BREAKING: What writers really think about AI

A large new study looks at writers' thoughts, and it's a doozy.

Author

Mignon Fogarty
November 05, 2025

Issue 94

On today’s quest:

— What writers really think about AI
— Beehiiv founder tested AI and found it lacking
— Author publishes book as a NotebookLM project
— WOTY watch
— I no longer trust scheduled tasks
— A new way AI could improve work

What writers think about AI

There's a brand new study from Gotham Ghostwriters and Josh Bernoff looking at how almost 1,500 writers and editors feel about AI and what they're actually using it for. It's fascinating information, and we have the very first look (podcast, YouTube)!

Big picture takeaways:

  • People are extremely divided, and they have strong (strong!) opinions.

  • The division is even stronger among fiction writers than nonfiction writers.

  • People with higher incomes are more likely to use AI.

  • Even writers who love AI have lots of concerns.

  • The most common AI tasks are suggesting titles, search, brainstorming, and finding words.

Top-line numbers for writing professionals:

  • 61% use AI tools.

  • 91% are worried about hallucinations.

  • 25% have considered quitting because of AI.

  • 75% of AI users say they are more productive (with an average productivity increase of 31%).

  • 43% of AI users said it improved the quality of their writing.

Beehiiv founder tested AI and found it lacking

Beehiiv founder Tyler Denk’s latest newsletter described the utter failure of AI customer support tools that the company was pushed to try by their VCs. He said they tested four different companies that built custom solutions that were fine-tuned on Beehiiv data, and “None of them could handle a fraction of the complexity of inquiries from our users, nor the simplest tickets.”

I found this surprising since customer support is an area where I see a lot of success stories, and I’ve also seen stories of these tools leading to higher customer satisfaction rates. I’m not sure what the difference is, but Tyler’s experience is an interesting counterbalance to all the “successes.” FWIW, he thinks VCs are so enthusiastic because they only see optimized demos.

I use Beehiiv to publish this newsletter, and I’m happy with the service. If you’re looking to start a newsletter or to switch from another provider (cough, Substack, cough), I recommend Beehiiv. Get a 30-day free trial + 20% off for 3 months with my affiliate link. Or sign up free to learn more at their big Winter Release Event.

Author publishes book as a NotebookLM project

George Kao has released the 3rd edition of his book “Joyful Productivity” only on NotebookLM, at least for now. It’s $10 or free with your email address and a promise to promote the book on social media.

He’s touting all the features of the platform — that not only can you read the chapters, you can listen to an audio overview, build a mind map, watch bonus videos, and ask questions about the book and get answers. I’m not sure if he’s the first author to exclusively release a book on NotebookLM, but I haven’t heard of anyone doing it before.

He’s likely getting extra publicity from making this unusual move, but I can’t help but think he’s also severely limiting his market by not releasing it in any other way. Maybe he plans to release it more widely later. (via Crystal Wood)

WOTY watch

I expect to see a lot of AI-related words in word-of-the-year lists this year. So far, we have the following from Dictionary.com:

  • Clanker

  • Agentic

I no longer trust scheduled tasks

I was excited about scheduled tasks from ChatGPT. I was using it to get a weekly update of new language books that are set to be published and to get updates of blog posts and articles with people describing how they use AI. In both cases, I was getting better results than I got from Google Alerts or Google searches.

Unfortunately, after a few weeks, both schedulers failed. I kept getting emails saying there were no new books and no new articles, and I knew that couldn’t be true. I put the same prompt in that it was supposed to be using and got lots of good responses, so I no longer trust scheduled tasks. I’m back to submitting the prompts manually whenever I remember.

A new way AI could improve work

I like Simon Willison’s thinking about AI, but his practical posts are often about coding, which doesn’t apply to me. But he had an observation yesterday that is making me wonder if there are parallels in the writing, marketing, or teaching world:

Simon said the code he writes with LLMs is higher quality code than the code he wrote before because the cost of making small improvements has gone to almost zero. Before LLMs, he would often skip small improvements because they weren’t worth the time. But now he includes them all.

I’m still thinking this through, and I’m wondering what you think. Could this apply to non-coding work?

Could non-coding work benefit from ease of making all possible improvements?

Login or Subscribe to participate in polls.

Quick Hits

Climate & Energy

The Future of AI and Energy AI’s rise is happening when we’re also at the dawn of the Solar Age — The Weekly Anthropocene

Google pulls Gemma from AI Studio after Senator Blackburn accuses model of defamation [An interesting side thread is the explanation that lightweight models aren’t meant to answer factual questions.] — TechCrunch

AI firm wins high court ruling after photo agency’s copyright claim. Ruling in case brought by Getty Images against Stability AI is seen as a blow to copyright owners. — The Guardian

Robotics

Inside Japan’s long experiment in automating eldercare [So far, robots actually create more work for caregivers.] — MIT Technology Review

Job market

Education

The business of AI

Is ChatGPT Maker OpenAI Becoming Too Big to Fail? — Wall Street Journal

Podcasting

“Nark” podcast uses AI to recreate the voice of a dead man (with permission from his estate) — Podcast Business Journal

Other

DeepMind's AI Learns To Create Original Chess Puzzles [This seems like big news in the chess world.] — Chess.com

What is AI Sidequest?

Are you interested in the intersection of AI with language, writing, and culture? With maybe a little consumer business thrown in? Then you’re in the right place!

I’m Mignon Fogarty: I’ve been writing about language for almost 20 years and was the chair of media entrepreneurship in the School of Journalism at the University of Nevada, Reno. I became interested in AI back in 2022 when articles about large language models started flooding my Google alerts. AI Sidequest is where I write about stories I find interesting. I hope you find them interesting too.

If you loved the newsletter, share your favorite part on social media and tag me so I can engage! [LinkedIn — Facebook — Mastodon]

Written by a human