Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

NotebookLM's automatically generated podcasts are surprisingly effective

simonwillison.net

Post reply on HN
Print view View original

311–320 of 511 posts

Re: NotebookLM's automatically generated podcasts are surprisingly effective

#313

Post by freedomben »

humansareok1 wrote: post #296
handelaar wrote: post #277

So basically what you're all saying is how it's technically impressive. Okay. It is also completely and utterly worthless -- an inefficient and slow method of receiving not-very-many words which were written by nobody at all. The one and only point listening to a discussion about anything is that at least one of the speakers is someone who has an opinion that you may find interesting or refutable. There are no opinio…

There's definitely not nothing of value here. This could be a useful new medium. I however hate the tone of the two hosts. It sounds like two pompous millennials talking about things they don't really understand.

Indeed, you nailed it.

The ridiculous overuse of the word "like" is as nails on a chalkboard to me. It's bad enough hearing it from many people around me, the last thing I need is it to be part of "professional" broadcasting.

I'm super impressed with this, but that one flaw is a really big flaw to me.

Re: NotebookLM's automatically generated podcasts are surprisingly effective

#315

Post by sodality2 »

mdp2021 wrote: post #308
sodality2 wrote: post #299

Earlier quoted context omitted.

Conversational audio form is really not an "industrial snack". If I had the chance to listen to podcasts about any topic, I would do so much more often - uploading PDFs of academic papers, manpages, etc.

Yes, but should not you wait for the generated content - the text - to be at proper level? We have "Francis Fukuyama vs John Grey" available... If the purpose is serious, of information access management, why did they elect the form of a pisstake ("like")?

I personally appreciate the lighter introduction/discussion into a topic. That may be all it's good for, and that's okay. I'm not replacing my reading with this any time soon, precisely because of the problems you mention.

Re: NotebookLM's automatically generated podcasts are surprisingly effective

#316

Post by bbor »

usaar333 wrote: post #303
RobinL wrote: post #291

Earlier quoted context omitted.

To take one example of where this is valuable: - Take some dense research paper or other material that is unsuitable for listening to aloud - Listen to it (via NotebookLLM) whilst commuting/washing up or whatever This way you'll have a big headstart on what it's all about when you come to read the details. I imagine in future we'll see a version of this where the listener can interject and ask questions too, that fee…

I tried that with a paper. It emphasized the wrong points and 8 out of 10 minutes were just filler. I like the idea of audio based formatting, but this particular implementation is quite inefficient

Interesting! I tried it with a (famous, tbf) philosophers book and it did pretty well. Absolutely not optimized for speed, but that’s on purpose. Could you share what field/type of paper you tried? I’m not doubting you at all — I’m sure it still has many topics it fails to capture, mathematics probably being one of them.

Re: NotebookLM's automatically generated podcasts are surprisingly effective

#317

Post by ben_w »

mdp2021 wrote: post #295
sodality2 wrote: post #285

Earlier quoted context omitted.

This is some insane catastrophizing. The value is that it turns it into a form factor that may be easier to consume, pay attention to, etc.

Since when industrial snacks are healthy food?

Indeed. But MREs, protein shakes, Huel etc. are also a product of industrialisation.

In this case, I could see potential value for a better iteration of this tech, making it a meal replacement shake rather than a candy bar.

There's too much interesting content for me to read it all, and I have a long commute. Right now I'm using that commute to learn German, and that is a good use of that time, but let's say I didn't need to because I hadn't moved country or I was already fluent: in this hypothetical, I'd gladly have a better AI than this(!) generate podcasts about the articles that I don't have time to read.

But the AI would need to be better than this one for that to be worthwhile — I just popped one of my own blog posts into it, and it was kinda OK-ish, but did make some stuff up. Now sure, the Gell-Mann Amnesia effect was written with humans in mind, but that's a shared disappointment and not a reason to let this AI off that particular hook.

Re: NotebookLM's automatically generated podcasts are surprisingly effective

#318

Post by humansareok1 »

freedomben wrote: post #313
humansareok1 wrote: post #296

Earlier quoted context omitted.

There's definitely not nothing of value here. This could be a useful new medium. I however hate the tone of the two hosts. It sounds like two pompous millennials talking about things they don't really understand.

Indeed, you nailed it. The ridiculous overuse of the word "like" is as nails on a chalkboard to me. It's bad enough hearing it from many people around me, the last thing I need is it to be part of "professional" broadcasting. I'm super impressed with this, but that one flaw is a really big flaw to me.

I really want to like it more, it could be interesting to drop in a textbook and get a dedicated series of podcasts about each chapter for example but the tone is so off-putting that I can't listen for more than a few minutes. Its pure cringe.
Post reply on HN