The best podcast I never recorded

The question of what artificial intelligence can actually do well is as old in the public debate as the question of whether Julian Nagelsmann is a good national football coach. Every day, new AI tools hit the market - most of them specialized for niche tasks but excelling at what they do. Curiosity drives our author Markus Sekulla to constantly explore and test these new tools. Today, he took a closer look at Google’s NotebookLM, which has recently been further integrated into the Google ecosystem.

AI is everywhere. It has come to the point where there are hardly any tasks left that I can imagine that it can’t significantly enhance. Whether it’s generating hyper-realistic business photos, chatting about political party platforms before an election, or turning written text into a podcast.

Let’s focus on the latter. When I first heard about NotebookLM, I assumed it was just another (probably unsuccessful) attempt by Google to catch up in the AI race. Sure, summarising text into a podcast sounds like a good idea, but it also felt like yet another tool that sounds cool on paper but would ultimately gather digital dust – too rough around the edges to be genuinely useful for a few more years. Kind of like Google Translate or Yahoo Babelfish, which held the translation throne for about five years (some of you may remember) before DeepL and ChatGPT came along and – excuse my language – made AI-powered translation actually awesome.

But since curiosity usually wins over skepticism, I gave NotebookLM’s podcast feature a try. Right now, it's only available in English, and I have to say – it’s ridiculously good.

It’s like… a New York City filter for my notes

NotebookLM can analyse documents, notes, or even videos and generate a spoken audio summary – formatted as a dialogue that sounds like an actual podcast discussion. Of course, it has that very American flair, full of “Wow” and “Really,” but hey, that’s part of the authenticity. And that’s exactly what surprised me. Instead of robotic, Oxford-English phrases with clipped endings, the AI effortlessly throws in the kind of “It’s like…” phrasing you’d hear from any Manhattan native.

The process couldn’t be easier: upload a text or multiple files, set a topic, and within minutes, I had a mini-podcast based on my own content. My favorite use case? Uploading those documents or notes I had been meaning to “go through at some point.” You know the ones. Just drop them into NotebookLM, then listen to the summary while taking a walk – or, if necessary, while cooking or brushing your teeth. You’ll find an example of this very text in the conclusion below.

My Aha Moment

The first time I created a NotebookLM podcast, I had one of those moments when you realise just how much is possible with new technology. We already know AI is transforming nearly every aspect of life and work, but this hyper-realistic AI-generated audio made that shift feel even more tangible.

Sure, this may not be the most cutting-edge AI out there, but it definitely reshaped my perception of how convincingly AI will mimic human skills in the coming years. The bar has been raised.

Conclusion

There are millions of AI use cases out there, and in the coming years, we’ll all find the ones that make our lives significantly easier. For me, ChatGPT plays a major role in my daily workflow, but testing new tools is also part of my routine. Google’s NotebookLM left an impression on me, opening my eyes to AI’s capabilities in a way I didn’t expect.

Maybe it will do the same for you – especially when you listen to the podcast generated from this very text in just a matter of minutes (Google Account required):

Text: Markus Sekulla