Mastodon Feed
Boosted by jsonstein@masto.deoan.org ("Jeff Sonstein"):
huey@social.ketupat.me ("Huey ☁️") wrote:
I am perhaps a bit late to the game but NotebookLM is pretty good. I had it generate a podcast based on a recent trade mark decision in Singapore that I am familiar with, and it picked up on nuances that GPT-4 didn't (such as the difference between well-known and well-known to the public at large).
(I gave it the PDF decision, as well as the IPOS trade marks infopack and the Act for context)