Link: https://www.wired.com/story/book-publishers-are-quietly-using-more-ai-staff-are-revolting/

According to interviews with more than two dozen workers, all of whom spoke under the condition of anonymity to avoid jeopardizing their employment, publishers are now using LLMs like Claude and ChatGPT to email literary agents, write publicity and back-cover copy, design cover art, and generate marketing materials, often at the direction of their executives.

At the same time, multiple editorial staffers say the widely publicized cases of authors accused of using LLMs to help write their books—like Hachette’s horror novel Shy Girl or Macmillan’s crime thriller Call Me, I’ll Hide the Body—are just the tip of the iceberg. “Every single imprint has a story of someone very esteemed who had a book canceled because of blatant AI use” without ever making the news, one editor says.

At HarperCollins senior leaders purchased Claude licenses from Anthropic months ago “but very quickly realized they didn’t know how to implement them,” according to a current employee. Dozens of junior- and senior-level staff were then “voluntold” to become “AI Champions” and were tasked with discovering use cases in monthly brainstorming sessions with staff across the company.

I'm so tired of every damn company in every damn sector racing to the bottom in order to save a buck with AI...

Minimizing cost to maximize profit is the essence of capitalism. What do you want instead?

Maybe independent audit to certify if, where and how a publisher used AI/LLM in their business processes? With the auditor publishing their findings publicly?

Should it be a legislated requirement for all publishers? Or a personal choice for the purchaser?

What other solutions are there?

They're also using AI to perform grammar checks. They're not firing editors, but why waste an editor's time having them fix grammar when an AI model will spot all the mistakes in seconds? It'll also spot the mistakes editors miss.

There's an old joke in publishing: The day you get the first (final) copy of your book is the same day you find a typo. With AI, that's not going to happen, so they can't not use it at this point.

Maybe a publisher that recognizes that good quality will attract more customers than pushing out cheap crap.

But I'm not trying to come up with solutions or even defending capitalism. I'm just sad everything is turning to shit in the blink of an eye.

The issue that I have with this is that, although I agree with the spell/grammar check with context use, they are using cloud models to do this. They are essentially giving a third-party a legal right to use an author's work to feed their model. Work like this is simple enough for a self-hosted offline LLM to do.

Hah! The first thing I thought when I read that they appointed "AI champions" was, "those people should learn how to use ollama with local models."

Though using the incoming book text for training probably isn't a big deal at this point because these companies are more concerned with the non-context part of the prompt (i.e. stuff other than the book text) and its output instead of the context itself.

Training their LLMs with incomplete/unedited works is generally not best practices. That's basically how you poison your model.

is there a list of publishers who have disavowed ai use?