Google’s NotebookLM, a note-taking app that’s infused with a large language model, is now available to everyone in the US. The company built NotebookLM to only pull information from a custom dataset. So you might use it, for instance, to help write a paper based on a specific set of PDFs and only rely on details from those documents.
The idea is that, by sticking to a custom dataset, the information NotebookLM handles will always be relevant to your specific needs as opposed to a more general chatbot that might draw from the entirety of the public internet. Google debuted NotebookLM as Project Tailwind at I/O this year before rebranding it and it says the tech is based on its Gemini Pro model.
Google also notes that personal data isn't used to train NotebookLM, so any sensitive or private details in your sources will remain hidden. That is, unless you share the sources with collaborators.
Along with making NotebookLM more broadly available (Google has been granting access based on a waitlist over the last five months), Google has announced a string of new features for the software. It plans to roll these out over the next few weeks with the aim of easing the transition between reading, taking notes and writing.
Among other things, the chatbot can format your notes into a newsletter, script outline or a marketing plan draft and export everything to Google Docs with one click. You'll be able to pin notes to a Noteboard space above the chat box, save and pin responses from the chatbot as notes and hide the source material if you want to focus on jotting down your thoughts with fewer distractions. You'll soon be able to ask the AI to focus on certain sources (of which you'll be able to have up to 100 in total) or jump to a citation from a chat response or saved note.
NotebookLM will also suggest some actions based on selected text or notes. At the outset, you'll be able to combine notes, summarize several of them or create an outline or study guide.