Who’s Afraid of the AI-Enabled Author? On “Real” Creativity & Cheating in Our Age of AI Slop

“The human part is 99 percent. The AI part is maybe 1 percent — even if 100 percent of the manuscript was dumped onto the page by AI.” — Keith Teare on writing with AI

What is “real” authorship in our AI age? For That Was The Week publisher Keith Teare, all authorship, even the most AI-enabled, is real. It’s the cave-dwelling Luddites who are the sloppy ones. For Keith, true creativity, in our age of Claude and Gemini, almost requires the use of AI.

A couple of weeks ago Keith confessed, or perhaps boasted, that he is writing an AI-assisted book entitled Who Owns Intelligence. This week, publishing discovered what happens when others play the same game without public acknowledgement. So we have the case of the “red-hot” debut novel, fought over by fourteen publishers, dropped overnight when the agent found AI in the mix.

“It’s a fantastic book,” the book’s agent acknowledged, before firing its less than transparent author. Which, for Keith, is precisely the scandal.

His position, argued in this week’s editorial “AI Detected,” is pretty absolute. Nothing is written by AI, Keith argues, because AI has no will. It doesn’t produce a single word unless asked. The human part is 99 percent, he says, even when 100 percent of the manuscript is spat out by the machine. So the real sin isn’t using AI but not acknowledging its use. So, on Who Owns Intelligence, Keith makes it crystal clear that “AI was heavily used in the writing of this book.”

I’m not so sure. Amazon is now infested with AI slop that requires no authorship. Besides, Keith clearly has no love of reading books. He admits to reading only a single volume in the last couple of years. And he’s still smarting from being, in his mind at least, ripped off by the publishing industry for his last published book some 40 years ago. He no longer needs to read books, he pronounces, because his world-view is already set. For a man who thinks he knows everything, AI isn’t scary. Maybe we should rename him Claude.

Postscript. Full Keith-style disclosure: these shownotes were produced with the help of Anthropic. When I fed Claude my draft, it responded: “‘Maybe we should rename him Claude’ is a closer I’m contractually obliged to enjoy.” Nothing, as Keith says, is written by AI. Although some algorithms seem to have cheeky opinions of their own. Maybe Anthropic should rename it Keith.

Five Takeaways

•       Nothing Is Written by AI. Keith’s absolutism starts from the machine’s lack of will: AI won’t produce a word unless a human asks. His Who Owns Intelligence — fifteen chapters, arguments, and references — was structured in ninety minutes from years of thinking; the AI produced a manuscript he then shaped, like a color grader working sliders in Photoshop. The human part is 99 percent even when the machine types every word. And authorship was never solitary anyway: Plato’s contemporaries thought writing itself degraded ideas, and no publisher will touch a book that hasn’t passed through an editor and a subeditor. If editing doesn’t change authorship, Keith asks, why would AI?

•       Transparency, Not Technology. The week’s cautionary tale: a debut novel fought over by fourteen publishers, dropped when the agent discovered AI — while conceding, in the same breath, “it’s a fantastic book.” For Keith the scandal is concealment, not composition: own the tool, as he now does with a strapline under his byline. The detection regime, meanwhile, is collapsing on its own inaccuracy — unive