Well, that [generative AI] escalated quickly...
Disclaimer: there’s a 100% chance this is obsolete the second I post it.
To quote one of my favorite movies, “Life moves pretty fast. You don’t stop and look around once in a while, you could miss it.” Our world changes rapidly these days—war, devastating weather, scandals, returning to the moon…one day rarely looks like the day before it. So, you could be forgiven for recognizing that generative artificial intelligence (AI) was a thing, without fully grasping the magnitude of it. Or, how quickly its evolved. Other, smarter people will opine on this technology and far better than me, so (for once) I’m going to stick to my lane and simply talk about the impact of generative AI on writing and publishing.
If you’re reading this, I’ll assume you know what Generative AI is. If not, click the link. I’ve used ChatGPT as research assistant for my books on subjects as diverse as quantum encryption, terraforming and repairing the ozone layer to helping me find books comparable to mine for marketing. While the research summaries it provides are high level, it would’ve taken me hours or days to compile the same amount of information and distill it into a paragraph or two for easy reference.
Once we see a machine do something we believed only a person could do, from summarizing a complex subject like quantum encryption to driving a car, it’s only natural to wonder, what else can it do? I think within the next decade, if not sooner, we will see tectonic shifts in multiple industries, where AI can augment if not replace outright human workers. So, it’s not surprising that the debate already started on whether generative AI can replace creatives. Some advanced generative AIs are already writing code, how long before they can write whole applications? Maybe even a computer game. What about writing something else, such as a novel? Some AI evangelists believe machines will be able to replicate nearly anything a human can do. So, if AI works by learning to do a thing, taking feedback and refining its output…why not creative works?
I used the prompt “write a short story in the style of Elmore Leonard about three crooks trying to escape a bank heist.” And, within two minutes, ChatGPT gave me a short story that was exactly that. I’d give the output 3/5 stars for what it is. It didn’t nail Dutch’s style, but it’s a serviceable microfiction. It checked most of the boxes of my one sentence prompt. Honestly, it was probably as good as the first short stories I cranked out in middle school. I wrote constantly at that age, and I improved quickly. By high school, I was writing short stories, novellas, bad poetry, whatever I could crank out. If I could teach myself to become a better writer through practice, study, and repetition, couldn’t an AI? After all, it’s not limited by time the way I was. Assuming they were available, an AI could consume every book I’d ever read, create a piece of fiction, and then teach itself how to improve that work. After all, that’s how AI works. It teaches itself based on a learning model defined by its creators and improves within that framework and based on input from humans on how it did. Given that, could an AI could read all of Elmore Leonard’s work and create a style guide replicating his tone, themes, favored words and turns of phrase? Could it write a serviceable story that sounds just like Get Shortyor Out of Sight? Maybe, probably. Within the next decade? Based on what I’ve seen in the last 5 months, without question.
But I firmly believe it would be bad pastiche.
Our (eventual) AI overlords may disagree with me, but I’m not worried about my job security. Creativity is ephemeral and intrinsically human. AI might be able to create a reasonable facsimile of Elmore Leonard’s work, but they’ll never get his wit. A machine is never going to come up with the conversation-within-a-conversation between Jack Foley and Karen Sisco in Out of Sight (if you’ve read the book, you know what I’m talking about, if you haven’t…stop what you’re doing and go get it). A machine isn’t going to write The Things They Carried or Snow Crash any more than they’re going to come up with Bohemian Rhapsody or Back to the Future. Those stories, songs and films are based on lived experience, emotion, and creativity. A machine cannot and will not replicate that. That’s not to say that an unscrupulous or lazy writer might not use the technology to cut corners. I can see someone try to leverage generative AI for an outline or even a first draft under the auspices of “saving time”. I think this is wrong and will only flood the marketplace with inferior work.
Okay, cool, none of the authors are going to get sacked. What, then, does generative AI really mean for us? AI will become a huge force multiplier for creatives, and business in general. Despite the indie revolution over the last fifteen years, publishing remains ripe for massive disruption. AI can enable writers in so many amazing ways, giving us tools to level the playing field and further democratize the publication process. There are many ways this tech is going to change the writing world, here is just a sampling.
Editing. A good editor is as creative as a writer. A good editor has a feel for that intangible quality that could make a good book great. They know how to suggest subtle changes to the craft so that it hits with the audience in the right way. A good editor is part coach, part shaman…maybe even a mystic, I don’t know. They help the writer improve their art, sometimes even helping the writer figure out what they’re actually trying to say. You’ll never replace that with technology. But, just as writers will benefit from AI, I think there’s a world where technology greatly enhances proofreading and copy editing. There are tools that do this now, and I use them, but I still rely on a human to put eyes on it before it goes to my readers. The technology needs to improve, but it will happen and within 5 years I think we’ll see fully automated proofreading and copy editing (in layman’s terms, this is checking/correcting grammar and spelling). Editing, like storytelling, requires judgement. There is a time to break the rules and an AI won’t know when to do that. I don’t think we’ll replace human editors, but I do think we’ll be able to make their jobs easier so they can focus on making our stories better.
I’ve already talked about how AI can be an effective research assistant and these capabilities will only improve. I’m supremely excited for this. Getting more detailed, in-depth, focused research will save significant workload. You’d be surprised at how much time I will burn researching a topic for what amounts to a single line in a story. The next evolution of this concept is fully featured digital assistants. An author is their own small business and it’s not a far cry to say that launching a book is like its own mini startup. Having a digital assistant to run the myriad tasks associated with this would be amazing. An AI assistant to triage emails, prioritize and organize my time, based on inputs around my creative periods is incredibly useful. There is tech that does some of this today, but not well. An AI assistant could also search out engagement
opportunities like podcasts, speaking events and guest blogging. It could also search for publishers to pitch. Say I've got a short story, my AI could save dozens of hours by searching for magazines that are open to submissions and interested in that type of story.
But, this stuff is table stakes and doesn’t scratch the surface of what AI can do.
I came out hard a few paragraphs ago that AI would never replace creatives. I do, however, think that there are areas in our industry that AI makes some inroads. Cover design is one. Book covers sell books. There are also designs that are proven to work. One of the best examples is the archetypal “shadowy man running away” in the thriller genre. Cover designers know the patterns at play and create accordingly. Given what we’ve seen in the last few months with AI generated graphics, an AI can easily learn these patterns and produce compelling covers. There is a massive body of source data available—book covers on digital storefronts and we can correlate that with sales figures to make conclusions on what works. A good cover is an investment (and one you should make), AI here can be a significant time and money saver for an independent author or a small press publisher looking to control costs. My hope is that AI generated art amplifies the work of a human cover designer rather than replaces it altogether.
The next area is in marketing copy—both those ad snippets (ex: If you liked Bad Monkey by Carl Hiaasen, you’ll love Proper Villains by Dale M. Nelson) to the summary on the back cover. As with cover designs, there’s a mass of source material and the machine can easily learn what is effective and what is not. As an indie, you can learn to do this yourself (through time, trial and error) or you can pay someone to write you blurbs, which can run about $300-500 / book. In addition to marketing copy, there’s the targeting aspect, which is simply finding the right readers for the right product. AI’s ability to ingest massive amounts of data, parse it and then target ads to the right demographics is what successful advertising is all about. Running an author business and being, admittedly, poor in this area, I’m super excited about being able to offload this to a digital partner.
Those examples, coupled with the editing improvements I see on the horizon, could be massive changes in the industry and further level the playing field by making it easier for authors to bring their stories to readers. The last major change and one that’s already underway, is AI narration. I know some authors are already experimenting with this. The technology exists and we’ve already seen how AI can be used to mimic someone’s voice and/or likeness. Deepfaking has already been used in advertising (with permission), in film (also with permission—Rogue One is the best example of this), but the technology has the potential for great peril. Audio deepfakes have become a staple of social engineering scams and cybercrime. The technology to produce high quality narration at low cost isn’t far off.
I’ve got more concerns here than with other applications of AI. First, I listen to a lot of audiobooks and a great voice actor delivering a narration is something to behold. I’m a big Star Wars fan and there’s a particular narrator I like who just nails Han, Lando, Luke and a host of others. What happens when Lucasfilm authorizes an AI to replace that narration with a replica of Harrison Ford, Billy Dee Williams, and Mark Hammill? Do those actors get compensated because the AI used their voice to train itself? Or, what happens when an unscrupulous author likes a particular celebrity’s voice, say, James Earl Jones, and directs his AI narrator to read is book and sound like Mr. Jones (but perhaps be just different enough that its somehow not illegal?). There are a multitude of repercussions here that we have not fully considered or reconciled, and are going to have to contend with soon. I do genuinely worry about my peers in the audiobook space because I think their livelihood is danger in the near term. I don’t think they should be, but I think the technology will be good enough to be a threat soon.
Generative AI is going to force sweeping changes across the entertainment and the arts. This Hollywood Reporter piece highlights the pressing concerns the technology is creating in the entertainment industry and some of the steps the writers and actors guilds are taking to protect themselves. It also crystalizes some of the broader implications of this technology.
If you’ve read the Firewall Spies series, you know that I’m keenly aware of AI’s destructive potential and am seriously concerned about our government’s lack of preparedness. I’d argue as a society, we are equally unprepared. That is not to say that I think we’re heading for a Terminator-style Armageddon, just that much like nuclear fusion—the promise and peril of artificial intelligence can be equally terrifying and exciting. I think this is happening much faster than anyone anticipated. That change will only accelerate as new technologies are deployed. We are going to have to contend with the rapid advance of AI and the implications on creative work, on ownership and copyright very soon.
This is a massive topic and I’ve barely scratched the surface. In focusing this post on just the positive technological changes I see, I’m not burying my head to the dangers (in fact, I think far more about that than the positive side) and will be back soon that chat about that aspect. That said, I'm not optimistic about the government's ability to proactively safeguard against the misuse of this technology. Nor do I think they will effectively work with the companies developing it to ensure it's done so ethically. There are few technologists in congress and if their Facebook testimony a few years ago is any indication, advanced technology is far outside the legislature's grasp. However, there are lawsuits underway now in the US, the UK and Europe which will certainly inform the intellectual property aspect of this discussion.
More to follow, for sure.