I’m Building a Bot, But Not for Books.

Authors are asking big questions these days, such as whether they’ll ever use AI to write a phrase or more of their books.

Maybe you saw that New York Times article about the romance author who wrote 200 books last year. Yep. Did it all with AI, and proud of it. She’s teaching other people how to do the same.

She’s on a Facebook group where I lurk and learn that some people now push a button, go to bed, and wake up to read what ChatGPT or Claude wrote them overnight, over coffee. Then some of them head straight to cover design, formatting, uploading, publishing, and marketing, right away. Some edit a lot and change every line of the bot’s draft.

Some churn out these books–a few a week–to the tune of six-figure incomes annually, and see their readers gobble up the stuff.

From my observations–don’t quote me, totally unscientific!–the folks on this group seem to write a lot of romance, erotica, thriller, cozy mystery, mystery, sci fi, fantasy, and more. I haven’t yet clocked a young adult author or literary novelist, but who knows, they very well could be there.

There’s a Formula

One YouTuber (wish I could remember who said this) theorized that the romance genre with its many tropes and formulas makes it easier for some authors to pivot to AI. I recall my aunt used to read multiple Harlequin paperbacks a week. Pulp fiction, books that blur into a baker’s dozen, easy. Same ingredients, same outcome, different colors and flavors of character and setting. Tropes aren’t exclusive to romance; you can be a writer of a certain genre with a formula you like to follow. And now we know that formulas are something you can feed to a bot.

There’s nothing wrong with an occasional Doritos binge. But if all my reads were like Doritos, formulaic and trope-driven books, my reading life wouldn’t feel very exciting. Same for my writing life. Outlines, rules, formulas help–just like I’m sure they helped the chemists design Doritos in the lab.

Also: I have to keep Doritos out of my house, or I tear into bags like my feral kittens do kibble. Those chemists sure know how to engineer my addictions.

(I also happened to binge a bag of Doritos on the night of Tuesday, November 5, 2024. So yeah, there’s all that.)

Books as widgets

For the romance author I’m referencing, books are more of a means to a salary. Books are products offloaded to the digital assembly line machines and workers. I’m not saying this author doesn’t do the work of prompting or master the craft of polishing every sentence a bot builds her; she very well may treat every word that lands on her pages like a precious pearl. What I’m saying is that she views art as a race, first and foremost. Something to be won.

“The New Fabio Is Claude” is a quite a read for our times–a read written by a human about a human using bots.

Show me an author who doesn’t want sales and productivity

I was raised to race–I’m a “good” American, a person wanting to grow a business, earn–and I don’t look down on artists wanting a livable wage/sustainable income. I value efficiency and productivity to stay alive in this business.

I would love to see a six-figure income from my books alone. Would LOVE to hop on that bucket list of travel, quality skin care, private chef, and regular philanthropy. Trust me. Yet I’ve always embraced slow art for many reasons. I’ve always hugged the inefficient for the sake of simmering creativity. So I don’t plan at this moment to use AI to write my fiction. Here’s why.

Worth the wait

Here Comes the Dog Parade, the book I just finished, took me six years to write. I wouldn’t trade a single minute of that process since 2018 for an AI program that could’ve popped a book out in a few weeks. 

That’s because this book was deeply personal. I wrestled with big questions about friendship, race, class, ans forgiveness. By doing so, I grew along with this book. I walked in someone else’s shoes. And somewhere along the way, I traded anger, resentment, and hate for forgiveness and love.

Some of us writers, we write for revenge. And after we work through the pettiest of our feelings, we surface with a realization: who we once thought was a villain was actually a really complicated human being.

I want to riff the old words of the Catholic liturgical rite–“Lord, I am not worthy to receive you, but only say the word and I shall be healed.” Only write the words, and you shall be healed. There’s only one way and that is through.

You must fall in love

There’s nothing like following the lead of your character off road and seeing where they might take you.

Back in 2022, I told my MFA professor, young adult author, A.S. King, “You have to fall in love with your characters.” That’s because I was falling for Ginny and Justine, two girls in Here Comes the Dog Parade who just want to be friends but are constantly thwarted by the subtle and blatant ways their school rigs its system for success. Success for the white kid, Justine.

Before I fell in love, Justine was a spineless, neurotic coward with too many thoughts. Ginny was cold.

I generated so many pages, so many details, all over the place. “Lyn, I’m not sure what this story is about,” I remember Amy saying. “I pace around, rereading your pages, trying to find the point.” I wasn’t sure either.

Knowing Amy was waiting for me to do better put a new spring in my step to wrassle with this problem, and wrassle I did for the next four years. That says something about us humans as social creatures needing human affirmation of our struggle.

The romance author described above might find my process highly inefficient. But I had to go through this kitchen-sink process. I had to climb out of the dish drain, make an outline, and then rip that up, and make another. Go back in the sink with all those characters and then resurface with some of those poor souls spiraling back down that drain. I needed to hear questions from my writer’s group. From Mary Jane, who bravely did a developmental edit. Beta readers. Erin, Jeff, Caroline, Morgan, and Mollie, I see you and thank you.

In my early drafts, I tried too hard to capture the everyday details of life back in the 1980s. Which is fine, but I’m not grabbing a panoramic photograph or making a documentary or a museum gallery. Amy and I talked about readers wanting to read about somebody exceptional: somebody unique or intriguing or powerful or extreme in some particular way. I’m writing a novel about an exceptional, unique set of people in standout circumstances who need to swipe at our heart strings, more than once, and take us on an emotional ride.

Besides the therapy Here Comes the Dog Parade has been, I know that I had to go through the process of taking every single photo until I could create the only ones that needed to stay.

Good news: Justine’s found a backbone! She’s got way more snark and spice to her personality. Ginny, who began a wax figure, eventually grew warm flesh, affection, fear, and pride. She’s pretty complicated. I kinda love them all the more.

What I’m investing in right now isn’t simply about beefing up my own brain cells. I’m sustaining my heart and my soul. Every time I wrestle with a book, I evolve. I grow up. 

AI is capable, yes

I suppose that AI could’ve told me how to do that, just as my professors and editors and writing group members have tried to tell me, but would I have listened? I don’t know. I know AI would’ve done the work for me, too, helped me skip a step of sleuthing out root causes and wrestling with character psychology. I’m talking to other authors who are having success with feeding their manuscript into Claude or some other AI and getting rounds of feedback on what’s working/what’s not. It does intrigue me.

I’m not eager to feed a whole manuscript into a machine just yet. I appreciate that AI could work with an intuitive writer like me and break out the plot beats masterfully, thus making me way more efficient. Plotting isn’t my strong suit. But last year I bought a plot board and played with Post-It notes, using a three-act structure, and used that to talk through the plot with my writer’s group. That helped tremendously.

I love the Post-Its and their physicality. I loved the community of voices weighing in on my story. AI can’t replace that right now. And even if it could one day project a holograph Post-It note board and a chorus of avatar voices, I wouldn’t want that because I’m not writing for bots.

I have a number of rules when I use it.

My mentor for screenwriting calls AI “Google on steroids.“ He says it’s a great research tool; I agree. With it I’ve analyzed things like the history of female country stars, the fencing security needed for an industrial hog farm, the details of medical treatment for stage two lung cancer. I find it far more efficient and direct than traditional search. 

I require it to give me citations and legitimate URLs.

I also use the same rules of academic honesty I teach my students: I never copy more than two words in a row. More, and you should start quoting and citing. I try to do less than that. I’m always rephrasing and rearranging information, the same way I would traditionally write a research paper, while embedding my own thoughts.

If you’re curious what standards I’ve taught my students and held myself to since 1989, here they are in my guide: Should I Use AI When Writing My College Essay?

Some things never change. Who’s writing is this? Mine, yours, or its? I hold myself to the 99.999% rule because I do not ever copy more than 2 words in a row of anyone else’s work. For my students, I recommend 95%, ideally.

LLMs are synthesis and transformation machines that seem to boggle some folks’ minds because of the speed and quantity of output. We have a thing about supersizing in this United States of America. Big is better, mainstream culture has always said.

But I’ve always been a pretty nimble goddess of synthesis and transformation myself.

My number-one goal when writing fiction is to be expressive of my own voice, personality, and soul. My words are my lens and perspective. There’s no point to having a robot generate my phrases and sentences; I’m a creator whose joy is in all the parts of making.

And I would argue to those in that Facebook group who claim they leave no AI word untouched that once somebody else writes the draft, you’re already limiting your playground to the AI’s choices. You are an editor, not an author.

I love editing. But it’s a very different gig.

I resent the AI company’s choices for good reason.

OpenAI, Anthropic, Meta, and Google are getting sued for good reason. All they had to do was ask permission and pay for others’ content. License it for training. Authors, visual artists, musicians, and any content creator with a blog, a web page, a channel. My friends and colleagues got their work pirated.

And yet these companies did not pay. They stole like many founders, CEOs, and other historical figures in America’s capitalist history.

Over 100 years ago, the fortunes of Leland Stanford were built upon the backs of Chinese laborers. Leland Stanford’s railroad fortune built my alma mater. I went to a school funded by a robber baron. We’re talking corruption, political manipulation, monopoly, government grift…never mind racism and brutality.

I can refuse to ride a train or refuse a car (Henry Ford’s politics, anyone?), or use any tools such as Microsoft Word built by a guy who chilled with Epstein. I live in a place where I could burn a lot of gas to do errands and shop, or I could have billionaire-squared Bezos deliver me packages (and sell my books).

I’m swimming in, steeped in, driving, consuming, swallowing the stuff of robber barons daily. So what is an ethical stance?

I hire humans.

I value voiceover specialists, real human actors for my first audiobook, and I will pay for them. I’ll pay for a boutique service run by humans who manage the audition, casting, creation, and distribution process.

I value videographers, who will help me build videos for my business.

I value human editors and literary agents, who will help me improve and sell my work.

I hire human coaches to help young writers express themselves fully.

I am offloading smaller tasks that help me and don’t steal someone’s job.

I’m using AI to gain efficiency with things that take me a long time as a small business owner.

• Building slides for presentations, something I’ve always done myself.
• Researching acceptance rates for colleges, something I do when assisting students sometimes.
• Designing an image for a newsletter that isn’t essential. Situation: have AI do it, instead of using one of my free stock photos that doesn’t match the newsletter’s colors or instead of troubling my web designer to spend an hour, when I need her for more important tasks.
• Brainstorming sales prices and sales language and bringing it to my dad, who’s a sales and marketing guy with tons of lived experience, so he can analyze it.
• Starting an email and creating a quick draft/structure so I can edit it (so far, it’s not the most efficient help because I always totally overhaul the text and keep some of the structure). Call me weird, but I grew up writing postcards and letters, so emails feel very personal.

I ask with every task I give AI: would a human do this better, and from whom am I taking business?

As far as my activism goes: I don’t have time or funds to challenge data centers or Duke Energy because my hours and funds are devoted elsewhere. But I’m well aware of environmental impacts.

Of the threat of super intelligence to our very survival.

We need regulation, stat.

So why would I even consider building a bot?

I coach young writers for a living. According to some, AI can purportedly do all that I do. Yet I know it can’t.

Let me take you on a little journey on the evolution of my thoughts and how AI can be my students’ support assistant, sometimes for some tasks. But always in service to human interaction, connection, and community.

Last year, I saw some students produce pieces that sound like a bad-sales-guy-meets-glossy influencer with pat little pearls of wisdom. Fake epiphanies and summations that sound so polished, you as reader start pulling away from the prose. AI was the culprit.

We discussed why their writing would not stand out and blended into the crowd. How an LLM tends to write like the scrapings of the entire Internet–a stew with way too many ingredients so that unique flavors no longer pop.

Meanwhile, starting in 2023, I’d begun to see vague guidelines pop up on various college admission website pages telling students that they could “use AI for brainstorming and outlining.“ That directive was very problematic to this English teacher who could just picture some students interpreting that as a green light to say, “Okay, cool. Hey, Chatty McChat! Give me five essay topics that will get me into selective schools.” Because that sounds like “brainstorming,” right? Or how about, “Yo, Chat: give me an essay outline that will get me into college.”

I feared for my reluctant writers, fearful writers, I-had-some-adult-who-crushed-my-soul writers. AI help made a bunch of sense for them. The overthinking gifted writers, too. They want reassurance just as much. So I worried about all of them.

Unregulated AI use isn’t for the underage or the faint of heart. To use it ethically, you need a workout mindset: a Write Hard, Learn More mindset. You need to know how to prompt it with tight, ethical standards that preserve your voice.

Heck, you need to know what voice even is. You need to have read widely to be able to withstand an onslaught of AI’s recommendations and walk away sounding like yourself.

But if you’re novice to writing, and fearful and hurting, or you’re really good and need to have every word checked because you overthink…and then you’re handed a free tutor who will do the whole thing for you in 15 seconds?

I mean, the temptation…

Back to widgets

If I had the time or your undivided attention to go into how college applications have become overprocessed product creation, aka widgets-meets-Doritos, we could go there. Some students who are churning out more than 12 applications could be more easily tempted to use AI.

Machine-like performance is being expected from kids who already have very busy schedules and stressful workloads. Adding Chatty McChat to the mix seems like the next logical step. Good leaders and managers delegate, after all. Why wouldn’t they dabble in a tool that everyone says they MUST master to be ready for the future?

What I teach is the joy of writing and creative expression. Success Story celebrates how writing connects with another human. How writing makes you feel. We dive into the specifics of what makes voice and the craft of how you can immerse a reader in your world so they’ll react. It’s the art of narrative-persuasive writing.

We talk about writing practice and neurophysiology hacks to write better. We talk about free writing for discovery and following your intuition. We talk about reverse outlining.

I’ve developed detailed proprietary methods and curriculum about craft, helping students embrace their own style and wisdom. If that curricular knowledge base can fuel a bot’s guidance of a student, and get a student who hates writing to work a bit more on their own when I can’t be there, and then bring me more of what I call Gold, then think about what might happen.

• students could practice more writing;
• my sessions with students could be even more productive, giving us more stories to work with;
• students could think more deeply about their lives;
• and students wouldn’t have to pay for as many hours.

On the other side of this learning curve are big questions to pursue.

I’m also wondering

• What does it mean to share ideation with a bot? Especially when you’re writing a personal statement /memoir?
• What does it mean to share ideation with a bot that tends towards sycophancy? One that is known to “hallucinate”?
• How do you standardize and ensure ethical experiences so that a student of any writing ability can ideate in a way that honors and prioritizes their ideas and thoughts?

I’m also aware of early studies such as this one from MIT, showing that humans revise more when they get human feedback. We’re social creatures, after all. Other studies are showing how we lose our perspective if we work too much with sycophantic bots. The human-to-bot-to-human process is key to this process I’m imagining. Essential. Not just because I want to keep my job. It’s just better for human learning.

And because AI is a synthesis machine, naturally it will “make” stuff for people.

• But if I can instruct it to be an excellent Socratic questioner, what then?
• What if in between human coaching sessions, a bot is a tutor there for student who might need a sounding board? For brief periods, not long-term engagements?
• Who might learn best in collaborative, interactive ways? 

These are the questions I’m exploring with my colleague Mindi Trimble. My answers are developing, evolving, solidifying. In my brain and in my community.

What I’m doing here–blogging–is not productive by some’s standards. It’s not very efficient. It’s stupid, some would say, since OpenAI and Anthropic and Google & Friends are scraping it as we speak.

But I have a stronger sense of self and purpose now after sharing my thoughts with you.

That’s because I write to think.

If you made it to this end, I hope you’re a bit of better thinker for it, in the most human of ways.