Goodbye ChatGPT: 7 Reasons Experts Are Running To NotebookLM

Everyone's talking about Chat GPT, Gemini, Claw, and Gro 4, but everyone is absolutely sleeping on this one underrated AI tool called Notebook LM. And in this video, I'm going to show you why you need to start paying more attention to Notebook LM than any of the other platforms. Because I believe personally that it's the future of AI and how we actually use it. If you

listen to a lot of the tech people talk about artificial intelligence, one thing that's going to come up is artificial general intelligence. And this is when a large language model is able to perform 80% of all human work. And that includes reasoning, problem solving, and so many of the other things that humans bring to the table when they perform their work. Another term that we don't hear nearly enough that's just as important as artificial general intelligence or AGI is narrow AI. So artificial general intelligence is many tricks. This is a large language model that can do everything. But narrow AI is basically a one-trick pony. It focuses on one specific domain. Now, I want to do a

thought experiment with you right now. And I want to ask you a simple question. How many books do you think that it takes for a human to learn, to read, to synthesize, study, be tested on, and apply to actually become an expert? I

didn't know either. And so, I ran the prompt with Chat GPT, but I also ran it with Claude and Gemini so that I can get some outside input to make sure that I wasn't operating inside of a vacuum and getting slanted data. Even then, I did it two different times. I did it as a typical prompt and then deep research and both times I came back with a general consensus across all three AIS. When I ran the simple prompt, the numbers were lower and then when I ran the prompts with deep research, the numbers were very different. There were about 100 books for level one, around 358 or 500 for level two, and somewhere close to 1,000 or more books to be an industry thought leader. And just to give you an understanding of what we're talking about, for level one entry- level experts, we have a licensed clinical social worker, a certified public accountant, and a licensed professional engineer. For level two mid-level specialists, we have board certified emergency physician, senior software architect, and then an investment portfolio manager. And for level three industry leaders, we have a federal court judge, Fortune 500 CTO, and a psychology department chair. And so now we can safely assume that it takes anywhere from 100 books in a particular domain to become an expert and anywhere up to 1,000 books within that domain to become a thought leader that influences how everyone else thinks in that domain. And I'm also going to make one more assumption and say that when people are looking at AI, they're looking for it to be a thought leader. have answers that we don't necessarily have. We're looking for it to at least be midlevel or more. So, if you're an expert and you're looking to create an AI expert that can help you with your work, then you're probably going to wanted to have access to or be trained on at least 500 to 1,000 pieces of data from your particular domain. And with your Notebook LM Plus account, you can upload up to 300 sources of data in varying sizes. Now, of course, it's not able to contain 1,000 books or 5,000 books of the highest quality data at the moment, but that's still far more than you're getting with chat GPT projects, Google gyms, or anything else that allows you to upload files. But imagine having a single sandbox or a single space where you had 500 to 1,000 points of data within your particular industry, whether it's how to grow a YouTube channel or anything else. And the AI only interacted with that particular content. And every time it spoke to you, it spoke to you with a citation so that you can hover over it and know exactly where it retrieved the data that it was basing its reasoning on. And for a lot of the

work that I do, that's one of the main reasons I find myself using Notebook LM more and more inside my workflow these days because it has practically zero hallucination. And another reason why I'm starting to lean towards Notebook LM as being the future of AI over ChatGpt and Claude and Gemini in general is because if you give Notebook LM a prompt and you request information and it does not have the necessary sources to answer your prompt, it's going to tell you that it can't answer. doesn't have access to the right data. And while you may think that that's a bad thing, I would rather someone tell me I don't know than make something up, which is hallucination, which is what chat GPT and every other AI does a lot of, which is why they have to have these disclaimers. Make sure you check the responses because the AI can be wrong. But with Notebook LM, it's going to always pull from the sources. And this brings me to the second reason why I prefer Notebook LM over the typical vertical chat. You can have

hundreds of resources in a single notebook, but you can ask a question that allows the AI to give you a single coherent answer based on information spread across those different resources. This is something that you absolutely cannot do with chat GPT, Claw, or Gemini because of the reasons I previously stated. You cannot upload this many documents and the size of the documents are severely limited. And not to mention, there is a rumor that the Notebook LM team has being able to cross chat across notebooks. It's on their product roadmap. So, if this is true in

the future, not only will you be able to upload 300 sources to this notebook, 300 to another, and 300 to a third notebook, but you will then be able to have Gemini 2. 5 Flash interact with each of those individual notebooks to give you a single coherent answer across 900 sources. The next reason why I prefer Notebook LM over Chat GPT and other AIS these days is because of context window.

When it comes to Notebook LM, it doesn't have a context window like we think of one where it begins to forget things because the conversation has gone on for so long. Because Notebook LM has a very specific rag system, a retrieval system where it goes and gathers information and comes back with it, it doesn't run into those problems. It's right there in the sandbox. And so you don't have to worry about sitting down and working inside of Notebook LM and the AI beginning to drift and bring in unfactual information or bias or anything else that can taint your research or cause you to waste hours of time, which I've done with ChatGpt and you probably have too without even knowing it. You can save any responses you receive inside your notebook as a note and then you can actually convert that note to a source so that it's never

lost. And that brings me to the next point because even though notebook LM doesn't have the creative space that say chat GPT5 or Claude or Gemini 2. 5 Pro have or Gro for have because it's contained in the sandbox. It still has a certain level of creativity that those AIs cannot have. Because of those particular platforms inability to handle large volumes of information like notebook LM can, they don't have the capability to actually find latent ideas that are hidden within a person's thought process. So for instance, I have a notebook that has 300 of Alex Hermosy's content uploaded and I did a video about it that I'll link to right here. And even though I know 300 videos is a small sample of his content and an even smaller sample of his actual mind, it's still enough for us to get an idea of what he thinks about some things and to pick up on some patterns. And so I asked Notebook LM look through his content and look for frameworks or ideas that frequently come up that Alex has not explicitly stated the frameworks. And one of the frameworks that it discovered is his ideas around constraints. Finding the constraints, solving for the constraints in a business and several others that he does not name as frameworks. It even anticipated what his next two books would be based on the pattern of his thinking. And you can check out that video right here to see it in action. And I touched on this next point briefly earlier, but inside of a notebook LM, your knowledge compounds and you're able to go deeper. So whenever you get a response across those hundreds of different sources and in the future if the rumor is true across those thousands of possible sources and you have that single response and you want to save it because you've discovered something or you've formed a new idea, then you can save that note into your AI studio and you can convert that note into a source. But this is where things get even better. After you turn that note into a source, you can turn off all the other notes. And then you can begin to study based on that single note only. Or you can turn that note into a mind map, an interactive podcast, a timeline, an FAQ, a briefing, or any other asset or document that they have inside of the AI studio. And based on the responses that you receive interacting with Notebook LM, based on that single source, you're then able to save those notes, place them in the studio, and then add them as a source. So that once you find an idea that you want to keep digging down deeper into and unfolding and exploring, it's so easy to do so in a way that keeps your drill bit drilling straight downward. Whereas with a typical chatbot, you have to copy and paste responses to save them elsewhere later or use my long-term memory protocol, which I'll link to a video right here where you can check it out, so that you never lose any of your memories and you always have them saved on your own private server. And one thing that I do believe would speed up this process of platforms like Notebook LM taking the lead over other platforms is if their AI studio was more like Claude's artifacts or at least had more breadth in what it created. The audio podcasts are awesome and especially the interactive podcast and I think those directly rival chat GPT voice mode because I'm able to have a conversation with two different voices and it makes me feel as though I'm in a group chat number one and number two it's based specifically around the topic that I want to discuss and there is no hallucination there is no drifting and so yes there's no creativity there's no exploration of the internet etc but notebook LM is not primarily for the person who wants to explore new ideas. They're get to the bottom of a particular idea, who are working in a particular field. And so, Notebook LM is perfect for experts. And it's perfect for people who want to become experts or thought leaders in their industry. If you want to become a leading content creator in your industry, there is definitely a use case for you to use Notebook LM. Now, there

are a few things I think would actually speed this process up for Notebook LLM to actually overtake Gemini Claw and even chat GPT and Grock as the preferred platform for experts. And I'll go ahead and say this too that if they were to make these few little tweaks to Notebook LM, I believe that more people would be willing to subscribe to Gemini Ultra, their $125, $150 plan, I think. And that's if they made these very small changes to Notebook LM that I believe would have a huge impact on the experience. And number one, that's if they embedded Gemini right there in the middle of the chat. Now, I know that there's probably going to be some problems with how you could actually pull this off. I don't know exactly how, but if I had access to Gemini in the sandbox and I had a toggle at the top of the conversation where I could turn the sandbox on or off and when the sandbox is on, it operates as typical notebook LM. it locks it down to the available sources and I could turn it off so that it's able to be more creative with the data that is present and then I could turn it back on. And something else I wish I was able to do inside notebook LM is to choose models instead of just using Gemini 2. 5 Flash for all of my queries. What if we were able to use 2. 5 Pro and Google Deepthink as well? Especially if you have a prompt that's going to span several different notebooks and it's covering hundreds and possibly thousands of data points that you have, you're going to need some Google Deep Think credits to actually get quality information from that because there's probably going to be some degradation of the responses based on the workload that you actually give 2. 5 Flash. That's just my theory.

Another thing that would be great is if in the process of embedding Gemini that we had some sort of memory. What if there was like a memory core or a notebook that actually gathered information about us personally that we could leave there? Kind of like the persistent memory that chat GPT has. But instead, we have a single notebook where there are certain facts that notebook LM is gathering about us or Gemini embedded and storing in that single notebook. Now remember, you can have up to 300 sources in a single notebook. And these are full on YouTube videos of varying lengths. And maybe it's pulling the transcripts from the channel, but you can still upload 300 documents. Those are a lot of words. detailed facts that an AI can write down about you and save there. So, Notebook LM, if they just had one notebook that was like your memory core, has the potential to be the most personalized AI that understood everything about you and everything that you were studying and interested in. I think I already mentioned the branching capabilities because as it is right now, even in Gemini, if you do a deep research and then you try to switch to another tool, it'll start another conversation. And so just the ability to switch between tools in a single conversation and the ability to allow Notebook LM to have tool calling when it's dealing with this information would just take the experience to an entirely different level especially when you can maintain that groundedness by turning the sandbox on so that it uses tools but it does not use knowledge or information that goes beyond the sources that you have uploaded to this notebook or another notebook. And I don't know exactly how all of that will work out, but I just see so much potential for small and inexpensive language models to do a lot of the work that we're actually

looking for AI to do. Instead of waiting for these companies to build this massive AI that knows everything, you can start creating your own little expert with Notebook LM for whatever your field of expertise is or whatever you're learning. And you can start working on becoming a better expert in that field or learning from expert knowledge in just a few minutes. And if you want to know how to actually set this up, then make sure you watch this next video right here that's going to give you a full walkthrough of how to set up a notebook with an expert's knowledge and then start extracting everything that they know from it or at least extracting the best ideas that they've had and publicly share so that you can start implementing them in your own life and in your own work and seeing the results. So, make sure you check out this next video right