Google’s NotebookLM – DEV Community

What is Google NotebookLM?

Google NotebookLM, released in July 2023, is a platform designed to help users better understand and engage with the information they already have. It aims to offer an intuitive way to extract insights and make sense of various types of data.

How Does It Work?

The platform allows users to analyze information from a wide variety of sources, including books, websites, YouTube videos, Google Docs, Google Slides, PDFs, audio files, and pasted text.

I personally like this platform for primarily three reasons

• It does not use the information you upload to train it’s model, unless you pull a review for a generated text.
• The unconventional method that information can be conveyed by a podcast rather than just a text
• It uses the data as context, so any information provided is more or less soley based on the data that has been provided, additionally, I appreciate that for each point in it’s response to a query, it references to the data source that was used for that point.

To demonstrate how the platform works, I’ll walk you through using an article from Wikipedia on the Boston Tea Party.

Step 1: Create an Account

Visit NotebookLM and you should see the landing page as shown below:

Click on “Try NotebookLM” and log into your Google account. After that, you’ll be redirected to the page below:

This page will display any notebooks you’ve previously created. As a new user, you’ll see an empty workspace. To start a new notebook, click Create.

Step 2: Uploading Data

A pop-up window will appear, allowing you to upload your data sources. Since I’ll be using a Wikipedia article, I will click on “Website” under the Link section and paste the URL of the article.

Step 3: Ask Questions

Once your data sources are uploaded, the page will look like this:

You have two options for interacting with the data:

Chat

As the name suggests, the Chat feature allows you to ask questions just like you would with any other language model. However, the key difference is that the responses are based solely on the data you’ve provided. This means the information you receive is directly derived from the uploaded content.

Studio

Studio generates an audio recording featuring two speakers who discuss the data you’ve uploaded, in a podcast format. Note that, as of the time this article was written, the audio is available only in English.

Instead of typing questions, you can “interrupt” the podcast hosts and ask them questions directly. After responding to your query, they will continue with the pre-generated script.

To generate a podcast episode, go to the Deep Dive Conversation section on the right side of the screen. Click Generate, and after a short wait, your podcast episode will be ready.

If you’d like the conversation to emphasize certain aspects of the data, you can click on the Customize button next to Generate. Specify your preferences here before clicking Generate.

Your section should now look like this:

Once you’ve generated the podcast, you can click on the Interactive button to engage with the hosts. Your screen should now look like this:

Here is the audio of the generated podcast on the Boston Tea Party.