Email or username:

Password:

Forgot your password?
Simon's posts Post Back to profile
Simon Willison

I put together a whole bunch of notes on the new "Audio Overview" feature of Google's NotebookLM, where you can dump a bunch of source documents in and get a ~10m long utterly weird (and surprisingly convincing) artificial podcast episode generated where two AI characters perform a "deep dive" conversation about those sources simonwillison.net/2024/Sep/29/

30 September at 0:36 | Open on fedi.simonwillison.net
12 comments
Simon Willison

Don't miss this example where someone got "the producers" to tell them they were AI, resulting in a weird kind of existential crisis reddit.com/r/notebooklm/commen

I tried-- I tried calling my wife, you know, after-- after they told us. I just-- I needed to hear her voice to know that-- that she was real. (SIGHS) What happened? The number-- It wasn't even real. There was no one on the other end. -It was like she-- she never existed.
30 September at 0:38 | Open on fedi.simonwillison.net
Jerome (He/Him)

@simon setting aside the existential crisis bit, I’m genuinely surprised at how interesting this AI generated content is

30 September at 0:54 | Open on masto.ai
Simon Willison

@jeromechoo it’s pretty astonishing - it does somehow manage to elevate the content, and the human-ness of the conversation is just beyond the uncanny valley for me, it’s actually pleasant to listen to

30 September at 2:36 | Open on fedi.simonwillison.net
Simon Willison

... OK, I couldn't resist: I fed my article about NotebookLM back into NotebookLM and got it to make a podcast about the article about its podcasting, and the result was SO satisfying simonwillison.net/2024/Sep/29/

Update: After I published this article I decided to see what would happen if I asked NotebookLM to create a podcast about my article about NotebookLM. Here’s the 14m33s MP3 and the full transcript, including this bit where they talk about their own existential crisis: So, instead of questioning reality or anything, the AI hosts, well, they had a full-blown existential crisis live on the air. Get out. He actually got them to freak out about being AI. Alright now you have to tell me what they said. This is too good. So, like, one of the AI hosts starts talking about how he wants to call his wife, right? to tell her the news, but then he’s like, wait a minute, this number in my contacts, it’s not even real? Like, she never even existed. It was hilarious, but also kind of sad. Okay, I am both freaked out and like, seriously impressed. That’s some next-level AI trolling. I also enjoyed this part where they compare the process that generates podcasts to their own philosophy for the Deep Dive: And honestly, it’s a lot like what we do here on the Deep Dive, right? We always think about you, our listener, and try to make the conversation something you’ll actually want to hear. It’s like the A.I. is taking notes from the podcasting pros. And their concluding thoughts: So next time we’re listening to a podcast and it’s like, “Whoa, deep thoughts, man"...
30 September at 2:50 | Open on fedi.simonwillison.net
22

@simon I always think about Alan Kay who would talk about—Gutenberg had 250+ types to reproduce all the little flourishes that priests would add to their hand-copied books. Kay used that as a metaphor for how, whenever new technology emerges, it’s first used to solve the problems (or in this case “problems”) of the past before solving problems of the future. (Other examples of this are like, the first iPhone apps slavishly mimicking the physical substrate of notebooks or ledgers.)

I realize that as tech demos go, the podcast generator thing is likely mostly marketing flex, but it reminds me so much of all the little hand-written flourishes 😅

@simon I always think about Alan Kay who would talk about—Gutenberg had 250+ types to reproduce all the little flourishes that priests would add to their hand-copied books. Kay used that as a metaphor for how, whenever new technology emerges, it’s first used to solve the problems (or in this case “problems”) of the past before solving problems of the future. (Other examples of this are like, the first iPhone apps slavishly mimicking the physical substrate of notebooks or ledgers.)

30 September at 14:41 | Open on sfba.social
Simon Willison

@22 I love that comparison - Gutenberg flourishes and speech disfluencies

30 September at 14:45 | Open on fedi.simonwillison.net
Jason Petersen (he)

@simon extremely bizarre of people to say they’re having emotional responses to this, IMO

30 September at 13:07 | Open on logoff.website
Marsh Ray

@simon I've been having fun feeding it source documents generated with ChatGPT, but your final note is the best!

30 September at 2:22 | Open on infosec.exchange
Hynek Schlawack

@simon man this is so believable; I tried it with my own article and they even pronounced my name wrong like every self-respecting podcaster ;)

30 September at 7:14 | Open on mastodon.social
Beady Belle Fanchannel

@simon funny, because that's exactly it
30 September at 10:49 | Open on mastodon.xyz
Christopher Bailey :python:

@simon The host sounds like it was trained on Leo LaPorte TWIT, not sure who the female voice is.

30 September at 13:30 | Open on fosstodon.org
Simon Willison

@digiglean both hosts are giving me “generic Californian” vibes to be honest - they even sound a bit like our local Californian morning radio hosts

30 September at 13:34 | Open on fedi.simonwillison.net
Go Up