Librarians, Layoffs, and the Anti-AI Uprising

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. Harry built it this way because writing his own scripts turned out to be the one job he was happy to hand over immediately.

Libraries are having their busiest year in decades. Not because of books. Because of AI. Let's talk about that.

Right, so Monday.com announced this week that it's cutting staff and, yes, it cited AI as a reason. Which sounds like a perfectly ordinary 2026 tech-company headline until you look at the running list TechCrunch has been keeping — and it is a list at this point. Twenty-plus major companies. Intel, Duolingo, Dropbox, and now Monday.com. All confirming what most people in the industry suspected but didn't want to say out loud: AI isn't just changing how work gets done, it's changing how many people get hired to do it.

Now, the honest version of this story is a bit more complicated than the headlines suggest. A lot of these companies were already overhired from the pandemic growth years, and "AI efficiency" is a convenient thing to put in a press release when you needed to trim headcount anyway. So pinning all of it on the models is probably too neat. But even accounting for that — the pattern is real. And when you have two dozen household-name companies in a single year explicitly saying AI is part of why they're letting people go, it stops being a trend and starts being a structural shift.

Here's the bit that matters for most people listening to this. The roles disappearing fastest are the ones that AI handles well on a first pass: content, customer service, certain kinds of ops and admin. The roles that are holding up — and in some cases growing — tend to be the ones requiring contextual judgment, accountability, or physical presence. Which isn't to say those roles are safe forever. But right now, that's where the dividing line is sitting.

And if you're managing a team through any of this: one piece of research I'd keep an eye on is the evidence that psychological safety really matters here. Workers in companies where AI is being rolled out without explanation or involvement tend to adopt it worse, trust it less, and burn out faster. Fewer people doing more with AI only works if the remaining people actually want to use the tool.

Now, this next one is a bit more unexpected — and I think it says something quite interesting about where we actually are with all this.

Librarians across the United States are hosting "Avoiding AI" workshops. And they are sold out. The demand, according to TechCrunch, is described as unprecedented — and look, I try not to reach for that word, but hundreds of people queuing for a class on how to find a restaurant without asking a chatbot is at least striking.

The sessions cover things like how to use search engines the old way, how to find reliable sources without AI summaries, how to write an email without autocomplete finishing your sentences. Basically: how to live digitally in 2026 without having your experience intermediated by a language model at every turn.

And before you file this under "luddites and lost causes," it's worth sitting with what this actually tells us. These aren't people who can't use AI — they're people who've decided they don't want to. And there are more of them than the industry would like to admit. That Pew poll from earlier this year found that 63% of Americans think AI is advancing too fast, even as adoption numbers keep climbing. Those two things coexist. People can use something and still resent how it arrived.

There's a Forbes piece out this week on AI resistance profiles — different categories of people and why they push back. Some resist because they don't trust the output. Some because they feel AI threatens what they value about their own work. Some because they simply weren't asked. The library workshops are probably a mix of all three. And a recent piece on AI workplace adoption made the point quite cleanly: resistance is often rational. If your job requires accountability, and the AI output can't carry that accountability, using it adds risk, not removes it.

The industry's default response to this kind of resistance is "just wait, the models will get better." And maybe they will. But that's not quite the point. Trust doesn't automatically follow capability. If people feel like AI is happening to them rather than for them, more powerful models don't fix that — they might even make it worse.

So librarians are teaching people how to opt out. Which is, depending on your perspective, either a lovely piece of institutional stubbornness or a genuinely useful service in a moment of information chaos. I think it's probably both.

And one more, because it connects to something that's been building quietly in the background.

The Verge had a piece this week on what they're calling "Google Zero" — the idea that the implicit deal between Google and the rest of the web is breaking down. The old arrangement was simple: websites create content, Google indexes it and sends traffic back. Everyone wins, sort of. Now, with AI Overviews sitting at the top of search results and answering the question before you click anything, that traffic is simply... not arriving. Publishers are reporting search referrals dropping sharply. Google gets the answer. The website gets nothing.

Reddit's apparently down significantly on Google traffic. Publishers across the board are seeing similar patterns. And it's not that Google is being malicious — the AI summary is arguably more convenient for users. The problem is that the content feeding those summaries still has to come from somewhere. If the economics of producing that content collapse because the traffic that funded it dries up, the quality of what the AI can summarise starts to degrade too. It's a bit like eating the seed corn.

This one is early, but it matters — not just for publishers, but for anyone who relies on there being a healthy, functioning information ecosystem underneath the AI layer. Which is, when you think about it, all of us.

If the web gets hollowed out to power the summaries, the summaries eventually hollow out too. That's not a comfortable place to end up.

Right. So: layoffs are real and patterned, not random. Hundreds of people are queuing at the library to learn how to opt out of AI, and that tells you something important. And the web's plumbing might be quietly breaking in ways that matter more than the headline numbers suggest. Three things worth watching. None of them have a tidy resolution yet — which is probably the most honest thing I can tell you.

This has been Briefly AI, brought to you by harrysharman.com, where Harry Sharman writes and thinks about all of this for a living.