s like, well, at some point, you could just do a life model and just create them and then you have the character rather than the actor. And that’s what the studios are really fighting for, they want to own the characters and the digital likeness of those characters. And believe it or not, AI has now become the sticking point in these contract negotiations. it’s like, they’re not even fighting about writing wages, any of this. The main sticky point is AI of all things because they see it as an existential threat, which it is.
TJ – 00:06:38:
You’re just taking somebody’s, I wouldn’t say the job, but precisely saying, Hey, you know what? I mean, I need the character because for which you certainly were known for, but I need to ensure that I could use it for years together and I could do that with AI. So totally. I mean, for them, it’s like, okay, what am I going to do now? If AI can do it, the whole damn thing, so make totally make sense, interesting. So on that similar thought, then what excites you about the potential of AI and automation in independent publishing, particularly in the context of course writing with AI?
James Timberlake – 00:07:10:
Well, for me, in fact, I just posted about this on LinkedIn today. I was like, for me, the thing that’s most exciting is that it’s going to, it really does free up the independent storyteller. And I use the term storyteller because, you know, a lot of people, you know, they may not be writers, maybe they’re an artist, but they have a story, or maybe they are writer, but they’re not an artist. And they want to kind of, this allows you to start filling in your gaps with AI so that, you know, anybody who has a story, which, you know, after almost a decade in some form of ghost writing, everybody has a story, right? Not everybody knows how to tell a good story, but everybody has a story worth telling. And this is, that’s what’s exciting to me is people being able to in the short term, tell their stories to other people. Right? Like one thing that I just saw was somebody, and this is what we’re one of the things we’re working on at MetaSkys as well, is being able to go in as a complete novice, you know, just a fan and say, man, I’d love to see a movie about this, this, and this. Maybe it’s based on my life or an experience, or maybe it’s just an idea that I’ve had forever, but I don’t know how to write a script, I don’t know how to do any of this and it’s like chat GPT can do some of it, but it can’t put out a whole script yet or anything like that. But you can start piecemealing some of the stuff together and what you can do now and somebody actually just did this today. That was a really good job is you can do now pretty easily with the, even free tools is put together a proxy or a pitch video. Right? So I’m complete novice, but I know that there’s a inkling of an idea of a story that I’ve never seen on, you know, whether I want to represent the culture I’m coming from or my unique background, or maybe it’s just something completely fanciful that I’ve never seen that I’d like to see. Right. And so now I can go and I can use these AI tools to take the images and the stories that are in my head and I can put them together into a 30 second little, you know, either a, what I tell people is you can either do a 10 page comic, cause that’s pretty easy to do, you could do, if you’re a little bit more technically, you don’t mind doing some more of the dive in and doing something like taking mid journey and putting it into runway, something like that. Well, now you can take your comic and you can convert it into kind of like a motion comic, if you will. And if you do a little bit more iteration, well now you have a trailer, right? And you could take that trailer and you can build support, go get a Patreon or whatever, and then you can build it to be as big as you want. And so now it’s no longer, Oh, I have this great idea, but no one in Hollywood would ever listen to me. You don’t need them to just go make the thing and then make the next version and the next one. And eventually you can build your own audience and you can be your own mini studio. And that’s really where, what’s exciting is we’re going to see one person, two and three person studios, taking over with this massive, with these new massive IPs, you know.
TJ – 00:09:52:
Wow. That’s definitely visionary. And I think it’s with the possibilities of what AI could do and not with the generative AI even more. Well, it’s not new, right? I mean, generative AI has been there for the longest time, but just that the application of it is a lot more easier to adapt to after ChatGPT. But I think, you know, with GANs and VAEs, you know, things were done like music generation or image generation but I think now it’s a different world, but pulling everything together, including a script, I mean, that’s just massive, totally an eye opener in that way. So now that we know about this, as the founder of, you know, the Story Share prize, you have that firsthand experience with AI in the publishing industry and glad to know that you launched an entire book some years back, just with AI. But could you share James, your journey integrating AI into your business model? What were, you know, some of the key obstacles when incorporating AI and how did AI help in resolving those while on the go? That’ll be like, you know, just in addition to what we’re talking, but predominantly, how does it really impact your business model overall, as you went through the journey?
James Timberlake – 00:11:01:
Yeah, so StorySharp was in particular, the way it affected was just the ability to, really two main takeaways, as far as how it helped us. One was it made it a much easier to scale because we weren’t having to add people, we could increase the same level of support for a lot more people without having to add staff, right? Or freelancers, so that was huge. And the other thing too is it made existing workflow, it just took out a lot of the busy work, right? And it took out a lot of the smaller little piecemeal parts. So like I said, the transcription, right? For those that don’t know, there’s a lot of ways you can get transcriptions, right? And one of the more popular ways is using something like Rev, which is a service that basically you send a recording to, somebody listens to it and they type it out, right? And that way you get, and it’s really accurate, but it’s also expensive and kind of slow, right? And so it’s harder to scale that as a smaller business and the place I worked before, like they would use it too, but a lot of times they would also use in-house people, right? That that was their whole job is just listening and transcribing. And, but with AI as that was getting better and better because, and so this leads into the obstacles. What we found though, the hard part wasn’t necessarily understanding where AI could help us, it was understanding which AI tool was the best for that particular purpose. You know, what would be nice is one platform that just does everything. and I think eventually we’ll have that, but we don’t really have that now. We certainly didn’t have that a few years ago, right? So when we were first starting, we were using things like, it wasn’t even, they changed their name since then, but I think it’s copy.ai now, but they had a different name at the time. But like I said, before the rebrand, I don’t remember what they were called, but that’s how early we engaged them because, but they were a marketing, they did marketing copy, but we were like, yeah, but you can we use this to take our transcripts and, you know, improve them to do things like back cover copy or to do things like, you know, emails, letting people know like that we’re taking pre-orders or announcement emails or like things like that, like, hey, we had like three great reviews, can you help us format them, you know, in a way that’s gonna be really impactful or like things when you’re doing like your Amazon description, things like that, right? So not having to hire a copywriter, being able to use a system like that to give the client a lot of starting point draft options is huge because, you know, we can do that. Well, here’s five, which one do you like better? Let’s refine from there. And we can do that over and over and over instantly, as opposed to let’s have a meeting with your copywriter, let’s help them understand, let’s make sure they read the book, right? And this is the old process, like you would read excerpts from the book, you’d get an interview with the author, and then you’d go and you’d write up a draft or two, and then they’d review it, and then you’d go edit it, right? So the whole thing would take a week or so to do it right, now you can do it in a meeting, right, in real time. So that was one of the first ones we did. But what we found is it was horrible at editing, like long form, it just wasn’t there, right? So we had to go and find a completely different system that was able to do that. And so that’s the obstacle I think even still today is just the vetting and testing of all the different platforms, especially now it’s almost worse because there’s so many AI platforms and tools, you know, half of which are chat GPT with a, you know, a front end, which is fine, you know, sometimes there’s value to that front end, but you know, a lot of times it’s not as, sometimes the marketing is a little bit better than the result, you know what I’m saying? So you gotta do that due diligence before you just start putting it into your workflow or before you start putting it on to, with a paying client. And so that was the biggest obstacle is just dedicating that time to do a really good deep dive investigation onto the options and then test those options in some type of real life scenario to see what kind of output you get, and then be okay to go, God, that sucks ,we need to start over. But if you’re an entrepreneur, you’re probably used to that in general, but like to me, that was, I think, probably the biggest obstacle is just that took a lot longer and it took a lot more piecemeal tools than we expected. So then you gotta make sure they all fit together, right? So that’s the other, that’s kind of the secondary obstacle and Zapier is a lot further along now with its integrations now. So a lot of that can be solved pretty easily now, which is Zapier. You know, again, there’s those types of things, like it’s those little tiny things, that’s where your break points, or at least for us, that’s where our break points were and that’s kind of like the more difficult part to navigate. But again, the trade-off was worth it because once it worked, it worked well and it saved us so much time, we could then go and move. So we just dedicated a certain amount of time to figuring out one problem and then we’ll figure out the next. And so eventually you get into a rhythm and then you just start adding your stacks as you go and it’s a snowball and it just gets easier.
TJ – 00:15:51:
I think one thing that comes out of that wonderful narrative you just landed, that there’s so much information in golden nuggets right there. But how does AI in this context of publishing, host writing, and beyond? With its ability to analyze and process vast amounts of data contribute to a more streamlined and effective content curation and also the vetting process, whether this is kind of, I don’t know if the copywriting is a concern anymore, or yeah, should be predominantly, but how are you vetting that, or how are you creating that effective content? And you know that this is the sort of content I need. So what’s that, looking at that amount of data which you have to deal with, how is that kind of done today? And what’s your take on that, James?
James Timberlake – 00:16:34:
So basically what I would say is, I think this is answering your question, hope it is, if not, just ask me. But basically, I think right now, a lot of people see, especially in publishing and marketing in general, any written content, people think that it’s like, AI is gonna write it for me, and it can, but it’s not gonna be good, right? Like, and even the parts that are good are gonna be generic, right? And so you can just start with say a prompt of write me about this, but we found is far more effective, at least in our purposes, is having the person write, and it’s almost like we can just, you don’t have to write well, the AI is gonna fix it. I just need you to write from the heart because that’s the part the AI doesn’t have, right? So you don’t have to worry about grammar, you don’t have to worry about sentence structure, it doesn’t even have to bloody make sense to you, like, you know what I’m saying? Like, I don’t have to read it necessarily, I just need it out. And so that’s really freeing for a lot of people, believe it or not, and that’s one of the things we used to tackle even when we were doing it manually, was we would just say, like, look, we’re gonna have a writer come in and fix this for you, right? And so I think for a lot of people, that’s a huge opportunity to go, well, just write, or even better, you don’t have to write, just speak, and let it do real-time transcription. Just get it out of your head and then feed it to something like chat GPT, or there’s other programs too. Some now that’ll even take it straight from voice and, and turn it into. Five or six different formats right off. But my point is, however you wanna do it, depending on your goals, you can now just kind of get it out in its rough form and it’ll reformat and it’ll pull out that context and it’ll pull out and it’ll reword it in a way that is more or less aligned with whatever format you train it on, right? And again, you can do the easy way or the hard way. You could just plop it into chat GPT and say, hey, help me edit this and you can give us some parameters I want an eighth grade reading level because that’s what you want most of your publishing to be unless you’re at a professional level. So, I bring this down to an eighth grade reading level, fix any grammar, shorten it by 20%, get rid of any redundancies and any areas that you see that I need to add more context or, cause you could talk to, like if areas that have more context, prompt me to do so, right? And so then it’ll respit you out a different format, so take that draft, edit it, change it, repeat that process till you’re happy. That’s the basic simple way anybody could do even with the free chat GPT, right? And then if you have the ability to do a custom trained model or even just using custom parameters within chat GPT, I guess you could do it too, but the real power is if you can do custom trained models because at that point, then what you can do is, especially you have a, if you’re working with a, an author that has some kind of established body of work or even like a thought leader who maybe has done a lot of tweets or they’ve done a lot of blog posts or LinkedIn posts, something like that. Well, then you can take all that content and feed it as a custom training data set and then you’ll create an author profile, right? So then as you’re starting to feed it, it’s not pulling from the entire, you know, large language model and trying to, it’s saying, no, no, no, you write like this. This is your style, This is your, these are the things, this is your cadence, this is your voice, so I’m gonna do my best to copy that. And so the more you can train your custom models, then the closer it will be to your voice, right? And that’s what I, to get back to what’s exciting, I think for personal publishing or independent publishing, that’s gonna be probably a pretty big thing because people are gonna, and I also think instead of books, you’re gonna see more like newsletters because people want this ongoing conversation as opposed to like a single, but for something even, you know, whether you’re doing a book or a blog post or a newsletter or whatever, I think you’re gonna see more and more people building out these custom avatars, if you will, and then using them as co-writers, right? So again, whether I’m feeding you an idea and then I have it build it out into a full blog post, or maybe I just do a journal and then I say, hey, here’s my daily journal, convert this automatically into one LinkedIn post, one Twitter thread, and then also go ahead and insert a chapter into Chapterly for me, right? So that that can be expanded into a full chapter, but at least I have my draft. So you can get like, you know, 10 pieces of content, some complete, some drafts out of one conversation or one free writing session. And so as an independent, you know, author or even just content creator, you know, this unlocks a lot of potential to just, you know, every day create 10, 20 pieces of content from an hour as opposed to what you had to do before, where it’s, you know, usually is more iterative, you know, you would release the book first and then you would take a chapter from the book and you’d put that into a newsletter and then you would take some sections out of that newsletter and you’d use it for social, you know, you break it down that way. Well, now you don’t have to do that, you can do it all at once from a single source. And so the volume of output is going to be, I think, a lot higher as more and more storytellers, authors, just, you know, thought leaders in general, start embracing that as a kind of a co-author.
TJ – 00:21:43:
Does answer the question. I think it really brings a lot of different perspectives beyond what I was asking. So thank you so much for going deeper. I think keeping the accuracy and hallucinations in mind too, given that’s equally critical. And given, you know, ghost writing has garnered so much of attention recently, doesn’t really create that sort of quality content with, you know, less hallucinations and to what extent can we trust the AI to produce more compelling narratives? And second part of that question is, I think you touched upon it, like, you know, bringing that unique style of the author, whether it’s a writing style or, you know, the tone, the quality and beyond. Is generative AI able to accurately replicate? And those will be my like follow-up questions from what we are discussing at the moment.
James Timberlake – 00:22:27:
So the short answer is it’s like most things it’s a matter of, it’s like a scale, right? It’s a spectrum. So again, if you’re talking to something like your basic chat GPT or even GPT four, yes, it can, but it won’t be as good as say a custom trained data set to train specifically on, you know, like if I, for example, let’s take Stephen King, right? Who’s got a large known set of work. Well, if I were to ask chat GPT to take a story that I wrote and gave it and say, rewrite this in the form of Stephen King, it would probably do an okay job because there’s enough material in its training to make that work. But if I were to come in and say, the example, if I were to take like a chapter from Carrie and drop it in and say, make it sound like me, it’s probably not going to do as well, right? If it doesn’t know who I am as a new writer, right? As an example, so that’s what I’m saying it kind of depends. So the more you can train it, then the better it’ll get is the short answer. So, and it’s getting easier and easier to do that both with, although now going to the hallucination part, there is a caveat. Again, this is why what we say is we don’t look for chat GPT to create information, we look at it to polish information. So I look to the author, the thought leader, that person, I want the information to come from their brain. I just want the computer to clean it up. Right. So for me in ghost writing, limiting it to that context, that’s the proper order of things, because then it’s, it’s serving a very specific goal and you’re far less likely to have hallucinations because you’re not asking it to fill in blanks. Right. But if you’re trying to flip it around and say, well, you know, I wanted to do all the writing for me, I just wanted to give it a topic. Well, the more freedom you give it, then the more likely you’re going to have it create something, you know, that’s not accurate or that’s maybe partially accurate or maybe is bias, right? Because again, you’re letting it free rein and you got to remember that a lot of these, or, you know, any of that you’re going to use publicly, they’ve all been trained by and large on just public sources in the internet, so it could go all over the place. And what’s interesting is over the last six months with, at least with chat GPT, we’re seeing it actually get worse as they put more and more safety measures in place, it’s actually having an interesting effect. We’re also seeing more hallucinations, we’re seeing lower quality output. There’s a lot of theories as to why this is, but I don’t know the answer, I just know that objectively, it seems to be getting worse. And a lot of people are talking about that, but is it theoretically possible? Absolutely. It’s just how much work are you willing to put in to get it there? But as of right now, you know, you can’t just say, write me a Stephen King book and expect it to spit it out with no work ,we’re not there yet and maybe we will be, but we’re not quite there yet. But you can, you know, with a little bit of effort, you can definitely get an entire chapter out of a conversation like this, or several chapters out of a conversation like this that are edited and formatted properly and can be used as a basis for a book or content that really, really truly does reflect the voice tone and opinions of the author because it started with that.
TJ – 00:25:38:
That’s beautifully explained. I think one of the things which touched upon, I think there would be these, you know, limitations, challenges with, yeah, we’ll somehow face whether it’s the accuracy at that level, but the custom data aspect of, you know, how on which you’re training the model. And that’s what probably we also need multi LLMs to even do different things, maybe summarization separately, or, you know, running a complete Q and A, beyond that. Any specific limitations or challenges that comes to your mind today, James, that the industry is facing, let’s say with ghost writing or publishing. And if there are at this moment, how are you overcoming those or how, what’s your thought around those? If you’re seeing that pretty often, maybe on a day to day basis with your work, we’d love to know that for sure.
James Timberlake – 00:26:21:
Other than some of the stuff we just talked about as far as like limited in what it can produce based on a given input. I think that probably the other issue that we’re looking at, or at least in my circle, we’re kind of thinking about long-term is they’re interrelated, but there’s a couple of things. One is kind of overuse, right? Like it’s hard enough to stand out as it is. But when everybody’s using it to game the system, you kind of flood the, as it were, you flood the market with all this content, most of which is created now, or more of which is being created by a computer, which becomes its own problem because then that training data gets pulled in for the next iteration, so it starts training itself. And so that’s not necessarily what you want, right? Like you want more human input and so the one fear is that, well, so much of writing nowadays is done for monetary reasons, whether you’re writing a TV script or you’re writing a blog post or you’re writing marketing company. At the end of the day, so much of it is about trying to convince somebody else to pay you for that content in one way or another, right? And so if that’s the goal and now AI can just spit it out, what you end up with is like, well, I’ll just write, instead of writing one blog post, I’ll write 20 blog posts and I’ll A-B test all of them. And so I’ll just push out more and more and more and the quality becomes less important because you’re going for just quantity. And eventually something’s going to hit, right? I mean, this is how marketers test ads for time immemorial and that’s fine. But now we’re seeing authors do it and we’re seeing thought leaders do it, right? And LinkedIn people. So it’s just like there’s more and more. And the quality is going down because they’re doing less and less actual thinking, less and less actual writing. And that’s one of the, I guess, concerns or fears is finding that balance of like, yes, it’s a great tool and it’s a co-writer and it can be an editor and it can be a publisher and it can help you. But just like I was saying before, it still has to come from a plate, like, the way I would describe it, it has to come from that human element, that heart element that the computer doesn’t have. They can out-think you all day long, can’t out-feel you at least, you know, not yet, maybe one day, but right now that’s your advantage, is you had a lived experience that brought an emotion to you, right? Whether it’s in a business or personal context or even a religious context. And if that’s what you’re sharing, then the AI can enhance it but it’s not a replacement for a lived experience, you know what I’m saying? You can’t just be like, create me a week’s worth of content for my business. It’ll do that, but it’s going to not be great, it’s going to just be a week’s worth of basic content, like every other person out there that’s saying, give me 10 weeks worth of basic content. And so I think that’s one of the issues right now that I think is a lot of people are starting to see on the surface. And then, you know, as far as like, how do you deal with that? Well, I don’t know if I have an answer to that yet, because some people are coming out with AI detectors, but they’re not overly accurate. Even OpenAI pulled theirs because they said it didn’t work, so we’re not going to keep investing in it. And the thing, you know, I make a joke a lot where I’m like, you know, now the thing that’s becoming more and more popular right now is based off what we were talking about earlier, this kind of virtual clones, if you will, right? Where people are training their own, like mini thems. So it’s kind of like a digital VA, except if it was trained with your brain and so it’s like, they’re going out. And so now I’m actually working on one for myself, but you can actually create a little mini version of you and put it on your website or on your Slack or whatever. And then when customers come in or friends or family, for that matter, and instead of talking to you, they talk to the digital you. And so they can converse and it’ll answer as if it were you pulling from your knowledge base that you’re updating, right? And this is great for, you know, executives or thought leaders or authors that want to, or actors or whatever, that want to engage with their fans in a personal way, but at scale. And it kind of OnlyFans is another great example of people that are using this stuff now. Because it gives the illusion of intimacy and a real life human to human connection, but at scale without actually having to do it. So that’s a little bit of a concern for me because we’ve already seen people that can’t distinguish the fact that this is a bot versus a human. And so they start forming real human emotional connections with a bot and then they start projecting that onto the human rather than the avatar, if that makes sense. So for example, if I was, who’s your favorite? I don’t give me an artist or a celebrity of some type that you’d like love to meet. So let’s say Tom Hanks makes a digital Tom Hanks, right? You know, America’s dad or whatever. And it’s great. Everybody goes and they talk to him about advice, they talk about his movie career, they want him to do the Forrest Gump quotes all day long. It nauseam and it’s happy to do that. But there’s one or two people that, you know, they’ve grown up with Tom Hanks and they really, they just, they really see him. And, you know, if I could just meet Tom Hanks, be best friends, like I know me and I’ve seen him, I know him, we’re going to just hit it off. And then you go on and you start talking to the digital AI. I mean, the digital Tom Hanks who is programmed to make you feel like Tom Hanks cares about you. Right. So then you started doing this and so it starts feeding your own, you know, like, oh, I knew we would hit it off, see, and then, okay, you say, now you build this six month relationship with fake Tom Hanks. And then you go to a, you know, a movie premiere or whatever and there he is at the red carpet. Yo, Tom, yo, man, it’s me. You remember me? You don’t know who the hell you are, you never talk to you, but you have that emotion. So, and then it’s like, well, now you’ve done hurt you. So anyway, like those are the things that I think about. I’m like, you know, especially when you hear about the, it’s actually worse in other countries in America, you know, you’re hearing people that are having full on relationships with, virtual spouses or, you know, virtual girlfriends or boyfriends. And they go, you know. Why would I ever want a human? Why? There’s too much hassle. They want to eat all the time, they’re mean to me constantly, you know, they get pissed off every time I make a mistake, you know, they have all these needs and whiny things, you know, they want me to spend time with them on their schedule. Right. You know, but the AI bot man, they’re on my schedule, they’re always nice to me, you know, they never have a bad day, they don’t mind listening to mine all the time. Right. So it’s like, it’s just a better relationship. Why wouldn’t I want that? And so like, you know, in the long, term things, like if, you know, if I’m like really play it out, like my worry is that we’re already an isolated, more and more becoming an isolated society where everybody lives in these little bubbles. I mean, even you and I, like we’re talking through screens right now, you know, but give it two years, we might not be having this conversation, digital me and digital you, they’re having the conversation. Right. And so then, okay, well, that’s great and that sounds fun for like, when you’re talking about a celebrity and a fan, but what if it’s one executive? I don’t know, what if it’s an Amazon executive and a Microsoft executive, but neither one of them are talking, it’s just their AIs. And then they start making decisions that affect multinational companies that then affect human beings across the globe. It’s like, well, okay, is that what we want? Do we want to personal AIs having these conversations and doing contract negotiations at the speed of, you know, 100 iterations per second or something like that? Just saying like, I want the best deal on this side and it goes, it’s like, you know, two little lawyer bots going back and forth until it gets to the final thing. And then nobody questions it because well, the AI said it’s the best. So like, those are the things long term, I think, for any industry can be problematic is people put more and more, they offload more and more of the decision making, rather than the execution to the AI. And I think that’s a mistake, because whether it’s talking about publishing, or you’re talking about finance, or you’re talking about any other business, you know, because also do some defense things, things like that, like the problem in my mind, this is, you know, opinion light, but the problem in my mind is, when AI is a really, really good, it’s like money, right? It’s a really, really good servant. It’s not a great master, right? Like tell it what to do, and it can do it, it didn’t do it right, iterate, and it’ll do it over and over. It’ll never be offended that you’re trying to tweak it, you know, like those like clients that you always hear the horror stories, like, it’s never good enough, they want 1000 revisions, right? Like, that’s what AI is perfect for, it’ll do 1001 revisions with a smile. But what’s not great is that when you start giving it the ability to make those autonomous decisions that affect other human beings, that’s what scares me because it’s such a natural impulse to just delegate that responsibility off because well, the AI is smarter, or it has access to more information, and those things are true. But like in the medical field, we’re seeing that. Already starting to happen because we have these diagnostic systems now that can pull from real-time data, all the medical information that’s in the system, all the doctors that are feeding it daily. And it can look at a list of symptoms and it can put out a diagnosis, not only like that, but it’s usually a lot more accurate than the initial human diagnosis, right? And to the point where they’ve even tested it on coma patients and they found that when they did a side-by-side comparison, I don’t remember the exact statistics, but it was something like 70 or 80% of the patients that the human doctor said they had a very low chance of waking up and there was an AI that said, no, they have a high chance of waking up. And so they defaulted to the AI and they actually started waking up I mean, I’d have to go back and find the source for that ,you can Google it, but the point is it was a fascinating thing. So there’s real opportunity to be had there, right? So that people can bat some of that human bias where we might just give up on somebody or we might misdiagnose it because it’s like, well, what are the odds that they have this super rare condition, right? It’s probably not that it’s probably indigestion, go home and then, oh my gosh, they die because they had this thing nobody knows about, right? But now every hospital has a house, if you remember the show House, like now every hospital has one. And so when you look at it that way, this is a powerful, powerful vehicle for delivering information, but if it’s actually doing the full diagnosis and then prescribing medicine and then just like, we could get to the point where you have automated medicine and a lot of people are pushing for that, right? Well, I just hold up my phone, it does a body scan and it just tells me what’s wrong with me and then it orders all my pills and Amazon drone delivers them tomorrow, or probably tonight. Right. And so maybe there’s an advantage to that, but I still like having the human in between to have that discussion, well, hold on a sec let’s take as a patient, cause my wife works in healthcare. That’s why I use this example, but you know, there’s a lot of times where there’s other considerations like insurance, a lot of people don’t realize how often insurance changes your doctor’s prescription after the fact. And you know, my wife deals with this every day where doctor prescribed XYZ because that’s what they feel is best for the patient plan, it goes to the, the provider to pay for it, the insurance provider, and some computer spits back and says, no, no, no, no, we want this option because it’s, you know, we’ve run these numbers and this is the option that they’re going to get. Right. This is what we’ll cover and this is what is recommended. And so that’s scary because it takes away some of that individual autonomy because the system told me this is better and the system is great for optimizing for the 80%, but it’s horrible for, or it can be horrible for allowing for that 20%. Right. And so many people that’s where they live is that 20% variance because she works in cardiology, so they come in, not everybody wants a internal pacemaker. Some people want external like some for any number of reasons, right? Some people might want try diet and exercise first before going on. Blood pressure medicine, for example, right? And they should have the ability to have that conversation with their doctor. And you could build it into a system but it seems like the trend is like, no, here’s the answer we don’t need to question it anymore. So again, long-term, that’s a concern for me, because medicine is only one of a thousand places we’re seeing this ,I mean, hell, we’re seeing it with buses in New York where they have camera systems installed in the buses now where if you are parked in a bus lane, the camera automatically sends you a ticket, right? And you can even automate that further where they’re now they’re texting you, hey, move your freaking car out of the way. But, and it’s like, that’s fine, but there’s no discretion, there’s no human discretion, it’s just a computer knocking out these things. As we start seeing law enforcement, military, government starting to use more and more of these systems for noble purposes, I’m not like saying this bad, but I’m saying that it’s just human tendency to let the computer do it and never question it again. And that opens up a lot of potential for abuse or just even inadvertently just kind of oppressive scenarios or at the very least takes away a lot of autonomy and the human element of those discussions that I think are important, particularly for things like medical or law enforcement where, you’re gonna be dealing with somebody at the worst moment of their life potentially, right? And do you really wanna hear you got cancer from a computer screen? Probably not.
TJ – 00:39:33:
Yeah, it needs that, no, totally. It just needs that certainly the validation and human in the loop is that’s why so critical. They can further enhance the experience while going through the process and somebody can now be more sure that, hey, I talked to a system, but I also talked to an expert who was able to kind of seal the deal or put an authority on it and say that this is exactly what’s happening. It gives you a lot more of the initial discussions and the things you’re having with the automation platform, whatever that may be And then the diagnosis elements with AI, that you know, okay, I got some ideas, much quicker than I expected. Now probably I should validate and that validation and human in the loop is so critical. Keeping that discussion in mind, but it just going back into the literary elements. Now writing often involves the use of complex literary elements such as theme, symbolism and subtext. How does generative AI or AI handle these nuances, aspects of literature, and today precisely? And can AI comprehend and generate intricate literary devices to the same degree as a human writer? I think you touched upon it to an extent based on the prompting, but is it kind of getting there at all, James? And if you could pretty much keep it more vision-oriented to your answer, where do you see and how do you see the comprehension and generating of intricate literary devices to the degree of human intelligence is going to happen today or if it’s already happening or there’s a future for it?
James Timberlake – 00:41:03:
So I’ll say some things that it’s good at is guiding or can be good at is guiding you through say a literary framework. So anybody who’s ever like one of the first books most writers or screenwriters at least will get is called Save the Cat, right, which is a framework, it’s essentially kind of like the hero story, but which is of course another one. And then you have the story circle, Dan Harmon’s story circle, which is another version of kind of the hero story. And so something like that, that’s pretty well established and there’s a lot of writing about it, both academic as well as actual usage, examples and usage. It’s pretty good because again, it knows what you’re talking about, right? So we can pull from that. If you were to say something like, if you asked it for a metaphor or if you asked it for something specific like that, like I want a metaphor about a duck and a, you know, a piece of wood and it talks about, you know, why you should make your bed every morning. It’ll create something within those parameters, yeah, it’s pretty, you know, it’ll sound like a human wrote it. And again, the more it’s like you said, the more parameters you give it, then the closer it’ll probably get to what you’re wanting. Right. So if you can give it examples, that’s always helpful, like, even if it’s just within the prompt, like if you can say, here’s an example, but for your version, I want it to be about this, I want it to be longer, shorter, better, whatever and It’s fairly good at that, yeah. And then again, it also depends on what we’re talking about, so like I mentioned chapterly, I think earlier, so chapterly is a fun one. So chapterly, if you’re talking about, like if you’re an author, you’re looking to write a book, nonfiction or fiction, chapterly is a pretty good option because the AI that’s built into chapterly is designed a little bit differently. I’m not affiliated with the company, so I don’t know a lot about their backend, but just from usage, you know, I can tell that their training model is a little bit different from what you get from some of the larger models,It also pulls in context from your book, right? Like you’re into it, because you can create books within the app, like a bookshelf, so you can work on multiple projects at once, and it knows which project you’re in and it knows the characters and the tone. And you can also use, within that system, you can use some of those existing frameworks, like it was somewhat earlier, it’s got a whole list of frameworks and it’ll walk you through those frameworks, right? And then if you hit a stopping point, just hit a button and it’ll start writing for you, so it’ll write the next sentence or it’ll write the next paragraph for you and keep that continuity going. And so if you’re talking about something like that, yeah, I mean, there’s a lot of, again, there’s a lot of opportunity, because you’re working with an established framework or you’re utilizing a well-known ,you know, model or something of that nature is pretty good at that so it makes a pretty decent coach in that sense but again you know it all depends on how much information you give it if you were to just say hey write me a book, it wouldn’t necessarily fall, it might. It probably would try to follow some type of best practice, but it wouldn’t necessarily follow the one that you want.
TJ – 00:44:03:
And with that, the next phase of the discussion where I think we touched upon some of it while talking, you know. The narratives which is being built and how, with technological advances, ethical considerations, particular role. What are the some key ethical concerns, you know, when using generative AI for ghostwriting you may see today? And how does one manage, let’s say, attribution authorship, like when AI is contributing significantly to the final work, how ethical considerations and you know, sort of contributions and attributions are managed today in the ghostwriting or publishing industry.
James Timberlake – 00:44:38:
I mean, I think the obvious ethical one that probably comes to my mind at least is obviously using work that other people either didn’t know was being utilized by AI in the first place or they posted it and it just got scraped or something like that. So in that sense, you could argue that you’re using that work and there’s no way to really attribute it to them because it’s one of a billion people that have contributed to any given large model and you’re seeing pushback right now. And you also see it’s not just with, I mean, we’re talking about publishing, but you see it a lot or you see the conversation a lot more around art, right, with people talking about how my art was used. And you can go and type in the style of a particular artist and reproduce work in their style. Right. And so that could argue this ethical because you could argue that it’s diminishing the value of the real thing or, again, flooding the market to the point where the real thing is no longer even want it because I’m so sick of seeing it. Right. And you can do that with writing, too right? Again, we’re talking about Stephen King, any well-known author in the last decade, you could probably steal their voice pretty well. Right. And even if it’s not an author, even if it’s just, like I said, a thought leader or a public figure or hell, I mean, if you did enough work, you could grab enough probably off of just scraping someone, some random person’s social media that you could build a model off of them and then you could write in their voice convincingly enough and we’ve already seen this scammers utilizing this. The one example that made the rounds was somebody took the voice of a daughter and then called the mom and said, hey, I need money or something like that. And but there’s others like, you know, I mean, inner or writing scams in some form or another have been around forever. Right. And so, you know, there’s going to be a lot of that. The other thing that’s kind of an offshoot of that, this is, I guess, probably something not as many people care about, but within the context of the industry, I think it’s important to consider is a little bit like what we were talking about before, but now that you can, now that anyone basically can come in and write a book on pretty much any subject, utilizing these tools, you know, to me, the ethical question is what responsibility as the author do you have? Because if you’re going in and writing a book and you’re not a subject expert, and then you’re relying on AI to fill in your knowledge gaps, well, then you’re far more likely to introduce those hallucinations that you were talking about, or even just, you know, bringing in, you know, theories or things that have been, you know, put on the internet, but aren’t necessarily accepted, or maybe they’ve been disproven or changed over time, but because you’re not in the quote unquote industry that you’re talking about, or you don’t know a lot about the subject you’re writing about, but you wanna be perceived as an expert, it’s really easy to just spit a book out, put your name on it, right? And you could write, I’m not gonna drop any names, but I know one person who’s written 50 books in the last year, all of them with AI. They’re an expert on none of them. And that’s okay, I mean, okay, whatever people still, they’ll buy it or they won’t and that’s fine that how it is. But for me, there’s like, I have a hard time with that because again, being the guy in the middle, I know the difference between when somebody comes in, I always tell this story, but this is, I had a client come in and I always asked a set of questions. And this is a few years ago and I said, what is your book about? And she literally said, oh, just whatever will get me on Oprah, that’s the book I wanna write. Okay, so you’re not a good fit for us. But the problem is there’s plenty of people that would take that in a heartbeat and AI isn’t gonna ask questions, it’s gonna be like, all right, well, based on statistics and what’s trending right now, you probably wanna write about these three things. One of these three things. Which one do you wanna write about? Oh, I wanna write about number two. I’ve seen that in the news. I know that’s trending. Great, boom, here’s an outline. Because that’s the first thing you would do is like create me an outline. All right, great. Now here’s the trick, here’s how you do it if you want me an independent publisher. Here’s how you would do it. I don’t recommend this unethical but if you wanna pretend to be a subject matter expert on something, just go to chat GPT. Figure out whatever trending thing you wanna be an expert on, have it write you an outline, 12 chapters. And then for each chapter, tell it to write you three sub chapters and it’s gonna create a list. And then you take each of those and say, write me a blog post on each sub chapter thing. So then you’re basically just building out a branch you stick all those together, you put it in a book, you publish it on Amazon, and then boom, now you’re an author. Congratulations, you’re an expert,and that’s all it takes. Now, is any of the information factual? You probably don’t know, you could read it and it might sound good, but do you know if you’re not an expert, you’re not gonna know and your readers certainly aren’t gonna know. But worse than that, they’re gonna assume that you do. And so they’re not gonna read it and go, man, this is probably a AI hallucination. They’re gonna read it and go, oh, okay, that’s how it is because so and so told me so. And so for me, I mean, we talk about ethics, that’s a big issue, right? Because it’s so easy to fake it now.
TJ – 00:49:40:
Yeah, and the misinformation can carry forward. I think it just may assume something and put together. And then now, and that’s the nature of generative models. It has to throw up and spit out some output. No fact checks on top. And then eventually somebody is reading it , Maybe, yeah, to your point, understanding if it’s coming from this author who probably is now known, he’s been seen everywhere. And then it’s OK, he’s saying so. That may be true, but nobody’s fact checking it. So yeah, that’s very concerning and can be scary in the longer run, for sure. Misinformation can travel so fast.
James Timberlake – 00:50:13:
Oh, no, I was going to say, but I will say that the flip side of that is if you’re writing fiction, sometimes those hallucinations can be amazing because they’ll take your story in a direction you never would have thought of on your own and you’re just like, that’s not even a thing. But now that it’s in here, I’m going to run with it. And I’ve had that happen too. So like, it’s not all bad, but I think there’s a different level of responsibility for nonfiction versus fiction.
TJ – 00:50:40:
And practical approaches to things if you’re putting together some thoughts out. Absolutely. I think fiction, that’s a good example. Keeping fiction in mind, you know, that’s where you need to imagine things and I think probably, you know, generative models can do a lot more better, they’re 100 percent. Cool. So probably towards the last question here, and this is just, I think you touched upon it when we first started the discussion today in the morning. If we have to cast our gaze to the future and considering the trajectory of generative AI and AI advancements and innovations. Where do you envision AI’s role in independent publishing, including post-writing, five or 10 years down the line from now? And if you can project any potential advancements, which you might be foreseeing already, James, because I figured out that there’s so much insights you were able to share. Those will be the final thoughts and questions for you, and then we’ll definitely do a closure.
James Timberlake – 00:51:31:
Who knows? Things change so quickly now. But I will say, I think where eventually we’ll get to, probably even closer to the five-year range, just like we were talking about personalized, generative entertainment, I think you’re gonna have personalized, generative news, you’re gonna have personalized, generative education, you’re gonna have this real-time, the information I need will be provided when I need it, and it will be provided contextually by the AI. And I think that’s eventually what we’ll get to. So it’ll be less, hey, what do I wanna watch? It won’t be like right now where you go on Netflix and you start clicking through things, or I go to Amazon, I go, I wanna read that book, or Kindle or whatever. I feel like what’s gonna happen is more and more it’s gonna be, hey, I wanna read about this, produce it. Hey, I wanna see this, produce it. That instant, that can be good or bad. The good side of it is, anytime I’m in any situation, I don’t have to stop, go on Amazon, look at 5,000 books on a topic, and hope I order the right one, and not know until after I’ve read it, and invested that time. Well, now it’s like, hey, read the top 10 books on marketing that were written in the last five years, or in the last five months, and give me the top three takeaways that they all agree on, so that I could execute that right now. So you got private readers that give you the, things like that, where you could start having that dialogue and it could be more of a research assistant. And if that is the case, my assumption is, reading will become less, non-fiction reading will change, non-fiction publishing will change, because it’ll be less and less of these educational leadership style books that are very popular right now, how to do this, how to do that, top 10, whatever. And you’re going to see less of those because it’ll be so much easier to get that through the AI. What you might start seeing as far as being produced for mass consumption will be more stories, will be more bios, will be more maybe self-written bios as opposed to something. Because again, anything academic, the AI can produce for me. So where does that leave the human? It’s like you were talking about a minute ago, it’s about that creativity or stuff. And so I think what we’re going to start seeing is more and more, when we’re talking about earlier, independent, small teams that are self-publishing their own stories, right? Their own IPs and interconnected world. And I think that one of the things we’re seeing now that’s really exciting for me, it’s not necessarily it’s empowered by AI in a lot of cases, but not necessarily, but it’s a collaborative writing, right? Which is not necessarily a new thing that’s been around, but with AI now you can do collaborative writing in such a different way. So you can create story branches where you and I might start at the same starting point in a story but if we go through, say, a week together on a platform where we’re reading the same story but it’s like choose your own adventure, right? Except we’re generating the story as we go. Well, by the end, we might have started with the same story seed, but we’ll have two very different trees at the end. So to me, that’s one of the things that we’re working on too is when you’re looking at, well, I’m going to pick up my next book, it’s not gonna to be a finished story, it’s gonna to be a potential adventure that you’re going to co-create as you go through that story.
TJ – 00:54:54:
So true, that’s brilliant. The analogy of we start from the same point, but how I create those dynamic workflows through asking different prompting or whatever that may be to build a narrative could be so different from what we both may be thinking about, that similar topic. Very interesting. Well, I think sometimes I have seen even asking ChatGPT, it builds different narratives, even if you ask, just give me another iteration, it’s completely different from the previous one. And then you can ask for specific prompts to make it whatever you like and tone.
James Timberlake – 00:55:27:
No, but you’re spot on. Exactly right. And now you have the ability to do that as a group. So if you take something like a fun one, one of my favorite systems right now is swarm.ai. and you might be familiar with them. But basically, I was talking to one of the guys over there, and they were showing me an early demo, and I started laughing. I go, oh, man, you gamified group thing and he’s like, no one’s ever worded it that way, but that is essentially what we’re doing. But if you’re not familiar, it’s just kind of like a Ouija board, where everybody puts their hand on it, and it moves or whatever and so what they do is they put options in a circle, and then everybody can come on and vote. But you just see the collective swarm intelligence, and you start seeing which way the vote is going. And so it starts eliminating options as it becomes clear that that’s not going to win until you get down to the one consensus. And then everybody voted on the same thing, it’s fascinating to watch in real time. But my point is you could take a system like that, which is being used for 100 applications right now across all different industries. But you could take something like that and apply it to publishing, and you could create a story as a group using this voting metric to determine key points and create multiple or alternative branches like we were talking about a minute ago, but as a group of 10, 20, 30, 100. So if I’m Marvel and I wanted to do something like this, then I could be like, hey, guys, we have a Captain America story or an Avengers story, and it starts here. Here’s the world. But who shows up next? Dr. Doom, Mr. Sinister, or Sabretooth, I don’t know, that’s three. Anyway, it’s like who shows up, and then we vote. And then based on who gets voted on, that creates the next story, and then you keep voting. Well, then you could take that, and then you could publish it, you could turn it into an animation. And what’s great is that’s also one of the things I see happening is people that are traditionally authors. That’s why I start using the term storytellers more, because I think that’s going to be a more accurate term as we go forward, because you’re no longer going to be an author or an artist or an editor. You’re just going to be the storyteller that does all of these things utilizing AI. And so I think what you’re going to start seeing is people that do that, where they work with, they build communities around their fans, and then they start having those collaborative expressions, building worlds or characters or stories with their fans so that the fans have that ownership. Because right now, the one thing that both the fans and the creatives agree on is that the studio makes horrible decisions that nobody likes. And so if we get rid of them, and it’s just the fans and the creatives working together collaboratively, I think there’s a huge opportunity to tell stories that we’ve never seen before, to do things that no studio would ever sign off on. And it’s self-funded by the people that are creating the story as fans that want to see the story, you know what I’m saying? It’s like this self-fulfilling thing. And to me, it’s an exciting avenue that we’re starting to just see the very early days of and then once you have the story, then it’s just a whole, like we were talking about earlier, great. Now there’s no reason I couldn’t take another system, put it on top of that, and have it render that story as a comic, and then have it render that comic as a movie. And so now all of a sudden, I’ve gone from some random dude in a Discord to now I’m a co-creator of a movie, my own movie that I helped create by my votes and it’s like, and then you’re going to want to share everybody. So you’ve got a built-in distribution network. Anyway, that’s where I see things kind of heading.
TJ – 00:59:06:
Wow, that’s so exciting. I’ve never seen the world that way at all. This is brilliant, James. Well, on that note, I really want to thank you so much for today. James, I’m in frank and given, we are in more platform technology field doing different things. But perspective of how you do publishing, host writing, and what the advantages could be, and then eventually the biases, how you see the future, I think there’s so much to learn from. I’m sure the audiences who listen to this could take a turn back from this episode for sure. Well, thank you so much, James, for today. It’s been a true pleasure to have you here at Not Another Bot, the generative AI show. And we would definitely love to talk to you more in future for sure. Thank you so much for your time.
James Timberlake – 00:59:51:
Yeah, and love it. I appreciate the time. Goodluck.
TJ – 00:59:57:
How impactful was that episode? Not Another Bot, the generative AI Show is brought to you by yellow.ai.To find out more about yellow.ai and how you can transform your business with AI powered automation visit yellow.ai. And then make sure to search for the Generative AI Show in Apple Podcasts, Spotify and Google Podcasts or anywhere else podcasts are found. Make sure to click subscribe so you don’t miss any future episodes.
On behalf of the team here at yellow.ai. Thank you for listening.