Content Library Steering Committee

I'm sure there are - or soon will be - dozens (hundreds) of similar competing products. A cool concept which , if it works, may save time querying knowledge.

https://9to5google.com/2023/12/08/notebooklm-gemini-pro-launch/

• Can only have up to 20 documents added to a notebook
• If you add a Google Doc to the notebook as a source - it's a static document that is imported. This means you can't add a Google Doc and then continue to update that Google Doc as a source. Once the Google Doc is imported it no longer updates for changes. At least in my testing, I couldn't see it updating.
• Just about every app invented since 1776 is going to have a "help me _____". So far this notebook doesn't go much further than embed an AI chat window very similar to ChatGPT or Bard.
  
  
  I wrote two documents using NotebookLMDocument #1 - Poor Result ( But might be better once I had more experience )
  • Write a blog post about a press release
  • Upload a number of documents that contain or support the press release
  • Upload a Wikipedia page (pdf) that outlines the history of the company
  • Upload some LinkedIn information about the person issuing the press release
    
    
  Admittedly I might not have known how to do this - however there was no easy way to say - write a blog post about press release X, quote person Y and give a summary of their background, and in the introduction, give me a summary of company z.THIS is what I would want for the product to be really useful. I could create bits and pieces and save, for example, the intro or the summary. But it would not write the whole thing. Not even a poorly written draft.
  
  Document #2 - Digest a more complex news release from SilverSun about their proposed transactionr
  • I was able to upload all the SEC filings in PDF - this is great because they don't change.
  • I also uploaded a transcript from a Q&A of the person involved in the transaction.
  • NotebookLM was very helpful at answering questions like "How much will management make from this" where I replaced the management with the name of the people. 
  • It also was very helpful in breaking down the transaction into simpler terms. Far easier than wading through SEC filing(s)
  • I didn't 100% trust the outcome - at least not enough to publish it as copy/paste. This is a BIG issue for me (and I suspect others)
  • However, it was very useful once I figured out to use the chat portion to create sections and not the entire post - for example, I could ask it to tell me how much the existing CEO stood to make - and it appeared accurate when I reviewed the SEC filing.
  • It was way off in assessing the transaction and thought that someone was buying SilverSun for $1 billion - which is not what's happening. This is a big issue. It looked like the AI took the press release narrative at face value and didn't do much analysis ( big problem, in my opinion). Sure this transaction is complicated and hard to understand - isn't that where AI should be helping? I found that it only partly helped and in some ways was misleading in its suggestions.
  This is good for a lot of background info where you might ask the AI for a quick summary. However, for analysis or asking, "is this a good deal for Silversun shareholders?" I rate it an F or incomplete. It's just not suitable for much more than background content unless you already fully understand what you're writing about and are only using NotebookLM as an assistant.