When authors use AI somewhere along the lines in the process of producing a book, watch out! In some quarters of the author community, the pitchforks come out. So my co-founder, Authors A.I. CEO Alessandra Torre, and I recently pitched the subject of AI and ethics to longtime friend, podcast host and author Kevin Tumlinson. (Go check out his impressive author site.)
He’d touched on the subject in the past but thought there was still a lot of misunderstanding about the topic, and so he booked us on the topic of ethical use of AI by authors. The episode with our wide-ranging conversation just dropped on Kevin’s podcast Wordslinger – a must-listen-to podcast for those serious about the author journey.
I started out by pointing out the different phases of our own journey at Authors A.I.
- The first phase, after our launch in January 2020, might be called the era of AI Puzzlement & Early Adopters. We spent good deal of those early years in the wilderness explaining to puzzled authors how AI could possibly be of use when crafting a novel.
- The next phase, Fear and Loathing of AI, was spurred by the advent of ChatGPT — our conversation took place during the week ChatGPT turned 3 — with all the angst and hand-wringing about the prospect that AI could soon put authors out of a job.
- And now we’re entering the third phase, which I’ll call the AI Opportunity Era, where authors, editors and publishers are recognizing the value-add that AI brings to the table – as long as certain safeguards and standards are met.
Here are some highlights from our 35-minute discussion:
The flood of AI-generated content
Kevin shared eye-opening insights from his time at Draft2Digital, where the platform was receiving upwards of 1,500 books per day from people generating AI content. Much of it, he admitted, was “sludge, crap, garbage” – outlines being passed off as novels, low-quality content clearly not touched by human hands. This flood of AI-generated material has raised legitimate concerns across the publishing industry about quality control and the potential dilution of legitimate authors’ work.
Understanding what Marlowe actually does
Alessandra explained the main use cases for Marlowe Pro, which provides developmental feedback on manuscripts. Think of it as a super editor that delivers developmental-level feedback in about 15 minutes – analyzing everything from character development and point of view to plot arcs, unresolved plot threads, and pacing. Authors use Marlowe to guide their self-edits or to complement feedback from human editors.
A key distinction, Alessandra emphasized, should be made between analytical AI and generative AI. Marlowe is an analytical AI focused on pattern recognition and feedback – it can tell you that your manuscript is 32% dialogue and show you exactly where the manuscripts lags in dialogue, and it can display a detailed pacing graph comparing your work to bestsellers in your genre. “There are certain things that humans just can’t do,” she said. “It offers data analysis that a human can’t do.” Generative AI, by contrast, creates text for you – and that’s where many authors draw their ethical line. Marlowe doesn’t do generative.
A mixed reception for AI
I mentioned the mixed reactions from the author community that I encountered firsthand. When posting about our Cyber Monday special to 25 different Facebook author groups, about half the moderators approved the posts while half rejected them simply because “AI” appears in our company name. Some commenters dismissed us as “one of those AI slop companies” without understanding what we actually do.
This polarization extends beyond social media. There are sites that won’t let you participate in their community or enter contests if you’ve used AI in any capacity – even for editing or for a human designer who uses AI as part of her design process. What these gatekeepers often don’t realize is that they’re probably using AI-powered tools themselves without knowing it, whether through writing aids like Grammarly or ProWritingAid, sales reporting tools, or countless other applications.
The education gap
The conversation touched on the educational process needed to help authors understand AI’s role in their work. Some resistance comes from an educated perspective about AI’s limitations and risks, but much of it stems from misunderstanding. As I noted, “It’s partly an educational process. We have to sort of educate authors about where to draw that line.”
Kevin shared an interesting revelation: some editors he hires are already using AI tools to generate their editorial reports. The editing itself is done by humans, but the formatted reports may be AI-assisted. If authors are going to take an absolutist stance against AI, he quipped, they might need to stop hiring editors altogether.
The thorniest ethical questions
Where do the ethical lines get truly blurry? Alessandra pointed to AI translations and audiobooks. Many authors simply don’t have the budget for human narrators or translators, and they’re struggling just to break even. Does using AI for these purposes take work away from human professionals? It’s a genuine dilemma – especially when considering that AI translations flooding foreign markets can range from “really bad” to merely “okay,” and experts recommend having a human review AI translations regardless.
The consensus seemed to be that when AI is used significantly, it should be clearly identified – though Kevin noted he’s wary of disclosing more than necessary to platforms like Amazon, suspecting they’ll find ways to monetize that information.
Working with traditional publishers
I mentioned the stance of traditional publishers based on conversation we’ve had: all the Big Five have drawn a firm line against generative AI being used to write the actual text of books. They won’t work with companies that supply authors with AI-generated words for their chapters. However, they’re perfectly comfortable with assistive AI for tasks like creating chapter synopses or query letters. If you’re planning to submit to a literary agent, you’d better keep quiet about any generative AI involvement in your actual writing, even as using it for marketing copy and supporting materials is increasingly accepted.
One point we emphasized: Marlowe is trained on a rights-cleared corpus. We either obtained permission from authors or purchased the titles used for training – something we did years ago, before the current debates around AI training data erupted. This matters because Marlowe’s core technology is pattern recognition, identifying the characteristics that distinguish bestselling novels from weaker performers. It’s not an LLM; it’s a homegrown AI built organically with machine learning, combining the expertise of bestselling authors with data scientists.
The next few years
As Kevin observed, when we look back five years from now, the current anxieties about AI may seem quaint – much like early fears about electricity “vapors” invading homes. Technology has always displaced some jobs while creating others, and AI follows that pattern. The question isn’t whether to use AI, but how to use it thoughtfully and ethically.
The practical takeaway from our discussion: authors should focus on understanding the difference between assistive AI (which helps you write better) and generative AI (which writes for you). Use tools that enhance your craft rather than replace it. Be transparent when transparency matters. And remember that at the end of the day, readers still want authentic human stories – AI should be your collaborator, not your replacement.
The pitchforks may still be out in some corners of the author community, but we hope that more thoughtful conversations can help separate genuine concerns from knee-jerk reactions. The authors who thrive in the AI era will be those who understand these tools well enough to use them wisely.
Listen to the full episode on the Wordslinger podcast (or watch it on YouTube) – and stick around to the end for a special discount code on a Marlowe 3.0 subscription.
