“Do I Have To Disclose I Used AI In My Query Letter?”

And other things about which I’m vehement.

Last night at an online event, a participant asked me, “Do I have to disclose I used AI in my book’s creation in my query letter?” I warned I was coming in hot before I answered (cussing alert) and responded, “If you are using AI to write your book, stay out of my fucking inbox.” I explained that I don’t think AI writing is “writing” at all—it’s created without thought, authorship, creativity, or intention, the very things that make writing valuable.¹ Many folks virtually applauded—most aspiring professional writers can get behind not outsourcing their writing to AI. I felt bad for cussing, but not too bad. Plus, I had the crowd with me.

Then came the follow-up, “But what about using it for research?” That’s the question I want to address here because the answer is perhaps less intuitive. You can go ahead and disclose that you used AI in the “research” for your book; it’ll tell me I have one less query to respond to as I don’t consider AI research to be research. Moreover, I consider it a race to obsolescence for nonfiction writers.

This is in part because I know AI research isn’t very good—it’s superficial at best, full of mistakes and hallucinations. It devalues something I deeply respect: the advent of AI has made the important work of real research all the harder. I constantly hear of incorrect AI research making its way into the student work seen by my academic clients and friends, and I worry about the rising generation of students thinking an AI search is research. I’ve also encountered its sloppiness personally. Next week, I am making the book tour appearance I am most excited about: I’m doing a drop-in visit to my partner’s “Nineteenth-Century American Fiction” class to discuss publishing careers and also, the ending of House of Mirth. (!!!) We timed my visit so I could discuss one of my all-time favorite books, one I first read when I was exactly these students’ age. To prep for the class, I wanted to refresh myself on the latest academic discourse on the novel. Let’s look at the AI results for “HOUSE OF MIRTH literary criticism.”

I wouldn’t settle for this as “research” (Cliffnotes links! Youtube!) in preparation for an (unpaid, informal) visit to a group of undergrads whom I hope to persuade that HOUSE OF MIRTH, publishing careers, and books matter, let alone for the beginnings of a writing project one hopes will interest other people. Note too that almost none of the suggested links are to actual books. Imagine you are a reader encountering Edith Wharton or literary criticism for the first time (lucky you!). Would these results lead you to believe that many people have written whole books on Wharton and have devoted years of their lives studying her? That such a devotion is worthwhile, or even possible? That you too could become part of such a community of readers and writers?

And that is more my salient point. I worry that in asking the question about using AI to research a book, a writer isn’t thinking too much about where the research results in AI searches come from, which, of course, is exactly what the billionaires who profit from AI want you not to think about. AI results are “scraped” (such an apt word) from the work of writers, writers you are choosing not to read and whose books you are not supporting, by using it.

Most professional writers don’t feel great about this! Per the Authors’ Guild, there are now nearly a dozen class action lawsuits by authors against AI companies “for copyright infringement based on the companies’ unauthorized copying of authors’ works to train their generative AI models.” In other words, AI companies mine authors’ work without their permission and without payment to make Frankensteinian texts that use the bits and pieces of that work to make a CliffNotes version of their writing and research that you, as an aspiring author, can then consume without engaging with other authors. By using AI to do research instead of reading actual books, you are saying the very thing you are trying to sell doesn’t have value to you when it is made by other people. Perhaps unwittingly, you are also accelerating the process of devaluing writing, research, and human expertise. Part of this is simply an economic question: how will we fund writing, reporting and research if the primary way we encounter it is as info scraped and spat out by Chat GPT? How do you expect me to advocate for you as a writer (who should be read and get paid) if you are not reading and paying other writers? This approach to what writing is is a race to the bottom, in which you create conditions for your own obsolescence, and to borrow from “Succession,” in which writing has No Real Writer Involved.

This is at the heart of why I am so dismissive of AI “research”---because of what it says about how you, as an aspiring writer, see yourself in relationship to books and other authors. Great writers become great writers by engaging deeply with the work of other writers, and ultimately, imagining themselves in community with them. Let me give you an example from my own files. My client Clint Smith is a poet who is also trained as a sociologist of education. HOW THE WORD IS PASSED: A RECKONING OF HISTORY OF SLAVERY ACROSS AMERICA, his first book of nonfiction, is in many ways a work of history, albeit poetically written. Knowing he was not a trained historian, Clint approached the discipline and the profession of history with humility and respect while writing and researching HTWIP. He read many relevant works of history. He also reached out to historians to ask if they would be open to chatting with him or even reviewing chapters of his book. All his hard work paid off. HOW THE WORD IS PASSED was one of the most acclaimed works of nonfiction in the year it was published. (Literally: Publishers’ Marketplace compiles a list of such accolades and Clint’s book was at the top.) It won the National Book Critics’ Circle Award for Nonfiction, was longlisted for the National Book Award, named one of the New York Times 10 Best Books of 2021, and was a #1 New York Times’ Bestseller. When you see Clint speak about the book, you can immediately tell you are in the hands of someone who has mastered his material. He knows it in his bones. Do you think he’d have such mastery of his subject, and have written such a book, if his research were rooted in AI?

That mastery is also important, and I am going to head off another “what about” question I intuit may come up in the comments: “What about using AI to help you with your outline? Your structure?” To that I say part of the work of writing is figuring out your own book’s structure and arc. You will never gain mastery of it by outsourcing this labor. (Again, I urge you to read John Warner’s book on this. Outlining IS writing.) I’ve now been on the road for nearly a month speaking about my book. TAKE IT FROM ME was not easy to write. I do find it quite easy—and fun!—to talk about. Part of this is because I really like public speaking, but part of it is because I did the hard, years-long work of figuring out my book’s structure and how to synthesize my own research, knowledge and experience. I couldn’t have known this without going on book tour, but one of the reasons I have enjoyed touring so much is I am able to appreciate all I have come to know. While I can’t promise you an agent or a book deal or publishing success, I can promise you the pleasures of mastery—one of the few parts of the writing and publishing process you can control. It doesn’t come from memorizing “AI OVERVIEW RESULTS.”

Now about that hat. One of the joys of working at the Gernert Company is my boss’s love of personalized swag. A few years back, a minor publishing controversy resulted in many agents calling a particular publisher to give their entirely unsolicited opinions. I was one of those agents, and I held forth Julia Sugarbaker-style with my thoughts. After I finished, I asked the publisher, “Have other folks called you about this?” The publisher responded, “They have. But you were by far the most vehement.” I told this story at a company meeting, and a few days later, a package arrived with this hat. I had no doubts who sent it.

I share this story because I do feel a bit bad about getting heated during Q&A; I never want someone to come to an event of mine in fear I will humiliate them. And in this case, my vehemence wasn’t directed at the person who asked the question, but at the ecosystem that is trying so hard to normalize AI. Billionaires have a lot invested in making us feel that AI is normal, helpful, smarter than us, and inevitable, with the goal of increasing our reliance on these products until we can’t imagine life without them—no matter how much they cost us in the long run, and those costs are high indeed. AI is marketed as a silver bullet, the only thing you’ll need. But I’m proposing something different here: we don’t need AI, we need each other. We need people thinking critically in conversation with one another, reading, researching, writing. And if want to write a book, consider how that task inherently connects you to other people.

So next time you Google something, scroll past the AI overview and read an article or two. (Here’s a great one we found on Wharton that wasn’t in that AI summary. Thank you, honey, for sending it to me.) Next time you’re not sure if your email has the right tone, get a friend or colleague to give you feedback. (Thank you, Gabe Sherman, for your help with hitting the landing in this post.) Don’t rely on AI search results to find the “best agents.” Get to know us by spending real time researching our books and lists. And learn to pitch your own work better by writing your own query letters, and revising them, and sharing them with valued friends before you send them out. I can’t promise I’ll respond, but I can promise that is time well-spent.

***

Like this post or this Substack? You’ll love my book and buying it is the best way to support me. You can order it here.

1. Thank you to John Warner for this framing. His book MORE THAN WORDS is the best thing I’ve read on large language model AI, what they can and can’t do, and what makes the intellectual and creative work of writing a value in itself to its creator and its recipient, one that cannot be outsourced.