This looks nifty and scary. I suspect that giving the user the ability to upload their "own" borrowed content may release Google from the liability of copyright infringement of the source material to build the language model (thanks to the DMCA [USA law] exemption for platforms that host user content).
@@erikals I tried adding the lightwave website URL and the vizrt legacy forums but the NotebookLM chat was not particularly successful in answering basic questions. I cannot tell if adding a web site instantly utilizes Google's existing index of that site or if it takes time for NotebookLM to "borrow" the web site content for analysis. The LW2020 manual exceeds NotebookLM's 500000 word count so I'll need to break that up and try again. Dan's "inside LW" book was "ingested" well so I'm sure he'll be happy that he's been replaced by our new digital overlords.
@@JamesOblak it will probably have problems parsing some websites / forums and html documents. i think pdf / doc / txt files is the most secure. and perhaps screenshots. (since nLM has an OCR scanner i think).
Q: so... you can upload LightWave related documents?
A: Yep.
Unbelievable.
This looks nifty and scary. I suspect that giving the user the ability to upload their "own" borrowed content may release Google from the liability of copyright infringement of the source material to build the language model (thanks to the DMCA [USA law] exemption for platforms that host user content).
agree, there are 2 sides to this. as for Copyrights... it seems Ai companies don't care, at all.
@@erikals I tried adding the lightwave website URL and the vizrt legacy forums but the NotebookLM chat was not particularly successful in answering basic questions. I cannot tell if adding a web site instantly utilizes Google's existing index of that site or if it takes time for NotebookLM to "borrow" the web site content for analysis.
The LW2020 manual exceeds NotebookLM's 500000 word count so I'll need to break that up and try again. Dan's "inside LW" book was "ingested" well so I'm sure he'll be happy that he's been replaced by our new digital overlords.
@@JamesOblak it will probably have problems parsing some websites / forums and html documents.
i think pdf / doc / txt files is the most secure. and perhaps screenshots. (since nLM has an OCR scanner i think).