My Books Trained AI. Then My Community Shut Down My Class About It.

A letter to my fellow liberals on neurodiversity, gatekeeping, and who the anti-AI crusade is actually crushing.

The kind of independent, local space that runs entirely on grit and goodwill—and the exact community room where a free beginner’s workshop became a modern purity test.

Hi. My name is Ella Barnard.

A few weeks ago, I was going to teach an “Intro to AI” class at a small independent used bookstore in the quiet, working-class North Carolina town where I live.

The class isn’t happening now. There was enough concern raised by customers that the owner, who is a dear friend of mine, decided it wasn’t the right fit for the space.

I understand that decision. Small businesses run on goodwill, and goodwill is worth protecting. But I’m writing this because I don’t think the people who raised concerns ever got to meet the real people behind that class, or understand the ground-level reality of why it was organized.

So—hi. We’re real. I’m real. Settle in, make some tea. This is going to be long. I’m not going to apologize for that, because the topic deserves it, and I think you can handle it.

Let’s start with where I live. This is a beautiful community with a rich history. It’s the kind of place where old textile mills have been converted into apartments, and where main street businesses have to fight like hell every single day just to keep the lights on.

Running an independent used bookstore with a community event space here is a daily act of defiance. My friend Sarah runs this shop mostly by herself, doing the work of fifteen people, not to make a fortune, but because she cares deeply about creating a gathering place for this town.

She was drowning under the workload. I wanted to help her host a workshop to bring some much-needed event revenue into her shop. And honestly? Sarah wanted to sit in the front row. She has ADHD, and she wanted to learn how a tool might help her handle the administrative tasks her brain finds hardest, so she could keep her focus on keeping the bookstore alive for the very community that ended up protesting the class.

I have skin in this game, too. I have ADHD, CPTSD, and depression. I take medication for two of those, and I do a lot of work on the third. I’m a full-time indie romance author—I write books with dragons in them for a living, which is exactly as weird and fun as it sounds, and I run my entire business completely by myself.

On a good day, I’m warm, funny, and productive. On a bad day, getting out of bed and making toast is a real accomplishment, and the gap between what my brain wants to do and what it can actually do is wide enough to grieve.

AI is part of my daily cognitive support stack—right alongside medication, therapy, and routines. It helps me on those bad days. It organizes my thoughts when they’re scattered. It drafts an email when my executive function is completely offline.

It’s not replacing my brain; it’s like glasses for my brain, helping it do the thing it is already trying to do.

When the pushback came in and the class was canceled, what I heard underneath the words was: we don’t want you, or what you have to offer, in this space. I know that wasn’t what people meant, but it’s what it felt like on the receiving end, and I’m trying not to pretend things didn’t hurt when they did.

I don’t feel as safe in that bookstore now, knowing that if I mention casually by the front register that AI helps me draft my marketing copy, I might get a look I can’t unsee.

So I want to talk to my own people for a minute. Because I am one of you, and I think we need to have a real conversation about this.

I know the core reason people have strong feelings about this technology: the training data issue. The belief that AI was built on stolen creative work, and that using it makes you part of the problem.

I want to take that seriously, because I take it personally. My books were used to train large language models. My actual work was fed into the process.

I am not some tech-bro telling you it’s fine; I am a working author whose creative work was used without her permission. But when I actually sat down and looked into what that meant, the picture got complicated.

First, the pirated databases used to train these models existed long before AI. They were built by regular readers downloading and sharing books for free on a massive scale for years. Every author I know has had their books pirated. It’s just the background noise of being an author in the digital age. I made my peace with piracy years ago.

AI companies are not innocent—they knowingly used pirated copies, and that matters—but if I’m going to be angry about my work being taken, I have to be honest that it was being taken long before AI entered the picture.

Second, there is a fundamental misunderstanding about what “training” actually means. And the easiest way I know to explain it is to tell you how I learned to write.

I’ve read thousands of romance and fantasy novels. Every single one of them shaped my syntax, my rhythms, my pacing. When I sit down to write my own book, I am not copying them, but they are all folded into how I think. We don’t call that theft; we call it being a reader who became a writer.

The actual view from the dragon-romance command center. Business developer on the left, accountant in the basket, nap coordinator in the back.

That is remarkably close to what training actually does. People picture AI as a digital vault storing copies of our books, pulling from them to copy-paste text. That’s not how it works. My books, along with millions of others, were used to adjust billions of numerical weights inside the model. Those weights encode patterns—how words tend to follow other words, how a sentence is structured, how a story unfolds.

The AI didn’t copy my book; it looked at it to learn how human sentences hang together. It didn’t keep the ingredients; it just memorized the recipe. My actual text is not in the model in any retrievable form.

The difference with AI is scale and speed, which is the legitimate heart of the consent and compensation debate. But the narrative that “the AI is actively storing my book and copying it every time someone uses it” simply doesn’t match the science. I am one author, and I don’t speak for everyone, but I am okay with my books being part of how this technology learned.

This brings us to the “complicity” argument that so often follows next: the idea that because the training phase was wrong, anyone who types a prompt into ChatGPT today is morally suspect.

The training already happened years ago. My books were ingested way back when. When someone uses AI to draft a letter today, they aren’t taking anything new from an author. Boycotting the tool now doesn’t return anyone’s book or compensate a single creator.

And here is where I need you to be really honest with yourself. If “benefiting from extracted labor” is our consistent standard for moral complicity, it catches almost everything in modern life. Google’s search index was built by scraping the entire web without asking. Wikipedia runs on volunteer labor that everyone else monetizes. Spotify pays musicians fractions of a cent. Most of the clothes we wear were made by people who aren’t paid a living wage.

When this framework gets applied aggressively to AI users, but ignored everywhere else, it ceases to be a principle and becomes a weapon. And it lands hardest on the people who can least afford it.

So let me tell you who actually gets hurt.

My friend Jeremy has diabetes; he uses AI to adapt his favorite recipes into versions that keep him healthy. Is he harming creative workers? Sarah wanted to use it to keep a local, independent bookstore from drowning. She isn’t replacing workers; there was never a team. There was just her, drowning. I am my own writer, editor, marketer, accountant, and web developer.

There is no corporate infrastructure supporting independent creators or small businesses in our culture. We are entirely on our own, trying to survive, and this tool helps.

Consider my grandmother, who was blind. She loved stories but couldn’t read my books, and I could never afford the thousands of dollars it takes to professionally produce an audiobook. With current AI narration, indie authors can finally make their books accessible to blind readers at a fraction of the cost.

And consider the writers I used to coach. For years, I worked with authors from marginalized communities—queer, Black, and disabled writers—helping them build sustainable creative careers. I loved that work. But I had to stop because running my own business took every ounce of my capacity. There weren’t enough hours; there wasn’t enough of me.

I am about to start coaching again. The only reason I can is that AI fills in enough of my administrative workload to give me my capacity back. That means more diverse voices in the world. More people quitting jobs they hate to do creative work they love.

The loudest narratives can only see the visible workers being displaced; they completely miss the quiet, distributed creative workers who wouldn’t exist without this infrastructure.

There is a vast moral gulf between a Fortune 500 company replacing two hundred illustrators to pad corporate profits and a disabled creator or struggling small-business owner using a tool to survive. Treating those as the same moral act doesn’t survive five seconds of real thought—and the people getting hurt by that thinking aren’t the tech conglomerates. They’re the disabled users, working-class creators, and local businesses being told that the tool helping them get through the day makes them villains.

Yes, AI is displacing workers. Voice actors, illustrators, and copywriters are watching their incomes shrink. These are real people, and their pain is real. Both of these truths co-exist.

Every major technology shift has displaced workers. Historically, the progressive response was never to demand that individual citizens boycott the technology; the response was structural policy—retraining programs, stronger safety nets, and support for workers caught in the transition. Refusing to use a tool doesn’t bring back a single job.

The argument that AI is purely “bad for culture” also assumes the old system was fair. It wasn’t. It was heavily gatekept.

Before ebooks and self-publishing, the question of which books got published was answered by a handful of executives in Manhattan. Their conscious and unconscious biases shaped whose stories were told. What got dismissed as “not real literature”? Genre fiction, romance, Black authors writing for Black audiences, queer stories that weren’t tragedies, and working-class voices.

Self-publishing broke that gatekeeping, and the exact same panic happened then: “It will flood the market with junk; real authors will be displaced.” Instead, the ecosystem expanded. More diverse, weird, marginalized voices found their audiences. AI is doing that same work today by lowering the cost of translation, narration, and design. It allows the people who were underserved by the old system to finally participate.

I am a liberal. I have been my whole adult life. Grew up in Northern California, graduated from UC Santa Cruz, and taught in low-income public schools in New York. My politics are not in question. So I am saying this from inside the family: we have used rigid moral positions to make vulnerable people feel unwelcome in our communities.

The way we talk about AI has real, painful effects on real people in struggling communities—people you would want to support if you knew their stories. I’m not saying any of us are bad. I’m saying we are better than this.

I am not asking you to love AI. I am not asking you to use it, or to drop your valid concerns about corporate accountability, labor policies, and environmental considerations. Those fights matter.

I am asking you to do something specific. Before you decide that the person using this tool is part of the problem, talk to them. Ask why. Ask what it does for them. Find out whether they’re the villain you’re picturing or the disabled author, the drowning bookstore owner, the grandmother’s granddaughter trying to make her books accessible. The same grace you would want extended to you.

With love, from inside the family,

Ella

PS. To be radically transparent, AI assisted in writing this letter. No, I didn’t just push a button—it took seven days of intense wrestling, organizing, and refining to get these complex ideas to line up. But on the days my brain wouldn’t cooperate and my executive function was offline, AI was the bridge. It didn’t do the thinking for me; it just helped me get my own voice onto the page. That’s what it actually does.