If you’d like to receive my blog in your in-box each week, click here.
I recently encountered an AI writing issue in my work that I should have foreseen yet somehow didn’t.
An author turned in a second-round revision of a story that was already very good when I saw it in its first-draft form, but needed development and deepening, as many early drafts do. This is a newer writer, but one whose work I’ve watched grow and deepen very quickly in our previous work together, and they have been consistent and dedicated about putting in the work to master this craft.
This second revised draft they turned in did start to delve deeper into some of the territory we had discussed in our first pass, namely expanding on the characters’ background and how it shaped them, deepening characterization, and continuing to develop some of the key relationships and arcs.
But the whole feel of the story had changed from this author’s usual style, which I’ve come to know fairly well. It was stripped of voice, a bland soup of pretty and overfamiliar phrasing, unimaginative clichés hampering its effectiveness. While I could tell there had been more development of the story, there were gaps in logic and continuity that had me struggling to follow the throughline, and I wasn’t more than a couple of pages in before I began to wonder if the author had used AI to help them in the revision process.
Let me offer the caveat that however an author chooses to write is their business, and my job is simply to take what they give me and help them mine out all the gold. I don’t question or judge how they got the manuscript to where it is.
But this story felt as if it had taken a giant step backward from what had been working well already and made it uniquely this author’s. They have a very distinctive style and approach and background that had been woven into the fabric of the story in their earlier draft and the previous title we had worked on together, all of which was missing in this one. It had gone from being a good story that could be made even better to a generic, thin facsimile, and while I was willing to work on it in our second-edit pass if that’s what the author wanted, I knew them well enough that I decided to first have a conversation.
When I asked if they had used AI for part of the writing, the author said they had done a deep revision pass and then put the final product of all that work into an LLM engine to give it more polish. This author’s strengths are more in story and character than in technical copyediting elements, and they wanted to sand off any rough edges.
I don’t know what specific command they fed into the machine, but it sanded the story down into a broad and superficial shine, stripping it of what had made it stand out even in that rough first draft, before the author’s efforts to continue to develop the story.
The Uses and Ethics of AI
My first reaction to this incident was to wonder whether I wanted to work on manuscripts that might be partly or wholly AI created, and if I should include verbiage in my website or contracts about AI-generated material.
And then I questioned that premise.
I’ve written about AI a few times already and its potential for and actual impact on writers and our industry. Per expectations, given the exponentially fast growth of technology and particularly this technology, its capabilities and reach have already continued to grow and expand since my posts beginning a couple of years ago. AI is ubiquitous now and has become a tool for many people, including authors, the same way most any technology can be used to support and facilitate the work that humans do.
I’m not particularly tech-savvy, and certainly not an eager early-adopting embracer of technology. I tend to rather reluctantly jump on board with new advances. But jump on board I do, and as early as possible (with the help of my savvy early-adopter tech husband), because I know that if I don’t I risk being left behind, the world and my industry evolving past me.
Like any other technology, AI has a place in facilitating authors in their writing careers.
So this isn’t a post about how you should never use AI in your writing. I think like any other technology it has a place. I have used it myself for things like generating an outline of a subject I’m developing a presentation on to make sure I have hit everything that writers might expect or want to be covered in my own outline.
And I’ve discovered my favorite use for it, which is to generate flawed or clumsy passages of narrative that I can use as examples in classes and workshops to illustrate my points and offer us material to analyze and edit together to improve it. (While I rampantly use examples from recognizable published works in my teaching, I rarely do as anything but positive examples, to avoid trampling any author’s work or feelings.)
It’s unrealistic to expect that authors may not utilize this powerful tool to make themselves more productive, to enhance their creative work and their creative product, to streamline some of the many processes involved in being a career author, both craft and in business, just as it would have been unrealistic of an editor to refuse to work on manuscripts created in word-processing software, when they first became ubiquitous, rather than typed on a typewriter.
Our world—and our business especially—is always changing, never more so in my experience than in recent years. Digging your heels in and refusing to evolve along with it or to accept those changes makes you a dinosaur, and also risks robbing you of advantages that may help grow your craft and career.
But it did make me sad to extrapolate about what this might mean for creatives and for our field. I think something important will be lost as we incorporate machine learning more and more into our own learning.
Creativity Is the Way, Not the Means
For starters, having worked in this business for many decades and with thousands of authors, I know firsthand that the only way to truly begin to understand and master this craft is by slogging through the swamp of learning it.
That means writing hundreds of thousands of words, probably many of them pretty bad that will never see the light of day, before you begin to understand all the many moving parts of this craft and how they fit together. You have to do the thing to learn how to do the thing, and “the thing” is the core of our identities as writers and as creatives.
And that’s the other area where I worry what overdependence on AI and other technologies may cost us as artists, may cost our creativity.
I’ve spent no small chunk of my own writing for authors in my books and blog posts and articles talking about that process: how the creative element of our careers is the only part of them truly within our control. How creativity is and must be its own reward if we are to create and sustain writing careers that fulfill and satisfy us.
Although our business is based on creative product, it’s more than just a commodity to be created and marketed and sold. I am of the school of thought that art has a higher purpose than that, a deeper one. It has the power to move and affect and change people. It has the power to impact society and our world. It’s not a widget to be mass-produced and automated at the push of a button.
What makes writing—and any art—affecting and profound is the very humanness of it, a quality that, for all the technological prowess I fully expect AI will develop, I have strong doubts it will ever be able to mimic or generate.
The same way I almost immediately recognized that something was off about the manuscript I was reading, I think people know the difference between art generated from our hearts and minds and souls and something regenerated from binary code at the chewed-up lowest common denominator of vast amounts of other people’s writing.
Read more: Human-authored Certification from the Author’s Guild
What Does Overreliance on Technology Cost Us?
A couple of years ago I watched the series documentary Limitless on National Geographic, where Chris Hemsworth, with the help of various experts, explored core elements of health and well-being to push the boundaries of human ability and longevity.
One episode that particularly stuck with me was about memory, in which a neurologist suggested that one of the worst influences on our resilience of memory is GPS. So many of us rely on it now to orient ourselves to where we are in the world and how to get where we’re going that our spatial abilities, mental mapping, and memory may be negatively impacted. I liken it to the way we so many of us lost the ability to retain phone numbers with the invention of speed dial.
Since that episode I’ve been using my GPS far less, and I noticed that not only does that seem to have a definite positive effect on my memory, but it makes me much more aware of my environment. I look for street names and notice landmarks, keep track of where and when and which direction I turn, and am more deliberately mindful of distances, knowing that if I want to travel this route again, or retrace my path back to where I came from, I’d better pay attention.
Not replying on the convenient technology of GPS makes me more present, aware of my environment and where I am within it.
Finding Balance with Technology
Back to my conversation with the author who had turned in the AI-polished manuscript. I gave my impressions of what I’d read so far and said that I thought the manuscript felt less effective in its current version, even though the language was technically cleaner, for all the reasons I mentioned above. I told the author I would work on whatever version of the manuscript they wanted me to, but that my suggestion was that they have me edit what they felt was the strongest version.
Based on our conversation, they sent me their original second-draft revision, before they had fed it into AI for the final polish.
There was still a bit more development and honing for us to do in our next pass, but the story was vastly stronger than the AI version they had originally turned in. I was gratified to see the author had indeed done the work of strengthening the story on their own, which gave us plenty to work with and continue to build on—and also revealed how much their skills were continuing to expand.
Best of all, the story’s voice was restored, its originality and uniqueness and depth. The things that will make it stand out in a crowded marketplace.
I’m not really sure what the takeaway is here. I’m not naive enough to think that all authors will be purists about the work and not lean on AI for more than logistical or technical support and facilitation for their careers.
But I guess my hope is that we’ll remember the shining soul of this work, which is our selves, our distinctive personality and our humanity.
I hope we will realize that the only way to truly master a craft about which most of us care deeply is to practice it, even if that means failing and failing and failing as you learn what makes for success. I hope we might realize there are no shortcuts in this process, or at least not ones that make a meaningful contribution to us as artists, as creatives—as human beings.
I hope authors might remind themselves that this is not a craft about results and product. Creativity is about process, and automating that to a machine undermines and circumvents the beating heart of any creative pursuit.
Overreliance on farming out such key parts of the creative process, to my mind, risks our losing a measure of wonder and uncertainty and unknown, all such fertile ground for our imaginations and for our brains to puzzle and ponder and learn—and grow.
Okay, authors, I know I’ve opened up a bag of cats here, but let me have it. I want to hear your perspectives on using this technology (or any tech) in your creative work: what you use and how, your reasoning behind it, and your own thoughts on the place of AI engines in our art.
If you’d like to receive my blog in your in-box each week, click here.
68 Comments. Leave new
Thank you! Very thoughtful and revealing information about AI and the inner workings of a creative process. I find the interruption of what little AI I have experienced annoying. Rather than adding to the potential of the creative process, it gets in the way. Grammarly is an example.
I have a book written that needs editing. I plan on checking with you to see if we can work together.
Robin Fay-McNair
Robin.faymcnair@gmail.com
I like Grammarly and don’t mind the little nudge from programs like that or Word’s–but yes, I get aggravated when they “correct” something wrong, or miss the nuance or meaning. (Of course, I spent about 12 years as a copyeditor, so I may get a little territorial in that area….) That said, I always think grammar, spelling, punctuation, etc., are part of the writers’ basic tool set, and merit more attention than I think they’re sometimes given in our creative process (AFTER the drafting, of course!).
I’d be happy to talk about the possibility of working together when you’re ready, Robin. Thanks for the comment!
I disagree that AI is just another tool, and that therefore we must follow and ‘engage’ with it. GenAI is terrible for the planet (even more important than its likely negative effect on the arts and communications) and quite unneeded. Rather, I suggest to ultimately preserve our craft, we will have to keep it as such – and develop and promote a niche for the human-made. There will be an audience. (I hope)
Thank you for not equivocating! You’re absolutely right, both on planetary & human creativity levels.
How writers, of all people, walk so blindly into this dystopic hellscape just because the billionaires put it before us to consume is just…dystopic! Not to mention the point someone else makes below about this idiotic technology having stolen *our own creative works* in order to exist.
Why do the wealthiest on the planet want to get rid of the artists? Hm.
Art has always had enormous power and value (it’s why oppressive regimes start with book banning, for starters). It makes people think and question and feel. And in this case, it constitutes the materials that these large language models couldn’t function without–however ill gotten that data is. I think there’s a lot of good that can come of AI, but we have to be mindful of the risks as well. Thanks for commenting, Angie.
This is such an important point, Fleur–I ALWAYS worry about the detrimental ecological impact of AI, and it’s why I keep my use of it to a minimum. It worries me how much people seem to default to it for that reason–I can only imagine the toll it’s taking contributing to global warming.
I agree that there will always be a market for “human made”–in fact Author’s Guild has already created a human-authored certification that members can display in their works and other materials. I think that has value. But I’m realistic enough to accept, too, that there’s a market for AI-generated work, in some areas and genres more than others, perhaps, but not every reader will care about–or even see–the difference.
History has shown us too many instances of technology overhauling industry, society, the world. No matter how much we might want to resist it, change comes and is usually eventually accepted, at least to some degree. Few of us wear handmade clothes, or drink from hand-thrown cups, or type on our typewriters, or any of so many other areas where modernization simply changes the world. I just hope, as you say, that we can keep sight of the core of our creativity and humanity, and figure out how to coexist with the changes.
AI is a tool and like any tool, can be misused. I remember reading blog posts where someone had stuffed keywords in an attempt to get more traffic to their website. Google changed their algorithm and all those sites got penalized, so people stopped doing that. Google revises it’s algorithm often to stay ahead of the latest cheats. I use ai to generate outlines and do research, but not to write. I remember the explosion of absolute dreck when self publishing became a thing. I think writers who create will always stand out over people who let ai do their writing. I imagine when the printing press came out a lot of scribes shared their fists at the “new technology .”
Ha–yeah, that’s what I try not to be: the fist-shaking tech resister. I know that way leads to obsolescence, and I always want to try to grow along with the world. But it’s damn sure not my instinct. Every time something like this shakes everything up, I grouse to my tech-savvy, tech-loving husband, “I fear and resist change!” And then I pull up my big-girl pants and learn about it as best I can. Because life is progress and change and I don’t want to calcify! Thanks for the comment, Stephanie.
More than a year ago, I tested AI for plot development, on a sample scene, and on meetings minutes. I fed it notes, and zip! Completed product! The plot (I had asked what happens next) was logical but bland and ignored the character’s deep desires. The synopsis lost its voice. The minutes required half the time they typically demanded—merely a matter of separating topics that had been magically blended together because it was possible from my notes to create sentences that tied discussion to unrelated action. I’ve heard the tools are much improved now. But AI is no more intelligent than autocorrect or predictive text. It may help. It can harm. Use with caution.
You describe very much my impression of the AI influence on this author’s story, Chris–it made it sound a little smoother, but at the cost of originality (which it of course isn’t yet capable of–obviously, as just basically a recycled cache of other people’s writing). And oddly at the cost of fluidity and sense in some cases–if I had to guess I’d say because it can’t yet grasp how a person’s background and experiences and psychological makeup and personality all interact to create a complete, cohesive character. I think “use with caution” is as succinct a summation as anyone could come up with.
One of the most important processes a writer should go through is finding their voice. This process is not something you should skip as a writer and sometimes it takes a lot of time to see/hear what your voice sounds like. Sometimes it takes years before you understand: ha, this is me. Or this is not me. AI cannot do this. AI provides you with texts that are written from an AI voice. If you were to use AI, you would have to be sufficiently aware of what your voice sounds like to edit AI into your own unique voice. And that requires a lot of creativity.
Oh, heck, yes–voice is so core to every author’s style and work and craft. And yes, can be so hard to find and fully free. I’m sure AI doesn’t really facilitate that process–and I know from firsthand experience like this one I’m discussing in the post that, at least at this stage, it works against it, homogenizing the author’s voice. That said, I’ve talked to a lot of authors–some who have commented here–who have found ways to incorporate it usefully into their own process in a way that doesn’t impede their voice. But as you say, it’s a skill and it requires vigilance. Thanks for the insight, Greet.
I am cheap so use the free versions of Chat GPT and Claude. One advantage is the limited number of questions I can ask in a given sitting. While frustrating at times, the limitation does force me to think about the next questions or get busy with writing based on the answers given. For example, I am working on a story set in France during and after World War I. Claude was able to give detailed information on things like the evolution of trenches, the Great Mutiny of 1917, shell shock (PTSD in today’s terms), what field hospitals were like, etc. After the war there is a death and Claude outlined the nuts and bolts of a Roman Catholic view on death and the funeral set in rural France in 1920. So, as a resource Chat GPT and Claude are outstanding. As a substitute for writing, I am selfish enough to say to those authors seeking a 100% AI-generated book, go for it. I believe my stories will fare better in the long run.
I do think AI can be helpful in collating unimaginably large data sets in offering the kind of information you’re describing, Keith–but in line with a couple of the other commenters, I worry about the impact of the tempting ease of doing so on our planet (it requires unbelievable amounts of energy). So far I’ve found I can meet most of my research needs by doing my own old-fashioned (ha!) internet searches, though of course now so many automatically push the AI-generated answers on us; it just takes a little more digging and reading on my part.
But that also lets me avoid the other flaw in the LLM models that another commenter has pointed out: that they can yield faulty, false, or even damaging information. Doing my own searches lets me verify each source of the info I’m getting, and I find I prefer that accountability and reliability. (It also keeps my brain working, which is one of my many concerns about this technology.)
I hope there still aren’t too many authors generating totally AI-created work, though I know there have been a flood of these titles coming onto the market. I think there’s going to have to be some significant regulation and law drawn up around these technologies–something I’m afraid I’m not hopeful is among the priorities of the current administration. Thanks for your comment, Keith.
For me, the most beneficial application of AI is creating content for social media. I prefer to spend my time crafting a novel, not FB and IG posts. I am learning to leverage AI to create a content calendar and branding.
ChatGPT is also a powerful research tool since it aggregates data so quickly.
As far as using it for writing, I LOVE the creative process. That’s why I write. I wonder if some “authors” tap into AI to get a quick product, therefore bypassing the lengthy process of creating a unique work.
Please don’t fall for the hype, or at least don’t help spread such false claims. ChatGPT is NOT a powerful research tool. And due to how it works underneath, it cannot be. To put it in simple terms: ChatGPT doesn’t understand the texts it consumes; it generates its output based on probabilities, i.e., which word is most likely to follow next. (This is why all the texts are so generic and bland.) If you want to test this, just let ChatGPT do some multiplications or ask it to count letters or similar – it’ll get so much wrong, you’ll never trust it again.
ChatGPT also constantly makes things up. Many companies have already learned this lesson the hard way, and some have done enormous damage to their own reputation (one of the earliest and most “viral” examples being that of the lawyers from Levidow, Levidow & Oberman, who submitted fake cases made up by ChatGPT in their legal brief to the court). If someone wants to use ChatGPT for research, this person must be very familiar with the topic so they can spot made-up information and false claims in ChatGPT’s output. But if you’re already an expert yourself, you wouldn’t need to rely on ChatGPT to begin with.
So please don’t spread the Silicon Valley lie that ChatGPT & Co. can help and speed up research. These LLM-based tools cannot do that. Everything must be fact-checked and verified.
I think there are valid arguments to both points. AI can be a potent tool for research–the power it has to draw on a vast array of data and correlate information is already creating advances in medicine, science, and other fields. And yes, as you mention, it’s also fallible and flawed. I do think it will continue to improve quickly–the pace at which it’s already advanced kind of blows my mind, and I suspect that within a decade or less it will be exponentially more powerful and capable. Thanks for the comment!
I find that both brilliant and disturbing. 🙂 As someone who doesn’t really love social media, I LOVE the idea of farming out the work of it to the machine. And yet…I also believe that what makes social media work as a viable method of outreach for authors to their readers is the genuine connections it can foster…which I worry may feel less authentic if we rely on LLMs to create our posts. (Not judging your use of it–I seriously think it’s brilliant.)
Chat GPT can be useful for research, yes, but as I commented to Keith above, there are ecological considerations, mental ones, and also issues of reliability with that.
I don’t know how widespread the use of it in authors’ actual writing processes is–but I know some authors I respect very much use it to facilitate at least part of their process (exactly what and how much, I don’t know). I’m really exploring my feelings and thoughts about this still. I think resistance is futile–the horse is out of the barn, and progress will happen whether we resist it or not. But I do hope we hold on to the best of our humanity and creativity. Thanks for the thoughtful comment, Dawn.
One of the most hilarious examples (to me anyway) I found to illustrate the unreliability of AI is this: I wrote a story about a Nazi bigwig. I made up a character, his fictional mistress. My sister asked ChatGPT who this Nazi bigwig had a relationship with in the 1930s. And he named my fictional character! Not that funny, really, when you think about it.
I can see the benefits that AI provides in many aspects of life, but I don’t believe that a tool creates. Humans create. If it assists with creating, shouldn’t it also have its name on the cover?
Despite the fact that I’m missing a third of my lung capacity due to a viral infection in my lungs over a decade ago that was incorrectly treated, I hike. I put boots on my feet and walk on a dirt trail. I carry a pack. I do it for the experience, but also so I can get better at it and experience more trails. I’ve learned to manage my lungs and do more than I probably have a right to do. I climbed two mountains last year.
I’m a writer, now an author. I create what hadn’t existed drawing off my experiences and using my imagination. It’s the literary version of “hands on.” I literally wrote a million words that’ll never be published just so I could learn my craft. I spent long hours finding ways to edit and revise so that what I crafted was a product of my creativity.
I loved every single minute I’ve spent doing those things.
Even if I didn’t publish, I’d still do everything I shared above. Would AI, a thing, do the same?
I’m a firm believer in learning to do, not learning to cheat the system. When it comes to art, cheating the system is fooling yourself.
It doesn’t really create at all, though, as you say–it regurgitates the collated product of others’ creation. That said, I’m of the opinion (colored perhaps by too many sci-fi movies) that it’s not out of the question that AI may develop the capacity to actually think and create eventually. And then…? I don’t know where we are. And like you, I am one who likes a hands-on approach to most things–not only is the doing of it usually where I deeply learn something, but it’s often the point of it: process, not product, to paraphrase your point, as I understand it. Thanks for this considered view, Christina.
Thanks very much for this. I think there are writing fields that are really threatened by AI, such as technical writing, but as for creative writing, the reader’s experience is one with a unique and individual author. It really is a *human experience*, a personal experience, of co-creation while reading a specific artist’s words and coming to know their unique voice and eccentricities, and even their flaws–the occasional awkwardness that somehow works, the breaking of a rule here and there–never a flattening out and “sanding off the edges.”
James Baldwin said: “You read something which you thought only happened to you, and you discover that it happened 100 years ago to Dostoyevsky. This is a very great liberation for the suffering, struggling person, who always thinks that he is alone. This is why art is important.” I would add, as far as AI (which neither writer had to imagine), that the “thing that happened” Baldwin speaks of only matters because of the personal experience of encountering it through the unique voice of another person through art.
It’s sad to hear of new writers so desperate for immediate success that they feed their ideas into a machine to “sand off the edges,” when so often those edges (grammatical issues aside, but even those to a degree) hold the seeds that make the writer unique. The edges are often where the voice resides. You don’t want to remove this material, but to develop it and learn to manage it. And yes, that takes time, and dedication to learning the craft, which further leads you into what is unique about your voice. That’s the amazing journey. You skip this step you are actually “sanding off” your potential. If it weren’t so sad, it would almost be comical to see writers throw away their uniqueness for a quick and imagined “success” because, after all, there’s hardly any money in this game. But it’s not about the money and never has been. Like you say, the process of creating art is the reward, and those lucky moments when we find out we’ve reached another human with it.
Beautifully worded!
I tend to agree with you, Steve, about the human connection of writing–or any art, really. It’s what I value in both consuming and creating it. But I do think there’s likely a segment of the market that would be just fine with AI-generated, formulaic stories that hit the right chord.
I love the Baldwin quote you mention, and the way art not only reveals and illuminates our humanity, but creates that invisible bond between artist and audience, and on an even broader scale with the characters who come to life for us and with whom we relate and connect. That’s the part of creative work that I think I love the most, and what I worry is lost when we farm out the process to machines.
But it is work, often, and I agree with you that I don’t think shortcuts to that get us that depth and resonance many authors are looking to create. Thanks for sharing these thoughts. It’s interesting to me to hear others’ perspectives on this–it’s clearly a galvanizing topic!
This is such a thorny subject. The world out there is full of AI products, and while we have the right to use them as needed, there is a deeper issue here. AI writing bots have been “trained” on stolen material from literally millions of published authors’ books. Current estimates are at 7.5 million books, taken WITHOUT the authors’ permission, knowledge, or payment to use their work for AI training. This is copyright infringement by AI developers. Three of my novels and a short story were stolen by Meta and fed into these AI bots. I gave no one permission to use my stories, received no acknowledgement, or payment. I paid considerable money to copyright my work with the US Copyright Office and still my work was stolen. So when writers are using any AI bots to improve or create their manuscripts, they are using stolen copyrighted material from other authors. Do writers really want to be a party to this kind of copyright infringement with their fellow writers? And while there are class action suits by these published authors against Meta and other AI companies, this likely won’t be resolved in the courts for years. I would ask all writers to consider the far-reaching effects of using AI writing bots and copyright infringement. The greatest writers in the world like Dickens, Hemingway, Poe, Virginia Woolf, Jane Austin, Bronte sisters, Mary Shelley, etc., all wrote from their heart and soul and honed their writing skills into a well-developed craft. This is the path to becoming a talented writer and producing quality material. I am, as many of my author friends are, sick over this violation of our work, infuriated, and feeling powerless to halt this AI-plagiarism. If you are a true writer on the marvelous journey of creative writing, then please follow your heart and soul into the fictive dream and the utter joy and rewards of the creative journey.
You bring up such a great point, Paula–and one I’ve written about before as well. The ethics of how these engines have been trained–without opt-in or compensation for authors–are pretty appalling (like you, all my books got fed in as well). And like you, I think about the morality of eating from that harvest by availing ourselves of the data misappropriated from other authors. I think we all have a responsibility to think about that and decide whether we want to be part of it–but I also know how easy it can be to think, “Well, if others are reaping the benefits of this technology anyway, why should I disadvantage myself by holding some moral line?” I think the same can be said about the ecological impact–every time I resist using AI for that reason, I admit I do wonder what impact my little abstinence has in the face of the millions using it.
And then I remind myself that by that argument, we just roll over and allow any detrimental force to take over. I call my congresspeople almost every day, despite knowing how little impact my thoughts are likely to have. I donate, I protest injustice, I minimize online shopping to help fight our towering waste problems and contributing to more cars on the road, I engage in efforts to get out the vote one person at a time. I used to volunteer at an animal shelter once a week, walking the dogs and encouraging adoptions. All those are teeny, tiny efforts in the face of massive problems that my single little contribution makes no real dent in. But if enough of us do it, then it does. That’s what I tell myself with AI too. Will we achieve critical mass? ¯\_(ツ)_/¯
But I know that making the effort matters to me.
Thanks for your words and your exhortations toward pursuing our own wonderfully human creativity.
Thank you, Tiffany, I appreciate your thoughts on this. I know the discouraging feeling about not making a dent. I guess it’s more about being true to ourselves and following our calling. If you are a true artist, then you love the act of writing, the struggles, the rewards, connecting to the sublime muse as we travel the fictive dream. Here’s where the writer’s intuition is the gemstone that AI will never be able to touch. There is a precious quality in such a personal journey. Catherine’s point below that AI will produce less skilled writers is significant. For sure, the AI journey will produce fewer pure artists in our world. And that is a great loss.
I think each of us has to find the process that works for us, and the tools we use for our work. I’m still figuring out my thoughts about generative AI–the comments here have been so interesting!
I keep coming back to the fact that it’s the doing of the thing that is such a value for us–and the way it connects with other people. I hope those things are timeless, and not replicable by machines–but even if they are, I can’t imagine creators will stop creating.
You know it, I use ChatGPT as a research assistant and for editing feedback. The keyword here is *feedback*, as opposed to rewrite.
I ask ChatGPT for feedback on my chapters or outlines, on tone, voice, pacing, or specific scene details.
Sometimes, despite telling it not to, ChatGPT will spontaneously offer rewrites that inevitably strip the text of anything that sounds like me. I then have to steer it back in line with something like “No, I don’t want you to rewrite it. I just want to know how you perceive it, whether the sequence of actions that takes place is coherent and easy to understand for readers.”
ChatGPT, being the sum of all average readers, is great at analyzing your story from the perspective of a market average. While I believe this provides me with a valuable external gaze as I write, it also means that if I strictly implement all of its suggestions, I will never achive more than an average work of commercial fiction in every aspect.
That leads me to something I keep repeating: do not let it think for you. Let it be the assistant that will pull and synthesize more data than you ever could in 2 minutes, ask it (as I do) how to make cockroaches glow in the dark, how much C4 is necessary to blow up a kitchen with 24 inch thick walls (my FBI agent has a field day everytime I open my browser); ask it how your story fits within a given genre, whether your character uses the proper military slang, ask it whether your scene hits the mark in terms of conveying X or Y, but ultimately…
YOU do the thinking. You decide what bits to keep or cut, what words to use or not. Verify the info ChatGPT gives you, and keep learning from sources other than AI. Let it expand the horizon of your knowledge, vocabulary or vicariously lived experience, but don’t copy-paste its answers.
And a final tip: if a single conversation gets too long, ChatGPT’s answers may become chaotic (and therefore mislead you into thinking that your writing missed the mark because the AI understood the opposite of what you meant), to avoid that, create a project and group shorter chats that each tackle a specific plot or research point. This allows ChatGPT to rely on the context of your WIP (it will remember who the characters are from one chat to another, what their personality traits are supposed to be) but remain coherent withing shorter, more targeted conversations.
Ditto. I do the same.
I’d say I’m a savvy user of ChatGPT. But I also know my story craft, having studied it for almost a decade.
I never use the AI tool to generate new prose because I’m an artist. I do the creating.
What tools like ChatGPT offer the creator is a chance to experiment with instant feedback. Experimentation and feedback are essential to learning and growth.
Even so, I’m not about to convince someone why they should or shouldn’t use them.
My view: do what works for you as the artist.
Sometimes, a pen and paper is still how I create my best work. For new scenes, it’s my go-to starting point.
I hear this from a lot of authors–that the programs serve as kind of a brainstorm buddy or sounding board. I haven’t tried using it that way, but I’m curious. Like you, I don’t try to convince or tell anyone how they should write or what their process should be–I’m always interested in hearing other people’s, though. Thanks for weighing in, Jane.
Thanks for sharing this, Camilla! I know a lot of authors who use the technology this way–as kind of a brainstorming partner or devil’s advocate. That’s in line with how I check my presentation outlines against it sometimes, to make sure I haven’t left out anything an author might expect to see addressed. I’m glad you keep the little bastard in line when it tries to take over your writing, though. 🙂
Your point about it generating average work is especially welcome–it’s what I meant about the lowest common denominator in the post, and I think it’s exactly why the author’s manuscript I talk about was so stripped of its voice and originality and life. AI can’t think for us, as you say–or anyway, we shouldn’t default to letting it. I think that’s when we start to lose some of the capacity we have for imagination and creativity and genius. Thanks for sharing your take on this. I was hoping you would, knowing you are a proficient user of the LLMs.
I am writing a hybrid memoir stuffed with research, and I am a trained and experienced qualitative researcher. My methods are rigorous–I read tons of literature (books, articles, studies), and I attend webinars, and I interview people, etc. etc. I collect and produce massive mounds of data that needs to be analyzed and made coherent and engaging.
I am finding AI a powerful tool to aid in summarizing articles and interviews. My primary AI sources have been Otter (will do now while transcribing), Zoom (will provide summaries) and Microsoft Word’s new feature, copilot. I don’t find Zoom very useful, and I don’t find Otter to be very accurate (checking and correcting takes almost as much time as I’d spend transcribing the work myself), but I must say I’m very impressed with copilot.
When accurate, AI summarizing is saving me hours of work. But I would not use it for content creation. It might do a decent job, be okay for a routine task, but at a deep creative level, it just feels so wrong.
My husband prefers Copilot too–and like you, I had issues with Otter, which I used to use as a transcription tool for my interviews. You’re right–I couldn’t publish them without extensive editing, which wound up taking so much of my time that I discontinued using it.
What you describe is what I mean by using AI as a tool–where it can save us time and streamline our grunt work to let us move the ball farther down the field, I embrace it. I now dictate a LOT of my blog posts and editorial letters, which I believe is an AI-powered transcription program that Word and Google use, and man, has it skyrocketed my productivity. And I find it enhances my creativity in those areas too–being able to freely speak my thoughts feels much faster than typing and lets me keep up with the pace of them in my mind more effectively. It doesn’t work with all kinds of writing, but where it does, it’s a major time saver and creativity enhancer. I’ve heard of people using it as you do too–to ingest and summarize major quantities of research, and collate it, and I can see how that would increase efficiency as well.
Yet I’m so torn by it, ethically, ecologically, and for what it means for us as individuals, for our world, and for society. That’s why I keep calling it a tool–most tools come with the potential for good and for evil. I think we have to remember that we are the ones who wield them in ways that may lead in either direction. Thanks for your thoughtful comment, Wendy.
Thank your for this piece, Tiffany. As always, very thoughtful and coming from the heart.
I’m a neo-Luddite (and I’ve been in tech for decades) so am totally against using AI to help with my creative writing. I watched the effects on my mind from social media (heavy use of AI) and got off it. The last thing I want is for machine code to get into my writer’s soul. Is it a slower process? You bet! Call me a dinosaur, I’m ok with that for now.
(AI probably could have caught that typo in the first sentence though! No way to edit my comment, sorry. :P)
I feel you, TR! I don’t use it to help me write anything (except the bad examples I mentioned in the post)–mostly because it feels like a cheat to me, and if I use it I think it would undermine my sense of authorship and creation. But I’ve been called a Luddite more than once–my husband jokes that I’d happily live in an Amish village, and he’s not too far wrong. I adapt, because I have a business that depends on a lot of the technology I’d have resisted otherwise, and that business is important to me. But a little part of me dreams of retiring to a desert island–no phones, no social media, no wireless… I often wish we could put teh genie back in teh bottle in a lot of areas, but alas…here we are. Change or die, right? Thanks for commenting.
I agree with everything you’ve written. I am often asked to beta read, and I have quickly learned to ask if and how AI has been used. If I’m going to beta read for free, I want to enjoy what I’m reading, and I have found that AI generated books often fall flat when it comes to story, character arcs, and dialogue (lacks humor, nuance, speculation, inferences). I find it tedious to read, so I’ve been passing.
AI is here to stay. I am curious to see how some authors might use it and similar to learning to sit down and finding the right words to write a book, how well some might do when it comes to fine-tuning the AI commands to reach the story they want to tell.
This is what I was contemplating myself, Teri–when I thought the author I mention in this post might want me to work on the AI-polished version, I really worried whether I would be able to effectively do it, and also how much we would be having to go back and reinvent the wheel along the way too–readdressing so many areas where this author had already advanced well beyond what the LLM version regressed his story to. I agree that they feel tedious to read, and I strongly considered specifying in my website and contracts that I wouldn’t work on manuscripts written with the help of AI.
But as we see just in the comments to this post, so many authors use it in so many ways–where would I draw that line? If they brainstorm with it? Research? Do grammar checks? Ultimately I decided against it. I’ve been fortunate that the authors I’ve gotten to work with are all deeply invested in growing not only their stories, but their own skills, and so I’m going to trust/hope that I won’t be getting a lot of inquiries about stories that more heavily lean on the tech in the writing itself. Thanks for sharing your thoughts on this!
I love this, to be honest. Let them all feed their work into AI.* Less competition. Less copyright competition. Less skilled writers.
An added bonus is that it corrupts the language model. Yes, these language models get dementia. One study said 5 years or less to unusability, which is why the thieves that made them are so ironically protective of their babies.
People don’t want that stuff, and for good reason.
*kidding, and yet not.
I wouldn’t be sad if these things made themselves obsolete in five years, but unfortunately based on how most tech innovations have gone over the years, it’s hard to imagine that will happen. I think it’s likelier that they may gain even more power and ubiquity. And a lot of people do seem to want it, from what I can tell. I do hope we harness it for good, and it’s already doing a lot of good things (its medical applications have been pretty promising so far, for instance). But like so many inventions before it, I think it could also lead to consequences we didn’t want or foresee. Thanks for weighing in, Catherine.
I agree with most of what you said. But on a related note, we all should strive to be clearer with the terminology. “AI” is just an umbrella term for a wide variety of techniques, and it has existed long before ChatGPT and the current generation of LLMs entered the scene. For example, “AI” also includes very valuable tools and systems that help in the medical sector (e.g., tumor detection and similar), in logistics (e.g., find the most efficient and least wasteful routes), or even when browsing the web (AI-optimized web traffic allows us to access content faster). But none of those tools are as wasteful, as harmful to the environment, and as much based on massive theft of copyrighted works as ChatGPT & similar tools in the “generative AI” subcategory of “AI”.
The techbro idiots are very much aware of this and have very purposefully highjacked the term “AI” to make it impossible to reject these harmful products in the GenAI subcategory. When writers, musicians, translators, voice artists, etc. say things like “we don’t want AI” or “AI is hurting humanity” or anything along those lines, the techbros will instantly argue that this makes you a technophobic luddite who doesn’t want any progress in medicine, logistics, etc. – because that’s also “AI”.
Essentially, the term “AI” has lost all meaning nowadays. It’s just a marketing buzzword now, and everything gets the label “AI” whether or not the respective product/service does anything “AI”-like at all. This makes it harder to differentiate between the good/meaningful/sophisticated uses of real AI and the bad/useless/immoral uses of GenAI. The latter is what we should all reject on principle until the environmental and legal problems have been resolved. And we can only reject it if we’re more specific with our words and terminology.
The TL;DR here: Let’s all be clearer with our words and not say “AI” when we mean “GenAI”/”Generative AI”!
You make plenty of excellent points here–and the one about being more specific in what breed of AI we’re referring to relative to the topics of the post, the large language models, is particularly well taken. Thanks.
Hi Tiffany,
From one writer to another: do you really, truly love writing–do love using English language efficiently and gracefully? Does this piece reflect your ability to communicate in as few and as precise words as possible? Did you use AI for any of it? Take a look at how often you repeat “work,” and variations of “depth” and “deepen” in the opening paragraphs. Are all of those the best choices available to you? If you want to claim the author-ity inherent in being an author, you might start with tightening your own prose.
Hi, Peter. I do indeed very much enjoy writing these posts each week, and no, I don’t use AI for any of it. The blogs are my own informal communication with a writing community who have come to feel like friends, not a heavily edited formal piece of writing.
Wow! Thanks for your valuable insights. I don’t use AI and don’t plan to use it…except when I google a plant in my flower bed (legit or weed?), AI Overview intrudes on the response. I’ll have to be very careful with internet research.
Yeah, it frustrates me too that the default results now seem to be AI generated. But like you, I just keep sifting through the sourced results the old-fashioned way. 🙂 Thanks for the comment, Margaret.
I just recently started to use AI with my writing. I resisted for many of the reasons you mentioned – art has a higher and deeper purpose and the power to move, affect, and change people, to impact society and our world. Our humanness, our very souls, is what breathes life into what we write. No computer, I don’t care how advanced, will ever be able to do that. I think of Data on Star Trek. As much as he tried and wished to be human, certain things still escaped his ability.
I use AI to help me learn writing skills and techniques, such as checking if I wrote something in third person limited, or to help me with awkward sentences or paragraphs, so that I can make it flow smoothly. Sometimes, I use it to help me with tone and provide examples that I can then extrapolate and use in my writing. One of the biggest areas AI has been helpful in is research regarding historical information, since part of my novel takes place in 1700s Boston.
What a great and timely piece. I believe that, in addition to writers and creatives, AI is impacting and affecting so many other professions as well, and trying to determine its appropriate and ethical use seems to be on many people’s minds.
PS. Ha! I love the GPS piece. I used to have many AAA maps in my car, but I have become GPS-dependent and appreciate the wake-up call to use my brain more when driving. The fact that I don’t even know all the names of the streets in my neighborhood drives home your point. Growing up, I knew all the names, even the ones in the next neighborhood.
That’s an interesting use of the technology, Emily–truly a tool for helping you deepen your understanding of craft. Yeah, I know a lot of folks and fields are grappling with the ethics and uses of AI. I see such amazing things it can facilitate–medical advances and diagnoses, helping people with disabilities–and I even read it’s useful in combatting human trafficking. But then there’s also issues with the misappropriation of writers’ original work, or the “hallucinations” and biases that seem endemic to the LLM engines, and some of the issues we’re talking about here.
And you’re right about street names! I realized when I first started doing this that I’ve been visiting my mom for six or seven years in her new house, and couldn’t tell you a single name of the roads I take to get to her house. When I started paying attention I could feel my brain waking up, and feel myself paying more attention–I like it.
Most everyone commented on everything already on my mind, so I don’t have that much to add. I do have this fear that books and other creative works will become so saturated with AI-generated content that the human artists will kind of be pushed out to the fringe, and will have to charge much more for their art. Human-authored books may end up being one of those things you find at art and wine festivals and farmers’ markets alongside handmade pottery and clothing. If AI gets good enough at satisfying readers, why would they pay $50 for a human-authored book when they can get an AI book for $5? My thoughts exactly when I need to replace a mug in my house. I’m not going to spend $35 on a hand-glazed work of art when I can get a perfectly serviceable one at Target for $3. Unless, of course, mugs are super important to me and that’s what I want to invest in. So let’s hope human-authored books remain super important to readers.
I think about that too, Cate—with almost exactly the same parallels–how we’ve transitioned away from artisan-made goods to mass-produced ones in almost every area of our lives. Like you, I desperately hope that novels and paintings and poems and music and sculpture and every other art form doesn’t go that same direction. My gut tells me it won’t–art is as old as humanity, and so much a part of that humanity. But what do I know…? I’m going to keep creating, though, because it’s the process that gives me joy. Thanks for commenting.
I won’t go near AI. Not for me, never, ever.
Ha! I can understand the sentiment, but you might be surprised how many elements of it are already seeped into our lives that we don’t even realize…!
I have some terrible writing habits that pro writing aid points out to me. It is incapable of writing an authentic voice and I disagree with 3/4 of the revision suggestions. It points out my bad habits and lets me see a paragraph differently. AI is useless on quotes, poetic languages, and dialogue. It can approximate a story and write the exemplar of a conclusion, it never a satisfactory conclusion. It will help communication in business and catch errors with details. However, I think every writer should cite it’s use when writing.
You sound like you handle AI the way that author Camilla Monk suggests in her comment–not letting it do the thinking for you. That’s what I mean by our keeping in mind this is a tool, and you’ve developed what sound like some helpful uses for it in your writing.
I love your point about it being cited! I think that’s part of what bothers me about the technology–I almost feel cheated or duped by work that incorporates AI-generated writing and doesn’t specify that. Thanks for the thoughtful points, Margie.
Having written only two novels, I’m leery of using AI for anything other than research, punctuation, grammar, and spelling. I fear its siren song that offers easy but mediocre writing in an undead voice.
However, MS Copilot usually gets me into the ballpark when I need to explore facts, myths, or lies. It’s much faster than Googling. (Actually, I use Duck Duck Go.) Ordinary web searching usually produces a rats nest of links that must be sorted and investigated. Copilot sorts that stuff and provides it’s sources as hot links, so I waste less time.
I recommend the novel, DO YOU REMEMBER BEING BORN? by Sean Michaels. Michaels is a bonafide AI expert and creator. His novel is about a famous poet who is hired by a Google-like company to collaborate on a poem with their new LLM software. Parts of this fine novel are actually written by AI, and the prose is clearly delineated by italics. I couldn’t help but to scrutinize and judge those parts.
That novel sounds fascinating, David. Thanks, I’ll check it out. That did you think of the authors use of AI in the writing–did it work? Was it part of the conceit?
I’m still enough of a Luddite that I check several of the actual individual sources, making sure that they are ones I feel are legitimate. I do worry about the misinformation LLMs can be so prone to–but I know the AI summary can be a time saver. Nice to see you here!
Using it for creating bad writing to use as examples – brilliant. They’re so painful to write from scratch! Thanks for yet another great post, and HOW do you have the time to respond so kindly and thoughtful to everyone, even those who are obviously jealous of your brainz (squints meaningfully in Peter’s direction)?
I KNOW, writing bad examples seems to take forever! (And AI is so good at being bad at this point… 😂) Thanks for this comment, friend. ♥️😉
Thank you for this timely, thought-provoking article, Tiffany. I have used AI to generate outlines for some of my recent blog posts. Because I’d been warned to ask for real references and that some of those will be bogus, I always check them. In 6 months, I’ve found that approximately half of the references are inaccurate.
When I asked my writing group to critique a blog after I’d used AI, their comments were reflective of your concerns: generic, lacking a personal voice, sounding “cliche’ed” (is that even a word?). They suggested that I add examples from my personal and/or professional life to round out and make the blogs my own. I followed that advice.
I understand that when I use google to find robust research findings, that, too, is AI. So, I’ve actually been using AI for five years. I can’t picture using AI to write a novel but have attended some webinars where presenters encouraged authors to keep an open mind, to learn more about providing excellent prompts, and to at least explore it. I plan to take some classes about using highly specific prompts.
Wow, half seems high! But I can’t say I’m totally surprised. I do think right now the data is flawed, and I am always careful to source my information anyway.
I do think generative AI I can be useful and helping with some of the busywork, but I don’t ever use it for any kind of writing. I think for me it distances me from the writing and doesn’t feel fully authentic. (Except for the bad examples I referenced in the post.) When I do use it, it’s as an adjunct, for instance double-checking that I’ve covered everything I should be covering in a presentation, as I mentioned.
I agree with you in that I try to keep an open mind though, knowing that everyone has to figure out what works best for them and what they are comfortable with. Thanks for the comment.
Tiffany, a week ago I would have said a resounding no to AI; however, I’ve used it and I think successfully, at least I hope. While writing my manuscript, I put my character in a WWII training plane. My father, a pilot, was coaching me on stuff I’d never figure out. But, he died. AI to the rescue. After asking ChatGPT all kinds of questions–the look of the cockpit to what would a pilot do if…–it asked me if I was writing a scene and would I like help. Gulp. Yes. It wrote several scenes for me, making all kinds of mistakes. The scenes, however, helped me visualize and suggested skills of a pilot that I would not have conceived. It was kinda like the encyclopedia talking to me. Have to chuckle at that one. Won’t use the scene it wrote. Some of the ideas? Yes.
I’m sorry about your father, Robin.
That’s an interesting use for the technology, allowing you to get a sense of an experience you haven’t had directly. I wonder how much of it was accurate, did you check with anyone? I’d be curious to know.
The way you describe using AI is the way that makes the most sense to me, kind of as a brainstorming partner. I think we have to do the heavy lifting ourselves though, or the muscles atrophy. Still, everyone can use a spotter now and then. 😉 Thanks for the comment.
Wow, half seems high! But I can’t say I’m totally surprised. I do think right now the data is flawed, and I am always careful to source my information anyway.
I do think generative AI I can be useful and helping with some of the busywork, but I don’t ever use it for any kind of writing. I think for me it distances me from the writing and doesn’t feel fully authentic. (Except for the bad examples I referenced in the post.) When I do use it, it’s as an adjunct, for instance double-checking that I’ve covered everything I should be covering in a presentation, as I mentioned.
I agree with you in that I try to keep an open mind though, knowing that everyone has to figure out what works best for them and what they are comfortable with. Thanks for the comment.
Hello Tiffany,
I would like to recommend an excellent book to your readership on the ethical use of AI in writing:
Story Ideas Unleashed: Story Idea Creation and Concept Design (third edition) by Peter von Stackelberg which is due to be released on May 1.
https://www.amazon.com/Story-Ideas-Unleased-Creation-Concept-ebook/dp/B0F438SJP4/ref=sr_1_1?crid=2XR5AGUTOA74D&dib=eyJ2IjoiMSJ9.nQl6Dkv5uPwKScoxUCz43rZpNF0Of1ETzZCmubISx6Q36TGwLOD0TErkMydsyqH0.mG3-ZsaLtFPnohHOkrFqwgIRQJ5YP0gxT6wKh2tH9rI&dib_tag=se&keywords=peter+von+stackelberg&qid=1745242443&sprefix=%2Caps%2C274&sr=8-1
In the cause of transparency, let me explain my involvement in Peter’s project.
Because I was agonizing over the second draft of my novel and because I was fascinated by some of his articles, I purchased his book.
Rather than digging down further into the holy trilogy of plot, character, and theme, it felt right to take a step back in the process and review story idea and design concept.
What I had not anticipated was the AI component that his book also addresses; I had bought it for basic review to reestablish my vision for my novel before I proceeded with my outline.
I wrote to Peter, and we have been corresponding ever since. At his request, I also read his manuscript and offered suggestions as I saw fit.
It answered my question: when I feel my foundation is uncertain, how can I move forward? But there is so much more – AI prompts to elicit answers that can augment the creative process.
For example I asked Chat GPT 4.0 what is the average word range for historical mystery thrillers in the last 50 years, and within 3 seconds I got a response!
I believe there is much work to be done on the ethical use of AI in writing. But then again, I imagine the original incarnations of papyrus, dictionaries, thesaurases, word processing programs caused great consternation as people began to use them and knowledge and comfort replace ignorance and fear.
Now, Tiffany, you also know why I am late in delivering my first 50 pages!
Best wishes,
Val
Ha, it sounds like you found a project that drew you! Thanks for sharing the story and suggestion, Val. I have complicated feelings about using AI, as you can tell–much of my reservation stemming from concerns about its impact on our environment, and on society. But I see it can be useful for a wide variety of applications. I hope we find ways to harness all its potential for good, and mitigate its deleterious ones.
Tiffany,
I totally agree!
Val