Google upgrades NotebookLM with more sources and faster fact-checking

Google has revealed a series of upgrades and an increase in availability for its research and study app, NotebookLM.

NotebookLM, which debuted last summer, is built on Google’s Gemini AI platform, and is touted by the company as “research and writing assistant”.

Google announced last week that NotebookLM is now available in over 200 countries and is currently running on Gemini 1.5 Pro.

The app has also enjoyed a bump in functionality. Google announced a handful of new features for NotebookLM this month, including support for Google Slides and web sources. At launch last year, the system only supported Google Docs as well as the user’s own PDFs and files, meaning users would often have to copy and paste material from the web if they wanted NotebookLM to exploit it.

Trusting NotebookLM

NotebookLM only ingests the inputs it’s instructed to by the user – such as a set of scientific papers or a person’s own lecture notes – and will create outputs based on those sources alone. That way, students and researchers who use the app to create a digest of information or help with writing a presentation, for example, can have greater trust in the reliability of the outputs.

This is particularly useful after the company’s recent debacle with its AI Overviews feature, which saw users’ search queries met with distinctly odd responses (a user who asked how to stop cheese sliding off a pizza was told to add glue to it, for example).

When NotebookLM launched last July, Google noted “It’s always important to fact-check the AI’s responses against your original source material” – to which end NotebookLM includes citations for the facts it presents, enabling users to verify how reliable they are. In this month’s update, the inline citations will now go straight to the particular sections of the source text they’re based on, to speed up fact-checking.

Data tweaks

Data uploaded for use with NotebookLM won’t be shared, Google said, and won’t be used to train other AI models.

As part of this month’s upgrades, NotebookLM can now convert the source material into a series of data formats, such as FAQs, study guides or briefing documents.

Google also posted a number of examples of how the model is already being used, including creating local newsletters, managing storylines in fan factions, and handling information for non-profit grant applications.

It added that author and journalist Walter Isaacson, known his for biographies of Steve Jobs and Elon Musk among others, has been using NotebookLM to analyse the journals of scientist Marie Curie, the subject of his next book.