How Google’s NotebookLM AI Chatbot Became My New Study Buddy

Did you know you can customize Google to filter out garbage? Take these steps for better search results, including adding Lifehacker as a preferred source for tech news.

Though its mobile app is quite bad, the web version Google's NotebookLM has become one of my most-used tools since I started playing around with the AI large language model last spring. I've found a ton of ways to use it to be more efficient in my everyday life, which is (much) more than I can say for other AI-enabled applications I've tried.

Instead of (poorly) doing the work for you, NotebookLM acts as a true helper, assisting in organizing information for you without being intrusive. And one of the most useful ways to interact with it is via its familiar chat interface, which has turned the program into my new study buddy, and a resourceful co-worker.

How I use NotebookLM's AI chat for studying

I've been studying for a certification exam and have been uploading my materials into NotebookLM whenever I get to a new chapter. From there, I've been able to make flashcards, mind maps, practice quizzes, and study podcasts with a few clicks. The podcasts are especially unique and helpful: Two AI-generated voices discuss the concepts from my notes and materials in an informational, conversational way that sounds just like a podcast—if said podcast were a little too pedantic and on-the-nose. (Sometimes, it even adds in tics like vocal fry to make things sound that much more realistic.)

Credit: Google/Lindsey Ellefson

Usually, I listen to a podcast on my chapter while I complete a chore, like packaging up my Poshmark sales. Then, I go back to the notebook and have it generate flashcards or a quiz to see how much information I retained. Each flashcard or quiz question includes a link to the source material, so I can click it and review the full passage in the left-side panel.

I find the tool so useful, I've become something of an evangelist for it. A friend of mine recently went back to school, and I taught them how to upload the slide decks, chapter scans, PDFs, and other materials from their online classes into NotebookLM. They now use the chatbot to summarize a given class's content and create outlines for discussion posts.

Using the chatbot to create outlines has been helpful for my friend, who struggles with writing but excels in thinking practically. They're able to ask the bot to generate key dates, important facts, and other concrete facts that they should be studying and referencing in their discussion posts, but which they find difficult to pull out of the dense material themselves.

The chatbot has also been useful for my own studying, but in a different way. Unlike my back-to-school friend, I have no idea what's going to be on my certification test. I don't get a study guide. I know that out of 23 chapters of material, a random selection of questions will be given to me, which means I have to study everything, hoping I adequately cover what appears on my final exam. So after listening to my podcast and drilling my flashcards, I ask the bot to summarize the main idea of each chapter and suggest five of the most crucial points within it. Even though I know I have to read the whole thing, this gives me a roadmap for where to start and what to look out for. In that way, it's similar to using a traditional study technique like KWL or SQ3R to formulate questions before a critical reading session.

I've also like to tell the chatbot to "have a conversation with me" about specific topics within the chapters. This is my version of the Feynman technique, in which you interrogate your mastery of a topic by trying to teach someone else about it, although I'm not "teaching" the bot so much as talking to it. By going back and forth conversationally, I practice describing the terms and ideas in a casual way, which I can't do if I'm not intimately familiar with them. If I find myself struggling to keep up the conversation, I know I have to study that section more.

How I use NotebookLM's AI chat for work

Back in May, I published a series here on Lifehacker entitled "Moving Made Simple." In writing and researching nearly a dozen articles for it, I interviewed something like 13 moving pros from around the world. I ended up with a stack of interview transcripts to sort through, and the idea of rereading them all to highlight the most important information, plus cross-reference reoccurring advice shared by multiple sources, was daunting, to say the least. I found myself struggling to remember who recommended what, and in what context, and determining which story each tip was best suited for.

But then I turned to NotebookLM. I opened up a new notebook and dropped in copies of all of my transcripts. Because NotebookLM only pulls information from the sources and materials you input, I could then ask its built-in chatbot questions without worrying it would spit out nonsense culled from a dubious, 10-year-old Reddit post. I asked it things like, "What do the sources recommend for carrying large furniture?" and "What order should someone pack their belongings in before a move?" and NotebookLM made simple lists in response, with each answer including a hyperlink to the exact transcript from which it came.

A number of pros recommended the same things, like packing hanging clothes in a garbage bag with a hole cut out for the hanger hook, and NotebookLM condensed all of those, highlighting that this was a top tip that I should definitely include. On request, it also included each speaker's name next to a given tip so I never had to go searching around for who said what, or risk misattributing something. NotebookLM made summarizing, organizing, and attributing all of this material so much easier than if I had spent hours going through the transcripts and highlighting everything myself.

Credit: Google/Lindsey Ellefson

Another benefit: Unlike ChatGPT, I didn't feel gross about using this chatbot, as it wasn't exploiting the work of others or risking producing questionable results or full-blown hallucinations—all out its output was based on my own hard-won input. I did the work of formulating questions, finding sources, and conducting interviews. The chat program just helped me organize it all so I could use it to turn the quotes and information into useful, actionable stories. (Note that at no point did I ask AI to help me draft articles—I used it for backend organization only.)

In the months since, any time I've had a large volume of materials to work through to complete a given task, NotebookLM has been a useful tool. I sometimes use the chat feature to ask it for summaries, main ideas, or outlines of my notes, but I most often employ it to pull out quotes or standout concepts. It has helped me become a much more organized, efficient writer.

How I use NotebookLM in my daily life

As long as there are relevant, concrete materials available to upload to the tool, you can NotebookLM for other purposes as well. I think of it like relying on a single source of truth (SSOT), an older productivity hack that involves putting all the materials related to a certain project into one computer folder—but better, because in addition to giving you space to store all your resources, NotebookLM can organize them for you. Personally, I've uploaded copies of my schedule and asked the tool to outline my travel and time commitments for the week. (Not everything you input has to be a file or link. There is a plaintext option where you can just type out whatever you need the bot to know.)

But you can get much more creative with it. Say you're looking for a new job. You might upload a job description and ask the NotebookLM chatbot to prepare 20 potential interview questions you might face. Or go farther: Upload relevant news articles about a prospective employer's culture, recent scandals, milestones, or the latest investor report, and use all of that to gain an overview of the company. The more information you provide, the more useful the output will be—and again, it will only pull from those sources, meaning you should be able to trust its summaries. To make sure of this, I periodically test the chatbot (because I'm a suspicious person). I'll ask it about a totally unrelated topic that I know isn't in the materials I uploaded, and so far, it has always told me my question is "interesting," but that there is no reference to it in the provided resources.

Other possible use cases could include summarizing meeting transcripts, asking it to suggest meaningful themes in your recent journal entries, or outlining a five-year plan for your career. Your options are almost endless, and bound only by the quality and volume of the source material you upload.

View the full article