[RobinL]

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 feels like a potentially very powerful way to learn.

[usaar333]

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

[bbor]

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.

[usaar333]

https://repository.law.umich.edu/mjlr/vol25/iss3/3/

Most of this is unlikely to be in training data.

Doesn't even mention the basics, like ethnic demographics of Fiji today. Confuses history as well (what happened in colonial times vs post independence)