How to Instantly Create an AI-Generated Podcast on Any Topic
Practical applications of AI are becoming more and more clear. It's wild
The goal of this article is to share an extremely interesting AI discovery and how one can try it themselves, while giving some examples of how I’ve used it and a little demo.
Over the last few days, I’ve found out about a quite amazing new way to engage with information. One of the latest commercialized implementations of contemporary AI has been the ability to create an ‘AI podcast’ for free, and practically immediately. It’s all thanks to Google’s NotebookLM, an experimental AI-backed project that allows people to interface with their data in new ways
Now, that might have you thinking - “what does that mean” - “how does that even make sense?” amongst other things - and I think it’d just be easiest for you to try it yourself to see. Here’s how you can do it:
Go to the NotebookLM website and sign in with a Gmail account (it’s free)
Once signed in, select New Notebook
Select a data source (you can use documents, Google docs, YouTube videos (it’ll take transcripts), or just copy/paste your data
Once your source is uploaded, expand the Notebook Guide if it isn’t already
Under Audio Overview, you’ll see ‘Deep dive conversation - Two hosts (English only)’ - select Generate
Give it 2-5 minutes to process
Once done, you’ll be able to play the podcast or download it for offline listening!
I’ve been absolutely mind blown since finding out about this. So far, I’ve uploaded and listened to conversations about the Starr report, a book about the history of Dance Dance Revolution, personal essays written in my youth, the Constitution, patch notes from a game a friend plays… even my own home health assessment. As you listen, you’ll hear interesting little artifacts (glitched out words, mispronunciations, possible hiccups due to how your document was formatted). Even with all of that in mind, this is a genuinely amazing thing to have access to and actually had me grinning like a child when listening to things.
What excites me the most is a new medium for accessing information - you can ‘listen’ to your documents now. Imagine wanting to learn about a new topic - downloading a bunch of reference materials and alongside your normal studies, you start listening to these podcasts while driving around - which are things you specifically wanted, about a topic you want to learn, whenever it’s most convenient?
I think the implications of this are large - this is only an experimental free tool Google put out there and it’s in its earliest consumable state… it only gets better from here. What does that mean for future generations - does the ability to create new content atrophy with convenient tools like this? And while these tools are great (and seemingly limited in scope due to the generations being from individual files) how will people come to discern truth from misinformation (or even inadvertent hallucinations)? Interesting to think about. For now, I’m going to listen to some more cool stuff!
For fun, I’ll be copying all the text I just wrote in this article into the system verbatim and will make a podcast out of it - you’ll be able to see an example in real-time and get a sense of how it processes what you upload. Here you go:
I hope you found this useful and that you are inspired to try creating your own ‘podcasts’. Happy listening 😊