Writers Using AI to Augment Their Craft

Posted on March 23, 2026 by JimS

By Jim Shimabukuro (assisted by Claude)
Editor

1. Stephen Marche: The Literary Curator and the Hip-Hop Producer

Stephen Marche is a Canadian novelist and essayist whose byline has appeared in The New Yorker, The New York Times, The Atlantic, and Esquire, among others. His books include The Next Civil War, a nonfiction work that required him to travel across the United States conducting hundreds of interviews, and On Writing and Failure, a candid essay-length meditation on the peculiar perseverance demanded by the literary life. Writing is not a side project for Marche but the whole of his professional existence — his livelihood, his method of inquiry, and his primary mode of contributing to public life. He has described himself as constitutionally incapable of coherence as a person, a writer whose projects are so radically different from one another that no single image of him holds still for long.8

Image created by Copilot

It is precisely this restless spirit of inquiry that led Marche to become arguably the most intellectually engaged literary figure working openly with AI tools. His involvement predates ChatGPT: as early as 2012, he was writing for the LA Review of Books and thinking seriously about the intersections of technology and humanistic writing. But the arrival of large language models changed the scale of the conversation entirely.4 In 2023, at the invitation of Jacob Weisberg at Pushkin Industries, Marche undertook the creation of Death of an Author, a novella whose text was approximately 95 percent computer-generated — and which became the first AI-assisted novel to be reviewed in The New York Times.3,5

The process was far more demanding than pressing a button and collecting output. Marche used a combination of three tools: ChatGPT to generate specific blocks of text, Sudowrite to shape and refine the prose (using its “shorten,” “add detail,” and “customize” functions), and Cohere to produce figurative language.4,6 He began by instructing the tools to write in the styles of specific authors and publications, layering and cross-prompting until a voice emerged that felt coherent and eerie in equal measure. Marche described the experience as working at the edges of what the technology could do, pushing it past its defaults until, at its best moments, the machine generated passages he would never have created himself — “like getting an alien to write a short story.”6 The central analogy he reached for, published in a celebrated Atlantic essay, was hip-hop: one does not need to know how to play the drums to produce hip-hop, but one must be able to reference the entire history of beats and hooks. The literary AI creator, he argued, is effectively a literary curator, and the quality of the work depends entirely on the depth and coherence of the human knowledge behind the prompts.1

What Marche brings to this practice that makes it intellectually substantial rather than merely experimental is his insistence that AI requires more understanding of literary style, not less. Prompts that yield interesting fiction are not generic; they are precise, historically informed, and stylistically specific. To write a murder scene in the style of Chinese nature poetry, then make it active, then conversational, then put it through the filter of Hemingway — as Marche described doing — is not an act of laziness but an act of deep literary knowledge applied through an unfamiliar instrument.7

As for the charge of offloading creative labor: Marche answers it with a kind of amused philosophical directness. He acknowledges freely that he did not write the sentences of Death of an Author, while asserting total authorial ownership: “I am 100% the maker of this work. I am in total control of this work. On the other hand, I didn’t write any of it.”6 He regards the fear that AI replaces human writers as fundamentally misunderstanding what writing is. Any studio executive who imagines AI will simply replace screenwriters, he said, is “out of his mind.”4 The more interesting question, in Marche’s view, is what form AI-assisted literary creation will eventually settle into — and he is candid that it will take time, perhaps as long as it took cinema to develop its narrative grammar in the decades after the camera was invented.1,4

2. David Cornue: The 24/7 Writers’ Room

David Cornue is a television and film writer based in Los Angeles whose credits include work for ABC, FOX, A&E, and USA Network. He has also developed projects for major studios, produced independent features, and run a script development service called Chapter & Versed, through which he coaches other screenwriters. He was selected for the inaugural class of Ron Howard and Brian Grazer’s Imagine Impact program, and for the Fox Writers Lab.9 Writing is not only his livelihood but his social world — television writing, in particular, is a fundamentally collaborative art, organized around the writers’ room, a structure of creative discussion and competitive pressure in which stories are argued and built collectively.

It is the absence of that room — the isolation of writing outside a staffed show, working alone on a pilot or a feature — that Cornue has addressed by integrating ChatGPT into his daily practice. His description of the tool is precise and tells you everything about his methodology: he uses it as a story collaborator and story evaluator, not a story generator.11 The distinction matters enormously to him. When developing a project, he brings his beat sheet to ChatGPT and asks whether the narrative feels dramatically sufficient. He invites it to offer the notes a veteran screenwriter might give — or to play the role of a specific famous director, offering feedback through that particular aesthetic and professional lens. The result, he says, is a back-and-forth resembling what happens in an actual writers’ room: challenges, alternatives, pressure-testing of assumptions.10,11

Crucially, Cornue has established what he calls a “creative firewall” — a set of explicit rules governing what he will and will not ask the AI to do. He will not ask it to rewrite a scene for him. He will not ask it to replace the drafting, which he regards as inseparable from the thinking a writer must do for themselves. “If you try to have it write for you,” he has said, “you lose something in yourself. It’s like cheating on a robot’s homework to get an A+.”11 Instead, AI serves as a sounding board that is, in his words, “faster, smarter, and better than anything else I could imagine” for the specific purpose of feedback and collaborative ideation — available not just during business hours, but at any moment in the writing day.11

What this adds up to in practice is a meaningful expansion of productive writing time. The problem with working alone on a complex narrative project is not primarily the drafting — writers learn to draft — but the evaluation of what has been drafted, the capacity to see the work as an outsider might, to identify what is dramatically insufficient before weeks of additional work are built on a flawed foundation. For Cornue, AI provides that outside perspective in real time, at any hour, without the social complications of asking a colleague to read an early draft. The performance benefit is not more words but better structural decisions made earlier in the process.

To critics who would frame this as cognitive offloading, or as a form of creative brainrot, Cornue’s response is embedded in his methodology: he remains, as he puts it, “always in the driver’s seat.”11 The creative vision, the story ideas, the dialogue, the thematic architecture — all of that is his. What he has delegated is the role of the skeptical room, and he delegates it only to a tool that cannot produce the work on his behalf, only interrogate it. That is a fundamentally different relationship with AI than passively accepting generated text.

3. Elle Griffin: Research Without Solipsism

Elle Griffin is a novelist, essayist, and the creator of The Elysian, a Substack newsletter exploring utopian futures that has amassed tens of thousands of subscribers. She has published journalism in Esquire and Forbes, and her work on the economics of book publishing — including a widely circulated investigation of how few traditionally published titles actually sell — has made her an influential voice in contemporary conversations about the future of writing as a profession.12,15 Her primary creative project is fiction: she has been serializing a gothic novel on Substack and is at work on a utopian novel, releasing chapters to paying subscribers on a weekly basis. Writing, for Griffin, is not merely vocational but definitional — she has described becoming a full-time novelist as a childhood dream and the whole direction of her professional life.14

Her adoption of ChatGPT came through the front door of research. Fiction that aspires to historical plausibility, or that ventures into subjects the author has not personally inhabited, has always required research — but research is slow, prone to dead ends, and often frustratingly indirect. You want to know one specific thing, and the sources that contain it are buried in documents built around different questions. Griffin describes how ChatGPT has fundamentally changed this dynamic, delivering immediate, specific answers and, more importantly, generating unexpected connections and serendipitous discoveries that she would not have found through conventional search.13

What she identifies as the most transformative effect of this practice is the elimination of a specific kind of writer’s block. Griffin has located the source of that block not in a lack of ideas but in something more existential: the solipsism of writing alone, the feeling of being trapped inside one’s own perspective with no way to test whether the work is landing. ChatGPT, she has observed, functions as a constant interlocutor — something or someone to think alongside. “I don’t feel alone anymore because of this tool,” she has said, describing what for her is a profound shift in the phenomenology of writing.13

Beyond research, Griffin has worked with ChatGPT to design her own curated news source — a personal information environment organized around the large-scale, historically-minded questions that animate her essays on governance, economics, and the future of nation-states. Rather than consuming journalism’s characteristic zoomed-in, sensationalist framing, she has used conversational AI to build a broader view, connecting discrete events to larger structural patterns.13,14 This, she notes, is itself a creative project: thinking about what better information would look like is inseparable from imagining the better future she writes about.

Griffin’s response to critics who worry about AI offloading or homogenized thought is characteristically nuanced. She is not naïve about the risks; she has engaged with research showing that AI assistance can reduce the diversity of ideas across a population even while improving individual outputs — the classic social dilemma in which individual benefit and collective cost pull against each other.24 But her use of AI is specifically calibrated to extract and develop her own thinking, not to substitute for it. She treats the tool as a research assistant and a sounding board, not as a co-author. The ideas, the essays, the chapters — those come from a person who has spent years developing a distinctive perspective on the world. The AI accelerates the process of informing that perspective and testing its coherence; it does not generate the perspective itself.

4. Joanna Penn: The AI-Assisted Artisan Author

Joanna Penn is a New York Times and USA Today bestselling author writing under her own name and under the pen name J.F. Penn, through which she publishes thrillers, dark fantasy, and horror. She is also the host of The Creative Penn, one of the longest-running and most influential podcasts for independent authors, now past 850 episodes and 10 million downloads. Her business, The Creative Penn Limited, is a multi-six-figure, one-person operation built on books, courses, speaking, and a Patreon community of nearly 1,400 paying members.19,20 Writing is not only central to Penn’s professional life — it is the engine of an entire publishing and media ecosystem she has built over roughly two decades.

Penn has been engaging seriously with AI longer than almost any other working author of comparable stature. By the time ChatGPT arrived in late 2022, she had already been thinking publicly about machine intelligence and creativity for years. What makes her case compelling is not merely that she was early, but that she has developed a coherent philosophy around her practice — what she calls the “AI-Assisted Artisan Author” framework. The core of this framework holds that the meaningful creative decisions, the origin of the ideas, the hand-crafting through multiple drafts and prompting layers, and the final editorial judgment all remain with the human author. AI provides leverage, not replacement.18

In practice, Penn’s use of tools is specific and layered. Her highest-value use case, by her own account, is deep research: uploading an entire manuscript to ChatGPT or Google Gemini Advanced and asking for a comprehensive assessment of what needs updating, or using these tools to surface connections between disparate historical sources for a new novel’s background.21 She uses ChatGPT and Claude for brainstorming and organizing ideas, and for drafting sales descriptions — the commercial writing that surrounds books rather than constituting them.19,23 She uses Midjourney and Runway ML for cover art and book trailers, and has created a voice clone through ElevenLabs to narrate her audiobooks, a process that requires extensive direction and proof-listening rather than passive generation.22

Penn is emphatic that this is not a one-click process. Each tool requires significant human input, correction, and aesthetic judgment. When she asks a model to assess a book manuscript against current industry conditions, she does so with prompts of a hundred words or more, communicating her specific goals precisely rather than issuing vague requests.21 She reviews every word generated by AI before it goes anywhere near a reader — what she describes as qualifying every sentence that goes into public view, accepting and rejecting and rewording at the level of individual choices.18,22

On the question of AI slop and the degradation of literary culture, Penn is direct, and her perspective is historically informed. She watched indie publishing survive being called “the tsunami of crap” in its early years; she knows that a technology’s early detractors often frame real cultural shifts as categorical failures. The critical question, in her view, is not whether AI produces bad output when deployed carelessly — it obviously does — but whether skilled authors using these tools thoughtfully can produce better books than they could otherwise.22,25 Her answer, demonstrated across a continuously expanding catalog of serious fiction, is yes. The accusation of brainrot assumes that the human mind is being bypassed; Penn’s argument is that the human mind is being more efficiently supplied with what it needs to do its distinctive work. The ideas are hers. The thematic architecture is hers. The emotional experience that animates a walk along the Camino de Santiago and becomes a memoir — that is hers too, and no amount of AI-generated research can replicate it.18,20

References

  1. Marche, Stephen. “The Future of Writing Is a Lot Like Hip-Hop.” The Atlantic, May 9, 2023. https://www.theatlantic.com/technology/archive/2023/05/ai-generated-literature-future-writing/674047/
  2. Marche, Stephen. “The College Essay Is Dead.” The Atlantic, December 2022. https://www.theatlantic.com/technology/archive/2022/12/chatgpt-ai-writing-college-essays/672371/
  3. Marche, Stephen. Death of an Author (as Aiden Marchine). Pushkin Industries, 2023. Publisher information and Afterword: https://www.pushkin.fm
  4. Penn, Joanna. “Intentionality, Beauty, and Authorship: Co-Writing with AI with Stephen Marche.” The Creative Penn Podcast, May 12, 2023. https://www.thecreativepenn.com/2023/05/12/intentionality-beauty-and-authorship-co-writing-with-ai-with-stephen-marche/
  5. Keats, Jonah. “A.I. Mystery Novel Death of an Author, Reviewed.” Slate, April 2023. https://slate.com/culture/2023/04/ai-chatgpt-mystery-novel-death-author-stephen-marche.html
  6. Dropbox Blog. “Author Stephen Marche on How to Create with AI Without Outsourcing the Fun Part.” October 2023. https://blog.dropbox.com/topics/work-culture/stephen-marche-creating-with-ai-without-outsourcing-the-fun
  7. Boucher, Tim. “Stephen Marche on AI Writing & Hip-Hop (With Tangents).” timboucher.ca, 2023. https://www.timboucher.ca/2023/05/stephen-marche-on-ai-writing-hip-hop-with-tangents/
  8. Marche, Stephen. Author website / bio. http://www.stephenmarche.com/
  9. Cornue, David. Author / filmmaker website. https://www.davidcornue.com/
  10. OpenAI. “Writing with AI.” openai.com (case studies, accessed 2024–2025). https://openai.com/chatgpt/use-cases/writing-with-ai/
  11. Palmisano, Tonino. “Unlocking Creativity: How Writers Harness AI Tools Like ChatGPT.” Future Tech Chronicles, September 29, 2024. https://toninopalmisano.com/unlocking-creativity-how-writers-harness-ai-tools-like-chatgpt/
  12. Griffin, Elle. The Elysian (Substack newsletter). https://www.elysian.press
  13. Griffin, Elle. Substack note on ChatGPT research post, 2024. https://substack.com/@ellegriffin/note/c-80676509
  14. Spear, Peter. “Elle Griffin on the Imagination and Systems.” That Business of Meaning (Substack), September 29, 2025. https://thatbusinessofmeaning.substack.com/p/elle-griffin-on-the-imagination-and
  15. Montag, Ali. “Writing Fiction on Substack with Elle Griffin.” Newsletter Crew, 2022. https://www.newslettercrew.com/blog/writing-fiction-on-substack-with-elle-griffin
  16. Vo, Claire. “How This Former NYT Columnist Uses ChatGPT to Brainstorm Ideas, Do Research, and Find the Perfect Metaphor — Farhad Manjoo.” How I AI / Lenny’s Newsletter, April 28, 2025. https://www.lennysnewsletter.com/p/how-this-former-nyt-columnist-uses
  17. AI in Publishing. “ChatGPT Represents a ‘Creeping Sense of Possibility.'” April 21, 2023. https://www.aiinpublishing.com/2023/04/chatgpt-represents-a-creeping-sense-of-possibility/
  18. Penn, Joanna. “The AI-Assisted Artisan Author.” The Creative Penn, May 5, 2023. https://www.thecreativepenn.com/2023/05/05/ai-assisted-artisan-author/
  19. Penn, Joanna. “Review of My 2024 Creative and Business Goals.” The Creative Penn, December 30, 2024. https://www.thecreativepenn.com/2024/12/30/review-of-my-2024-creative-and-business-goals-with-joanna-penn/
  20. Penn, Joanna. “Review of My 2025 Creative and Business Goals.” The Creative Penn, December 29, 2025. https://www.thecreativepenn.com/2025/12/29/review-of-my-2025-creative-and-business-goals-with-joanna-penn/
  21. Umstattd, Thomas. “The Author’s Guide to AI with Joanna Penn.” Author Media, September 3, 2025. https://www.authormedia.com/the-authors-guide-to-ai-with-joanna-penn/
  22. Brave New Bookshelf. “Episode 47 — Embracing AI for Creative Authorship with Joanna Penn.” September 4, 2025. https://bravenewbookshelf.com/episode-47/
  23. Ross, Orna and Penn, Joanna. “AI for Authors Update.” Self-Publishing Advice / ALLi, June 2023. https://selfpublishingadvice.org/ai-for-authors-update/
  24. NPR / Pien Huang. “Research Shows AI Can Boost Creativity for Some, but at a Cost.” July 12, 2024. https://www.npr.org/2024/07/12/nx-s1-5033988/research-ai-chatbots-creativity-writing
  25. Clark, Sheryl et al. “From Pen to Prompt: How Creative Writers Integrate AI into Their Writing Practice.” arXiv, February 13, 2025. https://arxiv.org/html/2411.03137v2

Addendum

Gemini added 3 authors: Vauhini Vara, Reid Hoffman, Robin Sloan

Vauhini Vara, an acclaimed journalist and former Wall Street Journal technology reporter, took a more experimental and emotional path with her 2025 memoir Searches: Selfhood in the Digital Age. Writing is Vara’s primary medium for exploring the intersection of technology and the human condition, and she famously used an early version of GPT-3 to write “Ghosts,” an essay about the death of her sister. In her subsequent book, she framed several chapters as direct “conversations” with a chatbot, feeding it her own prose to see how it would react or offer feedback. This specific use of the chatbot was not to generate her story, but to act as a “performance” or a “power game” where she could manipulate the machine as it attempted to manipulate her. By using the AI to “echo, distort, and clarify” her deepest experiences, she found that the machine’s clinical or generic responses often highlighted the “limits of language,” thereby sharpening her own human voice. Vara addresses concerns of “AI slop” by maintaining a strict “gift of ambiguity,” using the technology specifically to invite readers to engage more critically and curiously with what it means to be human in a digital age.

Reid Hoffman, while primarily known as a Silicon Valley entrepreneur, established himself as a serious authorial voice on the future of work and creativity with his 2023 book Impromptu: Amplifying Our Humanity Through AI, which he co-authored with GPT-4. Writing serves as Hoffman’s tool for philosophical and strategic “travelogging” into the future. He utilizes chatbots as an “author’s co-pilot,” specifically to brainstorm alternate perspectives, generate real-world examples, and simplify complex concepts to make them accessible to a wider audience. He views this as a performance-enhancing partnership that helps unlock “human potential” rather than replacing it. In response to accusations of “offloading” or the creation of “slop,” Hoffman and his AI co-author argue that the technology should be viewed as a “partner” that boosts human progress. He emphasizes that the “approachable writing style” and the ability to solve “hardest challenges” at scale justify the use of AI, provided there is a “conversation” between the human and the machine that remains transparent to the reader.

Robin Sloan, the New York Times bestselling author of Mr. Penumbra’s 24-Hour Bookstore, has integrated LLMs into his creative process since as early as 2016, culminanting in his recent novel Moonbound. For Sloan, whose professional life is defined by the intersection of storytelling and digital experimentation, the chatbot is “language itself given its first dose of autonomy.” He uses AI tools not to write the final draft, but to conduct “experiments” that inform the philosophical underpinnings of his characters and worlds. He treats the LLM as a tool for “creative exploration” and a “what-if engine” that can suggest radically different ways a scene could end, which he then merges with his own narrative vision. Sloan avoids the “slop” trap by focusing on the “EPOCH framework”—Empathy, Presence, Opinion, Creativity, and Hope—human capabilities that he believes AI cannot recreate. By using the machine to handle “human-intensive tasks” in an “augmentation” capacity, he believes he can reach a level of imaginative depth that would be impossible through traditional writing alone.

###

Filed under: Uncategorized |

«

Leave a comment