EP 383: NotebookLM Updates - customizable audio and a business edition?

This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.

In a very short amount of time. Notebook LM has gone from this kind of unknown tool by Google to a powerhouse like now, y', all, I'm, I'm watching the evening news and Notebook LM is being talked about. So we've already covered it once or twice, but they just released some brand new features, customizable audio, and a upcoming business edition. So Today on Everyday AI, we're gonna get you caught up on NotebookLM, if you haven't already done a deep dive, pun intended there. But we're also going to go over what's new and do some testing live. So I'm excited about this one and I hope you are too. What's going on, y'?

All?

My name is Jordan and welcome to Everyday AI. We are your daily live stream podcast and free deadly newsletter, helping us all keep up with generative AI and using all of this to get ahead, to grow our companies and to grow our careers. If that sounds like what you're trying to do, awesome. You're in the right place. We do this literally every day, weekday, Monday through Friday. It's unedited, unscripted, the realest thing in artificial intelligence. So if you haven't already, please make sure to go to your everyday AI.com if you're listening along on the podcast. Thank you for tuning in. We do this thing live too, so sometimes we ask for audience help. We're going to be doing that today, but make sure you check out your show notes to go to our website and sign up for the daily newsletter. All right, so before we get into all of these new juicy Notebook LM updates that I'm excited for, let's first start the day as we do every single day, by going over the AI news. So Meta has partnered with Blumhouse to generate short films using Moviegen, a new suite of AI tools that enables users to create 16 second video clips from text prompts. So this new partnership with Blumhouse features some prominent filmmakers, such as Anish Chagante and Casey Aflac, who are experimenting with AI generated content, which could really reshape how stories are told in the future with film. So while Blumhouse's CEO Jason Bloom sees potential for these tools to enhance creativity, the broader adoption of AI in Hollywood has sparked some mixed feelings. Right, the whole AI taking our jobs and it being, you know, too good, and replacing small teams so an interesting partnership that there from Meta and Blumhouse. All right, our next piece of AI news. OpenAI has, I think, launched something that's going to be very popular, a desktop app for Windows for chat GPT. So OpenAI has just announced the release of its new Windows desktop application for its popular AI chatbot Chat GPT. So this has already been out for on Mac for many months, but the new Windows app is currently available in Preview for chat GPT + enterprise team and edu subscribers. So you have to be a paid subscriber to chat GPT to get access to this. And OpenAI's strategy right now goes beyond convenience. They're really seeking to deeper, deepen user engagement, improve data collection for model enhancement, and also create a robust ecosystem that could be difficult for competitors to penetrate. So, yeah, this, this piece is pretty interesting. I've been using the desktop app for, for Mac since it came out, and it's something that really does help your workflow. Just having quicker access to Chat GPT, not always having to, you know, launch a new tab, but it being right on your desktop, being able to launch it with a simple keyboard command, pretty cool. So, hey, let me know in the newsletter, why don't you reply to the newsletter today? Let me know if you're going to be using the new Chat GPT app. And then last but not least is a good transition for our show today. The Notebook LM updates has everyone talking, including us today on Everyday AI. All right, let's get into the topic today on what's new in Notebook lm. What you need to know with these new advancements and if I think you should use it. All right? It's going to be a fun one today, hopefully, right? Doing these things live, right? That's what we're going to be doing. We're going to be doing some live examples. So for our live stream audience, I need you to be ready, right, you can drop some ideas now, but here's, here's some help that I'm going to need from you all. One thing that we do right, and even my background, I was a journalist before, all right? So I like to say when it comes to AI, we're investigative practitioners. So we never just take what a company gives us and read it to you on the air. We always test things, we always try to break things, we always try to see what's this story that the company doesn't want us to know. So we do things live, right? Which, if I'm being honest, doing any generative AI demo Live to thousands of people can not, not so much nerve wracking, but it can go terribly wrong. Right. So I, I hope I can count on you all for some good suggestions today. So here's, here's what I'm going to need a little later. I'm going to need a common skill, something that you would want to learn, right? So as an example, maybe playing the piano or quantum computing, right? So what is something that most part, for the most part, people at least understand what it is, but maybe they don't know how to do it. So I need a common skill or a common subject that you might want to learn and you can go ahead and get those ideas in now for our live stream audience that's joining us like Jay from Grand Rapids or Christopher from Bowling Green and Kurt from Ohio. So I need some examples of a common skill and then a common group of people, right? So basketball players, healthcare specialists, eighth graders. Right. It's kind of like that old, what was it, the ad lib game, right, where you, you have to kind of fill in the blank. That's what we're going to do live to test out these new features in Notebook lm. All right, so if you don't know what Notebook LM is, don't worry. Yeah, we're going to talk about all the new stuff here in a second. But I'm going to first give everyone a very high level recap of Notebook lm. All right. So it should be in your show notes if you're listening on the podcast. But go back and listen to episode 370. Did a complete deep dive again. That time pun was only half intended because I use deep dive a lot. All right, but here's in episode three 70. So you can go to your everyday AI.com, go listen to that. But essentially this notebook LM is both very similar to other kind of AI chatbots that you've used. And it's also completely different. Okay, so it is made by Google. Let's start there. And let's start by clearing some confusion. You might be thinking like, oh, what's Notebook LM by Google? I thought Google had that Gemini chatbot. So yes, it does. All right. And let me also be very honest, I'm not a big fan of the front end of Gemini. So when I talk about front end, that is, you know, if you're an everyday person, that's what you're doing, right? When you go to chat GPT.com or gemini.google.com and you kind of, you know, log into your account, whether you're on a free or a paid account and you use the quote unquote chatbot. I'm not a fan of Gemini on the front end. I think the backend is fantastic. Google has their AI studio. It's probably one of the better backends, right? You know, Chat GBT as an example, there's a front end, Chat GPT is the leader there, the back end, it's great, right? The, the assistance API, the playground, all of that. So I'm talking about front end here. So this is a new front end product. So what that means is with hardly zero skill, you don't need any skill, actually, you can go in, log in on the front end of Notebook LM with a Google account. So you can have a free Gmail account or maybe your company has a Google workspace, all right, you can log in, it's free right now, all right? And it's grounded. Okay? So I'll give you an example of what that means. But this is huge. And I think everyone is talking about these AI audio overviews, these deep dives, and that's really what's stealing the show. But what I think is when you really even want to know the difference of like, hey, what's the difference between Google Gemini and Google Notebook lm? Well, Notebook LM is grounded, okay? And it does so through. This isn't traditional rag, right? Rag was kind of the, the keyword, you know, the buzzword in AI a year ago, retrieval on meta generation. So that's essentially before your query goes to a large language model like Gemini 1.5, right? That's what runs NotebookLM is Gemini 1.5. So Notebook LM is a grounded model. So that means if you just go in and if you don't add sources, if you don't add information, you literally can't use it, right? Which is huge. This is huge, right? And this is, I think addresses one of the biggest problems out there with large language model and generative AI right now is it's trust, it's transparency, it's explainability, it's this black box of generative AI. Notebook LM is so simple, right? If you don't give it source information, it can't do anything. You literally can't use it. So it is grounded in the information that you provide. And you can provide information in a lot of different ways. You can copy and paste text in, you can upload YouTube videos, you can upload files off your computer, like a PDF as an example. You can connect to Google Slides, Google Docs, you can upload an MP3, which I like, right in that piece right there. I gotta call this out, right, Because a lot of people are saying, oh, it's, it's, you know what, if you're using copyrighted material, right, you can't have your cake and eat it too and then judge people who are eating the cake. So here's what I mean by that. If you are an advocate or a user of large language models, you've already essentially, right, which is everyone in the world, right? You can't not use AI. It's coming to your devices whether you like it or not, right? But the reality is the way large language models are work work is they scrape content every, almost every single thing that's ever been on the Internet. Copyrighted works, pieces of media, offline data sets, it's all in these large language models. So, you know, I, some people are like, okay, I love, you know, I love AI, I love large language models. But, oh, I'm not so sure about, you know, uploading content into Google LM, copying and pasting an article or using a YouTube video. That's not mine. Guess what? That's how all large language models are trained. So like I said, you can't do something with your left hand and then judge your right hand for doing the same thing. All right, let's just put that out there. And also, are there, is, is there a further discussion to be had on this? Absolutely, but that's not what we're getting into today. All right, so the way NotebookLM works is you upload all of your information, then it is grounded, it is a, a type of retrieval, augmented generation, and it's different than most chatbots. Here's, here's one of the reasons it doesn't save your ch, right? Which I think there's, there's pros and cons, but probably for most users, more cons than pros. So it works a little differently. And I'm going to show you, I'm going to show you a little bit of, on how that works here in a minute when we do this live. But you need to save things as notes. So it works a little different. Which I think, like I said, it's not necessarily a bad thing, it's just a little different. Also, the context is huge, is huge. So you can have 50 sources of up to 500,000 words a piece, which is 25 million words of content context, right? Which is mind boggling. All right, this, right now, the fact that this is still free for anyone, it's Wild to me, because, like I said, I think Notebook LM is a genius product. It solves so many problems, right? Large language models, hallucinating, not being able to easily work, large amounts of your data. Right? It solves those problems also. Right now. It's free for now. All right? And we have audio overviews. And that's really what has blown up the popularity of Notebook LM in the last couple of weeks, when last month in September, they announced audio overviews. These were essentially. It's an AI podcast where two AI hosts, you upload all of your documents, and then they essentially do a, you know, short, anywhere from 3 to 12 minute, sometimes 15 minute podcast on your content. And it sounds very human, very cool. All right, so what's new? So that's. That's what's been out there. And y', all, if you're not already reading our newsletter, we told you about Notebook LM, like 6 months ago before all this hype, right? So if you follow along with us every single day, you're going to be so far ahead, even of the people who are, quote, unquote, far ahead. Just putting that out there, right? That's why we do this every day. And we have receipts. I can be like, y', all, I told you about this way before. Way before. It was on CNBC at night or, you know, msnbc. Right? Because now literally everyone's talking about Notebook lm. So here's what's new. So it's out of beta. That's a small thing. So it's no longer an experimental product. They got a new logo. So it's. It's looking. Looking a little polished. Even a new landing page. Right. Also, there's some new features. When it comes to the audio overviews, I think that's the big thing. So, you know, people might call these audio overviews. They might call them the AI podcast. They might call them the deep. The deep dive. Right? So those three terms might be using them interchangeably. But now you can listen to this in the background and then continue to work within your notebook. All right, and again, I'm going to show you guys live what that looks like and how it works. The big new feature. Oh, my coffee's not good this morning. That's a bummer. Normally I'm. I'm over here sipping on coffee. Do. Do any of you guys. I'm just curious, do you guys have, like, a morning routine? If you try to listen, you know, once or twice a week to the show, let me know. I'm always curious. All right. So the big thing here, I think, is customizable audio interviews. All right, so this is kind of also, I think what made the product go viral is because a couple of weeks ago, you know, some smart prompt engineer out there essentially just uploaded one file, right? So like I said, you have to upload a file in order for NotebookLM to work. And what this person did is they essentially uploaded a. I won't say it's a, it was a jailbreak or anything like that, but they essentially just uploaded one file and it was essentially script instructions to these AI podcasts. Essentially, like saying something like, hey, guess what, you're an AI. And this is weird, right? React to it, you know, and then they had this almost crisis, right? The, the AI, the AI podcast, the, the audio overview deep dive was like the two AI hosts, you know, going through an existential crisis, like, oh my gosh, we're AI. I tried to call home, no one was there. What's going on? Right? And then I think internally, Google caught wind of this and they're like, wait, you should be able to steer these podcast hosts a little bit. Because before, which I don't think was. I. I think there's pros and cons before you just put on all your information. And then the, the two AI hosts just handled it however they, quote, unquote, saw fit. And I know that's weird, right, to give agency to AIs, but that's what happens, right? The, the, the, the AI within Google, Google's Notebook LLM decided how they could, you know, what they would make of your podcast, right? Whether you had 50 words in there or, you know, 5 million words. Essentially, Google decided, or NotebookLM decided how and in what format they would create your audio overview. So now you can customize it. So that's, that's a really big new update here that came out. And then last but not least, a couple other things that they're testing. This isn't out yet, but they're potentially looking at a mobile app. And then they're also looking at the possibility to add more than two hosts or additional hosts. All right, pros and cons on that. And then the other big thing here that they announced yesterday is Notebook LM for business. All right, this is not out yet. It's a pilot program right now. And you can, if you want, you can join the pilot. So presumably that means there's going to be pricing, because they talk about pricing. All right, so we assume that this is going to be used for teams. Google says that right now, more than 80,000 organizations and millions of users are using NotebookLM. And then there's going to be a Business Edition. Right now, we don't know a ton on what is going to separate this Business Edition from the edition that you're or from the free version that is already there. And we don't know will this still remain free after a Business Edition launches? We're not sure. Right. So all we know is that Google and NotebookLM is really pushing this centralized knowledge base for your team across, you know, marketing, lnd sales, customer service, et cetera. And the use cases are obviously abundant and they're low hanging fruit because you've probably tried uploading your data right into the front end of Google Gemini. Not very good. You probably tried uploading your data into the front end of Chat GPT, maybe using a custom GPT. Okay, but still not great. You probably tried the same thing in Claude, using CLAUDE projects a little better, but still not good. Right? No matter what kind of path you've chosen, you're always going to deal with hallucinations because the way that most models work is they're not grounded. So they might first check their own internal data, they might browse the web, they might use your data that you've uploaded, right? So it's kind of a mix, right, of how they retrieve the knowledge. Right? And because large language models are generative, it's going to do it a little different every time. So even when you're uploading your files, your knowledge base for your company into Claude, into Google Gemini, into Chat gbt, you know, whether you're doing kind of normal chats, whether you're building GPTs or projects or gems, in Google, results aren't always that consistent and aren't always that good. And there's hallucinations. And I think that's the problem. Problem that Notebook LM solves. All right? Also, both on the personal and the business version, Google does say this. They say, we value your privacy and never use your organization's data to train Notebook lm. All right? It says as a Notebook LM business user. They also say the same thing for personal users. Your data will stay private to you and whoever you share your notebook with. So let me just get the elephant in the room, right? People are confused. They're like, okay, well right now, if it's free, if you can upload all of this data, right? Essentially this is a state of the state of the art rag model, which is wild and it's free. If you would have told people a year or two ago that this would exist. Companies literally, in early 2022, you know, mid-2022, right around the chat GPT moment, companies literally spent 2, 3, $4 million building a version of something like Notebook LM. But it was way worse and it required a lot of time, a lot of money, a lot of resources. Now it's free, right? Like I said, you would have told people two or three years ago that this exists. They would have said, heck no, there's no way it exists. It's amazing. I talk about AI all day, it's fabulous. It's like I said, I said. I think I said this on, on the first show that I did on Notebook lm. For me, Notebook LM was more of an aha moment for the power of AI than the initial version of Chat GPT was, right? And you all know I love Chat GPT. But when the first version of Chat GPT came out, November 2022, I thought it stunk, right? We had been using the GPT technology through dozens of other AI tools at that point. And it was better, right, that OpenAI's technology was better in all of these other tools that had been out for, you know, a year or more. So when Chat GPT came out, I was like, not that good. When Notebook LM came out, I was like, wait, what wait, Zero tech, skill, free, accurate, grounded, right? Bonkers. So why did Google make Notebook LM then? Right? They have Gemini, their front end chatbot and they've spent a bajillion dollars marketing it, right? Super bowl commercials, it's there everywhere on the Internet, right? Like everyone knows about Google, Gemini. So why would they make what seems like, you know, not a competing product, but something that's kind of similar, right? Why would they do this? Well, here's one of the reasons and why it's free. Why I think, right, it's all about users, right? Right now, I don't think the big companies out there, so your Googles, your, Microsoft's, OpenAI, Anthropic, Amazon, etc, they don't care about money. They don't, they care about users, they care about retention, right? It's literally a gold rush right now, all right? And think, you know, oh, everyone went out to California in the gold rush, right? Everyone went west. Think of it, it's like different states competing. It's like, hey, come get this gold. It's free. Come get your gold. We just want you to settle here. We want you to stay here. And so far, OpenAI and ChatGPT have been crushing everyone else when it comes to hey, stay here. Right? All these companies make, you know, all this stuff available for either free or 20amonth, which is crazy, right? This is worth hundreds or thousands of dollars a month in terms of what generative AI and large language models can do. When you combine it with your data and. And you have a little bit of skill, it is literally worth thousands of dollars a month. But for the most part, everything's free or $20, 20, 30 bucks, in my opinion. This is why Google has started to shift and invest resources into Notebook lm. All right, so I'm sharing a little graphic on my screen here. This is data through June 2024. And this is essentially traffic, right, amongst chat GPT and then its rivals. So in June, chat GPT had around 3 billion web visits. Billion with a B. All right? And then if you combine. Combine Google, Gemini, Perplexity, and Claude, combine those three, they did not even get 1 billion combined. All right? You don't need to see this chart, essentially, you know, and then Character AI is another, you know, kind of widely used AI system. Right? All of the other biggest ones combined are not even a fraction of Chat gbt. All right, so in my opinion, that is. I mean, companies know this, right? Companies know this. They like, they know if they're winning or losing. And right now, even if they're winning, even if Gemini and Claude are increasing their user base month over month, ChatGPT is on a meteoric rise, right, in terms of being able to attract new users and retain new users. Right. Like I said, 3 billion versus the other biggest ones combined are not even at 1 billion. So in my opinion, people are always like, okay, well, why. Why did Google launch Notebook lm? Well, I think they're taking Smart Move. They're taking a completely different approach to how a large language model can work and the business value it can offer. Whereas I think, you know, Gemini is. It's, you know, much more steerable. It can be much more creative. That's not Notebook lm. Notebook LM is grounded in your data, which I think companies finally realized probably about a year and a half ago, where it's like, okay, cool. Large language models are cool. But what companies really want to do is they want to get their data in there and they want to actually set up some strict guardrails in terms of working with their data. And that's what Notebook. I think that's what Notebook LM right now accomplishes far better than anyone else. Right? And y', all, I spent thousands of hours inside all the large language models, so don't come back to me and say, oh, Jordan, have you tried Claude projects? Yes, yes. Hours a week, right? Hours. I try all these Google Gems, cloud projects, you know, open AIs, GPTs, and I teach, I've taught thousands of people how to use all this. Notebook LM is in a class of its own right now. All right, let's get crazy. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Genai. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for chat GPT training for thousands or just need help building your front end AI strategy, you can partner with us too. Just like some of the biggest companies in the world do. Go to your everyday AI.com partner to get in contact with our team or you can just click on the partner section of our website. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on gen. Let's get crazy. This could go awful. I need some suggestions, y'.

All.

And we're going to do this live and we're going to show you these new features in Notebook lm and at the same time, we're going to see a little bit. All right, so I need everything from our live stream audience. If you didn't get anything already, please get some ideas and I'm going to go through them right now. All right, so I need a common skill, right? Something that you would try to, try to learn. And that's what I think Notebook LM is so good at. And then a common group of people. All right, so let's, let's go ahead. I want to look through some of the ideas so far and please get your idea in here quickly. And then we're going to do this, we're going to do this kind of live. All right, so let's, let's read, let's read some of our examples so far. All right, so Angie says a group of people, college students. I like that Christopher saying, you know, the group of people, writers, poets or other types of artists juggling for Retired people. I like that one, Rachel. That one made me laugh. Okay, several people at the retirement center learning to play cribbage from Angie. Oh, I like that idea, Angie. My grandparents played cribbage, so I kind of like that one. And I've always wanted to learn cribbage. I think I learned it when I was like eight, but I forgot. Sabbatical life from YouTube says learning French. Common group of people, churchgoers people. Project managers. Okay, let's maybe combine some of these. So let's do learning to play cribbage for project managers. All right, so we're going to combine, we're going to combine here. All right, so here's what we're going to do, y', all, and apologies, I'm going to try to keep this tight, all right? But we might go off the rails, but hopefully you'll learn something here. And podcast audience. I'm going to be sharing my screen. Hopefully I'll be able to describe this enough, right? You see, y', all, I use Notebook LM all the time. Look, in the past two days I'm sharing my screen right now. Just in the past two days, I've already used it like eight times. I use this non stop, right? I use this to prepare. I use right kind of meta. I use NotebookLM this morning and last night to prepare for my Notebook LM show today. All right, so here we go. So here's what I'm going to do. I'm going to go into new Notebook, okay? And you'll see the first thing it says is add sources. So this is what I was talking about earlier. You can't just go in and start using this. It's grounded in your data. All right, so I'm just going to show you an example. Under here I'm just going to paste in some text because if I X out of this, I literally can't go down here and use it, right? So NotebookLM has a familiar interface, but on the left hand side is your sources. This isn't your chats. Okay? So this is a notebook. So when you go on a notebook, on the left hand side is all of your sources that you can upload and you can always add and delete them as you go along. And then on the bottom, this is where you will will chat with your sources, essentially. And then the other thing, chats aren't saved. So when you exit out of the page, all of your chats are going to be gone. That's important to know. So if you get something very useful that you want to save, you will add it as a note. All right? So anyways, I can't go down right now and start typing. I'm clicking down here at the bottom. You literally cannot use NotebookLM until you upload data. So let me just give you an example. I'm going to go copied text, okay? So the different ways you can copy, you can copy and paste text. You can put in a link to a website or a YouTube video connect Google Docs, Google Slides, or upload files such as PDF, text, markdown, audio. So I'm just going to go to copy text. I'm going to put in test, all right, Just to show you guys how this works. Okay, so now you'll start to see some of this information fill out. All right, so now I can start typing. So I'm going to say, tell me how to play cribbage. Okay, so here's what's going to happen. It's going to have no clue, even though in a normal large language model, this is in its training data. But again, Notebook LM is grounded. I know everyone wants to talk about how cool the AI voices are. No, you guys are sleeping. You're missing out, I think, on the most valuable thing in AI right now is a grounded large language model that has RAG built in a state of the art model because it uses Gemini 1.5. That only works in your data. So the only sources I put in there, I put in the word test, right? So then when I say, tell me how to play cribbage, it's probably just going to say, yo, can't do that. It's not in. It's not in my data, right? So I say, tell me how to play cribbage. And it says, the sources you provide do not contain information on how to play cribbage. Okay, so here's what we're gonna do. Here's what we're gonna do, y'. All. All right, so I'm in YouTube now. I'm gonna say how to play cribbage. Okay? So for a YouTube video, it needs to have captions enabled. And again, I'm gonna say this one more time because people are gonna be like, oh, Jordan, you're using, you know, you're using information on the Internet that, you know, you might not have access to. Guess what? It's all in large language models. It's all in large language models. All right, Just putting that out there. So like I said, you can't do one thing with your left hand and then judge your Right hand. All right, so I'm going to copy the link address to this YouTube video. All right? And hey, can. Can y' all let me know? Can you see. See my screen? I hate. I hate doing these things live. All right, so now we're jumping back in, and we'll see if this video is allowed. So I'm going to YouTube now. So I'm going to add a source. And I have the first video that came up when it said, here's how to play cribbage, I'm going to insert it. It may or may not work because it needs. It needs a text transcript, and also it needs to be a public YouTube video. So we might have to try a couple. This might not even work. So let's see. All right, perfect. The first one actually worked. All right, so there we go. We have a YouTube video. Okay, so now I'm gonna do something else. I'm gonna go how to play cribbage. I'm gonna bring in a couple other sources. Okay, so now I'm on the Internet. Here's some website. Okay, so from bicycle playing cards, I'm gonna copy this, this URL, and then I'm gonna jump back into my notebook. I'm gonna add a source on the left hand side. I'm gonna click website. I'm gonna put that website in there. This may or may not work. We'll see. All right, cool. It worked. So now we have two sources, you know, and you can rename these sources. So I'm going to say I'm going to rename this to bicycle, so cribbage by bicycle, so I can remember. All right, there we go. And let's do one more just so we can see that you can pull in information from all different places. So now I'm going to copy and paste some information. I'm going to go one. One more. So here we go. Here's this beginner's guide to cribbage. All right, cool. Here we go. I'm going to copy and paste all this information. So this is from Instructables. I'm going to go to sources, add new. Go to copy, text, paste it in, insert it. All right, and then over here on the left hand side. This is important. Name your sources. All right, so I'm going to rename this one. And this is Instructables Cribbage, y'. All, if there's one thing I hate, I hate typing live on the show because the way my desk is set up, I'm like, crammed And I got my little T. Rex arms. Right, okay. And I'm actually going to do one more thing. I want to provide a little bit of information to Notebook LM on what a Project manager is. All right, so I'm going to go to Project Manager. So I'm just saying, what does a Project Manager do? All right, here we go. Haha. I'm getting a Google AI overview. All right, let's do this. Let's go to what does a Project Manager do? Here we go. From teamwork.com I'm going to grab this URL, I'm going to go into our notebook. I also am going to name our notebook.

Okay.

To cribbage. I'm going to add this one more source as a website. Let's see if this one works. Perfect. All right, so now we have four sources. Okay. So we have some copy and paste text, we have a couple of websites, and we have a YouTube video. Okay, so now let's go ahead. I'm going to generate this right away. Okay. So I'm going to show you because I'm going to hopefully play the difference so you can see how these, this audio overview works. So right away I went ahead and started to generate the audio overview. All right. And this one's going to take a couple of minutes. It might take, you know, three to five minutes. So I probably should have done this a little sooner because I also want to show you all how you can kind of change this and you can influence it. So the way that Notebook LM talked about this is they said it's like passing a note, right? So it's like passing a note. So think, you know, you have your two AI podcast, you know, hosts, and this new customize feature is like passing them a note on what they can or can't do. All right, so this is why at first I want to show you kind of the, the default podcasts that will happen. Because right now it's just going to use all of the different information in there and it's going to use its own kind of algorithms to determine what should the audio overview actually be. And this one might not turn out the best. Why? Because the source information is a little mixed. Right. It's going to see three sources on how to play cribbage and then it's going to see something about project managers. So we're probably not going to listen to the entire AI overview because it's probably going to be a little, it's going to be some mixed results. Right? It might be, it might still work. It might kind of understand. Okay, well, it looks like they're trying to learn cribbage, and it also looks like there's some information about project managers. So maybe they're trying to teach project managers cribbage, but maybe not. All right, so here's what we're going to do next. We're going to let that go. All right? We're going to let that kind of warm up here. It's going to take a minute, and then let's go ahead and start typing up. We're going to start typing up our customized instructions. So we're going to do a test. All right, again, so we're passing the note here. All right, we're passing the note, and I'm going to do my best to keep my eye on our notebook. And once the AI overview is done, and you know what, we do this live, y', all, it can be all over the place. So if you have any questions about, about Notebook lm, get them in now, because when we regenerate our second podcast, I might have three minutes to kill. All right, so I'll, I'll try to fill it in with some facts and show you around other parts of, of the notebook and some other features which I think are great. But you might as well get your questions in live stream audience, and hopefully I can help get some education going as we, as we go along. So let's work on some of these instructions and I want to test them. All right? And I'll say this, I did a, a quick reaction video yesterday. And I'll say this, working in text, very, very rarely get any hallucinations. Okay. And previously, when you left the AI hosts to their own accord, right? They were very accurate for the most part. So I did find maybe this is some bugs yesterday, but I actually got some hallucinations in the Deep Dive audio. All right, not in the normal text. Right. That was still 100% super stellar. So I don't know if by allowing users a little more freedom to steer the AI audio overview, if this is causing maybe some problems with accuracy. So as an example, yesterday I did the video. I'll just tell you the quick takeaway. What happened was I uploaded a bunch of different sources. One of them was avm, which was Advanced Voice mode. And all of that was the transcript of, of a podcast that I did on Chat gbt's Advanced Voice mode. And I asked the audio overview. I gave it some custom, you know, some of those custom instructions to customize the Deep Dive. And it made A lot of stuff up. It said AVM was an AI company and it said, oh, it's working with agents and it's not Right. So it struggled, right? It struggled a little bit. So, you know, always keep that in mind. You always need human in the loop and just because this does great and it's a hunt. In my experience, I've very rarely seen any hallucinations in text. Right. But with this new introduction of kind of this steerable audio overview, at least in my limited use case, I mean, it hasn't even been out for 24 hours. It kind of went off the rails just a little bit. All right, so luckily we now have our. Our first round, so we're gonna do our second one pretty quickly here. We're not gonna listen to this whole thing. It created a 12 minute podcast. All right, I'm gonna go ahead. Sorry, I'm gonna put this on like 1.5 and we're just gonna listen just to a little bit live stream. Audience, help me out here. Please let me know if you can hear this audio. Sometimes streamyard, which I use, doesn't stream the audio. So Here is the 12 minute podcast just a little bit on cribbage. So let's see how it handles this information. Here we go. All right, awesome. Cool. You guys said no audio. Love when this happens, even though I test this all the time. Let's try it one more time. All right, here we go. Share the audio. Share the audio. Let's try it again. All right, take two. Let me know, y'. All. This is. You know, people always say that. They're like, jordan, do more demos, do more demos, y'.

All.

Do you know how hard it is to do demos live when technology hates me? All right, here we go. Let me know.

Ready to up your game night?

Because today we're taking a deep dive into the world of cribbage.

You've shared some fascinating sources with us.

Rulebooks, strategy guides, even a visual how to.

So by the end of this, you'll be well equipped to impress your friends with your cribbage skills.

What's fascinating about cribbage is that it's not just a game of cards.

It's a game steeped in history and strategy.

It's like a blend of poker face.

And mathematical puzzles all wrapped up in a charmingly old school package.

I love that poker face in mathematical.

Speaking of old school, our journey starts back in the 1600s with a poet, Sir John Suckling, who apparently had a penchant for gambling and inventing Card games he did.

And quite a legacy he left behind.

Imagine a smoky pub, tankards clinking, and.

This poet gambler introducing his new creation.

Cribbage, which actually evolved from a game called Naughty Naughty.

It does have a certain ring to it.

Maybe Sir John.

Okay, so I ramped up the speed there. So you'll see it didn't do anything with the project managers. Right. Which is fine. It's actually probably better. It kind of probably realized, oh, okay, one of these things looks out of place. All right, so now here's what we're going to do. And I was adjusting the speed a little bit. Here's a little. Here's a little tip, y', all, right? Now, you cannot duplicate notebooks. So you might spend a ton of time getting all this, you know, everything in a notebook and all your different sources delete, Right? People don't know this. You can delete the audio overview and then essentially just regenerate it. Okay, so now I'm going to go to customize, and we're just going to customize this slightly. So I'm going to say, please start the Deep Dive audio. So again, this is the. The new feature. Everyone's talking about this, you know, kind of passing a note to the host. So I'm going to say, please start the Deep Dive audio overview with. A intro to project managers and why they should learn cribbage. I'm gonna say, please start with a shocking fact that grabs attention and a funny joke. All right, and then I'm going to say, please have the host be. I'm gonna say witty. And then I'm going to say, please have the host avoid platitudes, truisms, cliches, and generalities. Do not use space fillers or needless jargon. Make sure all words mean something. And then I'm going to say, please end the Deep Dive Audio overview with. I'm going to say an analogy. No, I'm going to say with a. I tried this yesterday. It didn't really work. So I'm going to say with a pirate joke that resonates with project managers trying to learn cribbage. So this is incredibly difficult. So I know it's going to take a second, so let's just go straight in. Let's start. Let's start generating this. All right, so now when I go to customize, I can just paste in all that information. All right, cool. It's not letting me generate. It might just be too long. Let's see. Okay, so I'm going to delete this. So I'm going to have to delete that part about having the host avoid platitudes, cliches, all that. All right, so now it's generating. All right, so now what I'm going to do is I'm going to show you some of the other features.

If.

If there weren't any questions. Let me just double check, Let me double check here. Okay, so Leslie asking, is there a current limit on how long the podcast can be? Great question, Leslie. So you don't have a ton of leeway in the length. So you can try with the new customizable instructions, I try to make some shorter. I'd say generally depending on how much source information you have, it's going to last anywhere from like 6 to 15 minutes by default. I tried with this customized customizable feature yesterday. I, I was trying to create a shorter one. I got it down to like three minutes. So I don't think as an example, you can create an hour long podcast off of, you know, 10 million words. I haven't specifically tried to make a very, very long podcast off a lot of words, but, you know, who knows, maybe, maybe that's what I'll do next week. Tara says it would be amazing if Notebook LM would save these as curated by blank, similar to perplexities pages. Yeah, it's actually a great point that Tara brings up and I have mixed feelings about this because you have these, I'll say this, you have these very visible and influential people in tech now who are essentially just releasing podcasts right, on Spotify and everywhere else and not very well done. You know, they're just, you know, releasing these, you know, 10, 15, 30 part series using AI and using these hosts. So, yeah, I wish, I actually wish you couldn't download these because right now you can download them, which is, you know, great use cases, right? So you can take it on the go. You don't need to be connected on the Internet. You know, I'm going to be doing some traveling, so I might download some of these and listen to them on, on the plane. But you know, the downside is now people are just spamming, right? Literally there are thousands of these podcasts now. And you can tell because it's the same two voice voices, but you essentially have, you know, popular, you know, quote unquote, highly visible people, people with large audiences just releasing this, which, I mean, like I said, there's good parts of that and there's bad parts. But I think unfortunately we're gonna get spammed. So if you are a podcast fan. You're gonna get spammed with a lot of these podcasts and I wouldn't be surprised if even some of the top podcasts, right? I just realized this because thanks to you all, Everyday AI is usually a top 10 podcast in the US on Spotify for tech podcasts. And there's like 90,000. And I saw the other day, there's a bunch of now just AI generated podcasts up there. And I'm like, and there people with like big followings and it's like, okay, part of it, it's like, okay, I can see that. But part of me is like, oh, man, that's, that's terrible. Right? All right. Anyways, we got our second podcast done already. Thankfully, I don't have to keep, you know, trudging on here. So again, as a reminder, I said please start the Deep dive overview with an intro to project managers and why they should learn cribbage. Please start with a shocking fact that grabs attention in a funny joke. All right. And then I said, please end with a pirate joke that resonates. So we're essentially going to see how, how much can you steer? How much can you actually steer this thing? All right, so let's go ahead, take a quick listen and again, please let me know, audience, if we can hear this. Here we go.

Ever heard of a perfect 29? It's a cribbage score so rare you're practically guaranteed to see a unicorn riding a unicycle on a rainbow before you ever achieve it in real life. But more on that later. We're diving into the wild world of cribbage today and you might be thinking, cribbage. Isn't that a game for, well, let's just say folks who enjoy a leisurely pace?

That might be the common perception, but you'd be surprised.

Exactly. Because today's Deep Dive is for all you project managers out there juggling deadlines, resources and expectations like a dos. We're going to show you how this centuries old card game is practically project management boot camp in disguise. We've got our trusty Cribbage by bicycle guide, a how to play cribbage video by gather together games, and even an instructables article for those who like to.

Learn by doing quite the stack of sources. What's fascinating is that while cribbage might seem like a simple card game, it's been around for centuries. Cribbage by bicycle tells us it evolved.

All right, so again, I, I cranked the speed up there so we can double check. And let's just listen to the beginning one more time. Because it actually hit our what we told it to. We said, hey, start with a funny joke or a shocking fact, the combination. And then talk to project managers on why they need to learn cribbage. Let's listen one more time. Just the first couple seconds.

Ever heard of a perfect 29? It's a cribbage score so rare you're practically guaranteed to see a unicorn riding unicycle on a rainbow before you ever achieve it in real life. But more on that later. We're diving into the wild world of cribbage today and you might be thinking, cribbage, Isn't that a game for, well, let's just say folks who enjoy a leisurely pace?

That might be the common perception, but you'd be surprised.

Exactly. Because today's Deep Dive is for all you project managers out there juggling deadlines, resources and expectations like a dos. We're going to show you how this centuries old card game is practically project management boot camp in disguise.

That's wild, right? It hit it right. The instructions are a little limited, so you can't put in like pages. I'll have to do a character count, but it looks like you could do about maybe 100, 150 words of instructions. So pretty impressive. Let's see if it ended. So I said, please end the Deep Dive audio overview with a pirate joke that resonates with project managers trying to learn cribbage. Very specific. All right, let's see if it. I'm gonna go to the last maybe 30 seconds here and let's see if we can pick it up. And again, I'm playing this between 12 and 15 speed just so we can get through it.

You might even discover a hidden talent for counting by 15s.

And let's be honest, after a long day of daunt charts and stakeholder meetings, a little cribbage could be the perfect way to unwind and flex those mental muscles in a whole new way?

Indeed. Though, as you play, I challenge you to consider this. Do your cribbage strategies reflect your project management style? Are you a bold player going for big risks and bigger rewards? Or are you more calculated and methodical in your approach?

Oh, that's a good one. It's like holding up a mirror to your decision making style, but in a fun, low stakes setting.

And hey, even if you don't become the next cribbage champion of the world, you'll gain valuable insights that you can apply to your projects and maybe even impress your team with your newfound card playing prowess. So that wraps up our deep dive into the world of Cribbage. We hope you've enjoyed this journey through.

The history all you project managers out there.

All right, I didn't see it in the last like 30 seconds.

I have one question. Are you ready to add a secret weapon to your project management toolkit? Cribbage might seem like an unlikely hero, but trust me, it's more than just a game.

It's a fun, engaging way to sharpen the.

All right, I'll have to go through and listen to this the, the full thing here, but I didn't on first glance see the ending pirate joke. So maybe it kind of lost track there. But I'll make sure to go ahead and recap this all in the newsletter. So let's wrap this one up and live stream audience, what did you think? And I hope I don't need to explain this to you, right, But I hope you can already see the enormous potential, right? I think an easy part is learning new subjects, right? So let's say you're a project manager and let's say you need to learn about a new client, you need to learn about a new skill set. There's a new technology in your field and you're a project manager, right? Being able to upload that information. You know, you can have a source that's about yourself. You know, you could upload a source that's your, as an example, your disk profile, you know, so you can upload source information about yourself and then you can upload source information about whatever project, new project you're working on about a new technology, you're trying to learn about some new news that's impacting your industry. And then in the customizable instructions, you just tie it together, right? And then that's one thing AI is great at, is personalized learning. But when you can ground it in facts and stats that are relevant to you, that's when I think we, I think this is a bigger aha moment than the initial chat GPT because now this is your data and it's personalized to you. And like I said, you know, the, I think the customizable audio overviews, you know, there's probably going to be some bugs early on because it, when, I mean, when you think about it, it's actually wild technology. So you know, it had some troubles when I did some testing yesterday. It had some hallucinations. So this isn't something that I think right now you should be hands off the wheel and just trust that every single thing is going to be correct. So you know, you get still got to do your due diligence. Still needs to be human in the loop. Right. But at least when working with the actual information here. Very good. So I told you I would, I would quickly show you around the rest. I'm going to do like a 30 second overview. All right, so even when you listen to this thing aligns, you can continue.

To work resources, timelines, stakeholder binding.

You need a very simple combination generate all these different starter card and the other three fives and the jack of clubs.

I know that'll be hard to hear, but just to show you, you can listen to the audio overview and continue to use other parts of Notebook lm. So, so you can click and create these little notes. Right? So an faq, a study guide, a table of contents, a timeline. Right. You can create these all with one click. And then I can also go in here and chat. So I can say, you know, tell me the three most important tips. Tips for a project manager to learn cribbage. Okay. And then I click enter because I do need to show you kind of one of the most. Okay, interesting. So it says this question cannot be answered from the given source. Okay. So it actually can't combine technically, I didn't. In my source material, even though there's information about project managers and there's information about learning to play cribbage, this is a good example. There's no stats or facts in there specifically on how project managers can learn cribbage. So when I said tell me the three most important tips for a project manager to learn cribbage, it can't tell me. So now I'm going to do the same thing. All right, so I'm going to do the same thing and I'm going to say, tell me the three most important tips to learn cribbage. And now it should tell me. Let's see. All right, so there we go. So now it's saying, here are the three most important tips to learn cribbage. That's a great example on how it's not going to infer, it's not going to make things up when you are working on the text version. Right. When you are chatting with your notebook. All right, so the good thing is, is you can go and click the sources. So it's saying, you know, hey, the three most important things are under understand the scoring combinations Master the pegging system, strategize your discards, and then it has little citations for each one. So for the first one, understand the scoring combinations, I can click right. So here's citation one. And then it takes me to the source. This was the cribbage by bicycle. And then it highlights. So it takes me to that source and it highlights it. It. So you know, you can always have a level of confidence. I can click number two. That's in the same source. Let me click number five. So a lot of this is from the same source. That's okay. It's probably the best source, I'm guessing. So, yeah, at least in this question, it took everything from the same source. That's okay. So now if I want to save this, because if I come back, it's going to be gone. I can just click this save to note. All right, so then you can toggle your kind of notebook guide on, and then you can toggle the View chat or close chat. And then here's all of my notes, right? Here's all of my saved notes. So here's the one that I just generated, the faq, the timeline, you know, those are default options. And then the question that I asked, and then I saved it as a note. So now when I go back, you know, I'll still have all of my sources, I'll have my saved notes, but this chat history will be gone. So a little different than standard large language model. So I hope this was helpful. I know this was a long one, so if this was helpful, please let me know. It takes us many, many hours to plan shows that I hope are helpful for you and your career. So if you're listening, whether on on Twitter or LinkedIn, please click that repost or share button. Share this with your network. That would mean a lot. If you're listening on the podcast, please leave us a rating. Subscribe to the show and go to your everydayai.com Sign up for that free daily newsletter and we'll see you back for more AI in five. Thanks, y'. All.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going for a little more AI magic. Visit your everydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.