NotebookLM: Will Instant Podcasts Transform Learning?

--- type: transcript date: 2024-10-03 youtube_id: LrJxmVp_JVA --- # Transcript: NotebookLM: Will Instant Podcasts Transform Learning? in this video I'm going to be showing you notebook LM as well as the brand new illuminate experiment from Google notebook LM has caught a fair bit of attention over the past couple weeks on social media and for good reason I've been playing around with this Fair bit and it is pretty remarkable what they've set up one thing that's interesting with this release is they haven't really been pushing it they just put it out there and people have organically just started to gravitate towards the tool I think even Andre karpathy said something to the effect of this is like a chat GPT type of moment from his experience of using notebook LM essentially how this works is you can go to notebook lm. google.com and then you can create a new notebook the way that it works is you can just drop in your files here if you have anything locally whether it's a PDF or markdown or even audio files you can put them all within here you can connect it to Google Drive you can put in links even YouTube links if you want or you can even just paste in copied text and right now you can put up to 50 different things right now which is quite a bit of context that you can feed the thing I'm going to show you one of these preset examples that they have here within here you have all of the different sources that were added you can just check them out all on the sidebar here you can look into it you can see all the different key topics that it organizes for you now you have the ability to have a conversation with it which is something we've seen a ton of tools do already be able to upload different files and have some sort of rag chat with it you can ask questions you can say what year was the light bulb invented and then just like a chat GPT sort of experience it will just take a moment it will search through that context and then it will give you a nice succinct answer with sources you can see exactly the different pieces of text that it came from if you're just putting in all of these different random documents that you haven't necessarily even read before you could imagine how something like this could be really useful and versatile in learning really anything especially for like college or university students or even high school students this is going to be something where you can just put in basically whatever you want to learn about and it will give you a really good platform you can interact with it in addition to the ability to ask questions it does have the follow-up question similar to something like you'd see within perplexity which is really cool as well platform is where if you click the notebook guide once it's generated this notebook for you and you click load conversation what this will do is it's going to generate an MPR style podcast based on everything that you have within here essentially it is it's going to be a conversation between a couple different hosts and they're going to be talking back and forth if you've listened to NPR Podcast in the past this is definitely going to be something that resonates with you I'm going to go ahead and play some of the audio here I'm just going to play a little bit just so you get a sense on it I'll click play here it really is funny when you think about it flick a switch bam instant Sunshine no matter what time it is but getting to that instant light wasn't instant ATL and the stuff you sent over about the light bulb itself about Swan about Edison man it proves how anything bitty simple the whole thing was it's so true we're just so used to electric light now that most of us don't even think about what went into it and what's cool is what we've got here it lets us bust some myths wide open like that whole lone Geniuses thing people love to slap on invention yeah not so much totally it's like we're about to show everyone that making a light bulb wasn't just one a moment it was decades people getting close some serious competition and in the end it's actually a story about teamwork making the dream work 100% that's just a really quick example of the overall platform now the other great thing with this is you have the ability to share these as well next I'm going to be showing you illuminate and this is an extension of notebook LM this just came out and what this allows you to do is you can go over to the generate Tab and you can paste in a PDF so let's just say we go to artificial intelligence and we just grab a random PDF here so here we see a 12-page PDF and what we can do with it is we can past it in here now before you click generate you have the ability there's this tool tip if you hover over all of these you can select a beginner audience General audience or an expert audience you can determine whether you want it to be quick medium or long so less than 5 minutes 5 to 10 minutes or 10 minutes and then you can select the tone whether it's semi-professional or casual for instance you can just type out casual within there now at time of recording when I logged in it said it will give me 20 Generations per day and it only takes a few moments to set this up I already did this example previously I'll just again show you a quick look on what this looks like this conversation is powered by Google illuminate check out illuminate. google.com from hey everyone welcome back today's discussion is all about Babble bench a new Benchmark designed to test how well large language models llms can handle different types of data like images and tables all while using Code sounds complicated right let's see what it's all about what exactly is Babel bench trying to achieve imagine you want an L to solve a real world problem like figuring out the busiest time of day based on picture of rush hour traffic and the table of traffic data Babble bench creates these kinds of challenges it's not just about understanding words it's about understanding pictures tables and using code to combine that information and get the right answer wow that's like a really tough test that just gives you a really quick idea on what it can do it's pretty exciting that now we can have these podcasts that are erated on the fly on basically whatever we want you can imagine say a new model comes out and there's a really dense research paper that comes out say a new llama model comes out like llama 4 comes out next year what you could do is you could feed in a PDF that's maybe a 100 plus pages long it will do is it will synthesize the information almost as if a team behind the scenes read the research paper and then came up with a nice little polish podcast that you can listen to I'd really encourage you to try this out put in some information that you're trying to learn whether it's links or documents or what have you really throw whatever you want at it and you can even put in your personal data within it that's something that I haven't tried too much but it could be interesting right maybe if I fed it my YouTube videos maybe it' be able to give me potential ideas on videos or advice on the Channel that I might have there's just a ton of different use cases for something like this it's really neat to see these new creative tools coming out there and the great thing with this is right now as it's an experiment it's completely for free I don't know if this will remain the case or how long it will be free for but at least while free I encourage you to check it out I'll put all of the links within the description of the video and if you found this video useful please like comment share and subscribe otherwise until the next one