NotebookLM is becoming the podcasts app I didn’t know I needed

Many people are already AI-weary, including me. From producing bland content and replacing our jobs to water-hogging data centers and the existential doom risk, there’s a lot to hate about it. ‘AI slop’ has quickly entered our vocabulary, and for good reason. It’s high on that list of reasons to dislike the tech for many people, but we shouldn’t tar absolutely every bit of media churned out by AI tools with that brush. After my recent introduction to AI-generated podcasts created with NotebookLM, I’m ready to defend them — at least to a point.

It was Google’s excellent app that opened my eyes, and unlike half of my colleagues, I’ve never even used it myself. Welcome to 2026, where my second-hand AI experiences count as research.

I made this realization because I recently visited South Korea for the first time, and thought it might be useful to get a feel for the place before I arrived. Podcasts are my medium of choice for learning these days, so my natural instinct was to jump into my podcast app and see if there was a pod dedicated to Korean culture. After all, there’s one for almost everything else. After scrolling through a few that didn’t quite fit the bill, I came across ‘Understanding Korea, One Story at a Time’ by Jiwon Yoon, Ph.D. It seemed to be just the explainer on the country’s history and culture that I was looking for.

The thing is, Jiwon wasn’t the host of the early episodes I was listening to. She’s a Korean-born writer and former tenured professor who writes a Substack about Korean society. This involves extensive research, so she decided to dump all her notes, including those that didn’t make it into her Substack articles, into NotebookLM. As you may know by now, from there you can just use the Audio Overviews feature, and you’ve immediately got yourself a podcast episode. Jiwon explains that this is exactly what she’s done in the episode description.

I was in two minds about this. Instinctively, it didn’t feel quite right to have Korea explained to me by two hosts who are not only not Korean but don’t even exist. On the other hand, NotebookLM generates its Audio Overviews solely from the source material you give it, so the risk of sketchy references or outright hallucinations is much reduced. In this case, extensive notes were provided by a Korean native who knows the country better than the vast majority of people. I gave it a try, and was quickly sold on it.

If you’ve ever tried Audio Overviews yourself, you might agree that Google has generally done a very good job with the AI hosts. They’re far from perfect, occasionally saying slightly incongruous things, and you also start to notice certain patterns in the structure that are reminiscent of AI writing. However, they’re also really quite engaging and surprisingly good at emulating a rapport. They play off each other, marvel at the facts about Korean culture, and give Jiwon due credit for her work. Google has clearly done its research on the dynamics that make top podcasts easy to enjoy and has programmed them into NotebookLM.

Let me be very clear here: I’d much rather listen to a well-made podcast with human hosts. For example, I’m a big fan of The Rest is History podcast, and podcasts generated with NotebookLM covering the same material wouldn’t be half as enjoyable. We build a quasi-relationship with the human hosts of our favorite podcasts, and AI can do precious little to replicate that.

Having said that, there are tons of bad podcasts out there, and I think NotebookLM does a better job than many of them. It at least has the decency not to spend the first 12 minutes discussing the hosts’ weekend. That’s just my personal preference, by the way — I’m fully aware that this isn’t a good thing for the podcast industry at large. But if it is going to happen anyway, then budding podcasters’ loss will at least be academia’s gain.

Academics often have the best story to tell, or access to it. They have little-known facts and groundbreaking insights. There are countless discoveries and hot takes in academia that would interest a mainstream audience, but they may only get as far as a dense essay or an impenetrable thesis. Some become entertaining substacks or decently selling books, but it’s a very small percentage that make it past the written word.

One reason is that academics often aren’t naturals in the entertainment business. Were the bookish nerds also the charismatic class entertainers in your school? No, mine neither. That isn’t to throw any shade on academics, and there are exceptions to the rule — the aforementioned Rest is History podcast is a good example of bona fide historians who also know how to engage the listener.

However, countless other academics don’t have the inclination or the presentation style to convert their work into something for a mainstream audience. If Spielberg or Cameron doesn’t turn their research into a Hollywood blockbuster, the average person on the street is unlikely to hear it. And even then, someone probably has to be chased by a dinosaur first.

And for any scholarly types who want to make a podcast, it still takes some investment and know-how. You need the right space and equipment, which is becoming more accessible but still costs money. Then you need knowledge of how the medium works. Again, this isn’t hugely prohibitive, but it takes time and commitment.

NotebookLM and apps like it present an obvious answer. An insightful and well-reasoned essay on South Korean culture can be turned into a podcast for free in a matter of minutes, and having AI hosts is the trade-off. Jiwon Yoon now writes and presents her podcast, but it was NotebookLM that helped her get the ball rolling.

I can’t blame anyone for detesting the idea of the AI revolution, and I’m certainly not an advocate for it. But it’s happening to us whether we like it or not, so let’s at least embrace the opportunities it presents. If NotebookLM podcasts can bring ideas out of academic corners I’d never normally reach and into the time I’m doing chores or stuck in traffic, then maybe that’s one small part of the AI future worth keeping.