Blog

Podcast with NotebookLM

By Michael Faber // October 4, 2024

Google's NotebookLM is a pretty interesting tool - it allows you to organize a bunch of research in the form of your own writing, external documents, AV, internet links, etc, and then can build out summaries or other outputs based on that corpus of information.  It has been around for nearly a year, but it's the new "audio overview" feature that has been gathering some headlines lately.  This tool essentially turns all of that research into a 'deep dive' podcast between two seemingly human hosts.  It does a pretty fantastic job of making it sound like a real podcast, right down the to (sometimes overly dramatic) banter we love and expect.

There was some recent buzz around this podcast feature where, in a recent "episode," the AI realize they are not human and have an existential crisis.  It is worth a listen, even if the creators kind of juiced the situation - there is no real sentient awareness here, just a really good simulation of it.  

That said, I thought I'd give the feature a try.  So I took this blog post that I recently wrote (shameless plug incoming) about the Co-Lab's partnership with artist Sherrill Roland and his new exhibit in the Nasher of art created entirely at the Co-Lab (so much plug), and stuck it into NotebookLM.  I CLICKED ONE SINGLE BUTTON, and it produced an 8 minute "deep dive" podcast that sounds so unbelievably real.  Check it out:

The moments of true AI shock have been getting fewer and farther between as I use it daily at this point, but this one took my breath away.