WEBVTT 00:00:00.000 --> 00:00:01.960 It's like having a personal tutor for your complex 00:00:01.960 --> 00:00:06.440 ideas, available anytime. Hey everyone, welcome 00:00:06.440 --> 00:00:08.880 to the Data and AI with Mukundan show. I'm your 00:00:08.880 --> 00:00:11.119 host Mukundan Sankar and today we'll be talking 00:00:11.119 --> 00:00:14.400 about Notebook LM and how it can be your newest 00:00:14.400 --> 00:00:18.500 and your most knowledgeable teacher. So I don't 00:00:18.500 --> 00:00:20.800 know if you've heard about Notebook LM, it's 00:00:20.800 --> 00:00:24.239 one of the newest products by Google and so the 00:00:24.239 --> 00:00:26.739 thing about Notebook LM is people have been using 00:00:26.739 --> 00:00:30.800 this for podcasting. By people, I mean content 00:00:30.800 --> 00:00:34.759 creators. However, what I wanted to talk about 00:00:34.759 --> 00:00:37.640 today was Notebook LM can be used for something 00:00:37.640 --> 00:00:41.320 more than podcasting. There's so much more to 00:00:41.320 --> 00:00:43.939 just generating like AI -generated podcasts from 00:00:43.939 --> 00:00:47.679 this tool. And that's basically the focus of 00:00:47.679 --> 00:00:51.340 this episode. So before I even go into anything, 00:00:51.579 --> 00:00:54.960 let me just talk about Notebook LM. What is Notebook 00:00:54.960 --> 00:01:00.000 LM? Notebook LM is this product by Google. It's 00:01:00.000 --> 00:01:04.379 a really cool AI -based product. So it's powered 00:01:04.379 --> 00:01:07.640 by a really AI -powered chatbot. It's a super 00:01:07.640 --> 00:01:12.299 advanced chatbot. And in the backend, it is using 00:01:12.299 --> 00:01:16.319 Gemini 1 .5. So it's basically designed for research 00:01:16.319 --> 00:01:20.280 and also for note -taking, but it excels at making 00:01:20.280 --> 00:01:26.189 written content into audio. So basically the 00:01:26.189 --> 00:01:29.390 key features of this product is uploading content 00:01:29.390 --> 00:01:33.709 from various sources like web links, basically 00:01:33.709 --> 00:01:39.629 URLs, YouTube videos, Google Docs. It can generate 00:01:39.629 --> 00:01:42.010 audio conversations that sound like real human 00:01:42.010 --> 00:01:46.250 dialogues, which I mentioned. And it can provide 00:01:46.250 --> 00:01:50.109 really insightful summaries and suggest questions 00:01:50.109 --> 00:01:54.599 based on the uploaded text or web link. So basically 00:01:54.599 --> 00:01:56.620 any kind of text that you upload, it will suggest 00:01:56.620 --> 00:02:02.780 questions as well based on that. So my initial 00:02:02.780 --> 00:02:06.260 impression when I first discovered this was I 00:02:06.260 --> 00:02:09.620 was amazed by its ability to transform blogs 00:02:09.620 --> 00:02:12.539 into podcasts. And yeah, I mean, that is a game 00:02:12.539 --> 00:02:16.439 changing tool for sure. So the way it works is 00:02:16.439 --> 00:02:20.360 you have two people, a man and a woman, and they 00:02:20.360 --> 00:02:23.939 are having a conversation. So in this conversation, 00:02:24.240 --> 00:02:27.419 they are talking about the piece of content that 00:02:27.419 --> 00:02:32.060 you wrote. And they are explaining it in a really 00:02:32.060 --> 00:02:39.219 easy to understand way. So basically, they're 00:02:39.219 --> 00:02:41.120 taking something really complex. In my case, 00:02:41.219 --> 00:02:44.280 it was a complex blog post. And it was turning 00:02:44.280 --> 00:02:46.800 it to something which is very easily understandable. 00:02:47.439 --> 00:02:50.460 So I thought that was amazing how they did that. 00:02:52.120 --> 00:02:54.460 Because I felt like, well, I couldn't have done 00:02:54.460 --> 00:03:01.960 a better job myself. And to maybe think about 00:03:01.960 --> 00:03:05.000 it in a more fun way, think about it like two 00:03:05.000 --> 00:03:09.460 of your favorite podcast hosts and they are, 00:03:09.580 --> 00:03:13.719 you know, having an espresso and just, you know, 00:03:13.719 --> 00:03:15.759 having a casual conversation. So that's what 00:03:15.759 --> 00:03:19.580 it sounds like. So, I mean, this kind of a game 00:03:19.580 --> 00:03:23.479 -changing tool, I think, would be, you know, 00:03:23.500 --> 00:03:27.039 where the podcast world is headed now. So speaking 00:03:27.039 --> 00:03:30.039 from my personal experience, so I uploaded a 00:03:30.039 --> 00:03:34.960 content about retrieval augmented generation. 00:03:35.199 --> 00:03:38.219 So it's like one of my blog posts and I'll link 00:03:38.219 --> 00:03:43.039 that in my show notes. So in this blog post, 00:03:43.159 --> 00:03:47.780 basically what I did was I had broken down digital 00:03:47.780 --> 00:03:50.819 augmented generation and how it can be used for 00:03:50.819 --> 00:03:56.460 news. So basically how to generate AI summaries 00:03:56.460 --> 00:04:01.819 and audio from the text. So for this particular 00:04:01.819 --> 00:04:05.819 news case, which I had, I used this blog post 00:04:05.819 --> 00:04:08.840 and I generated a conversation. So there's an 00:04:08.840 --> 00:04:11.139 option to generate a conversation, which means 00:04:11.139 --> 00:04:13.879 you can generate a podcast episode from a piece 00:04:13.879 --> 00:04:16.519 of text here. And that feature felt really intuitive. 00:04:17.319 --> 00:04:21.300 So what Notebook LM did in this use case was 00:04:21.300 --> 00:04:27.139 it created a podcast episode from my blog post. 00:04:27.500 --> 00:04:33.139 And it made retrieval augmented generation sound 00:04:33.139 --> 00:04:36.459 so easy. And any 10 -year -old could probably 00:04:36.459 --> 00:04:42.459 pick that up. So it just sounds like two experts 00:04:42.459 --> 00:04:44.680 are having a conversation. And they're making 00:04:44.680 --> 00:04:49.040 it very easy for even a 5 to 10 year old to understand 00:04:49.040 --> 00:04:52.560 the conversation. It's like a general conversation 00:04:52.560 --> 00:04:55.180 they're having and you get to be a part of that. 00:04:55.500 --> 00:05:02.800 So what I think makes Notebook LM unique is its 00:05:02.800 --> 00:05:06.540 audio creation capability. So basically you're 00:05:06.540 --> 00:05:10.000 converting text into naturally sounding dialogues, 00:05:10.100 --> 00:05:14.579 which sets it apart. And it's like turning your 00:05:14.579 --> 00:05:23.000 Word document into a really super interesting, 00:05:23.220 --> 00:05:29.339 super intelligent TED talk. And there's so much 00:05:29.339 --> 00:05:32.100 education potential as well with Notebook LM. 00:05:32.860 --> 00:05:37.180 Like I said, it makes a complex idea sound very 00:05:37.180 --> 00:05:41.420 easy. So any complex subject that you want to 00:05:41.420 --> 00:05:45.420 feed it, you can get like a very easy answer 00:05:45.420 --> 00:05:48.740 from it so it's like a supplemental learning 00:05:48.740 --> 00:05:51.300 as well in that process right like in addition 00:05:51.300 --> 00:05:57.279 to your youtube your um google chat gpt and all 00:05:57.279 --> 00:06:01.879 of that other um you know search tools basically 00:06:01.879 --> 00:06:04.339 so these tools they help you to understand complex 00:06:04.339 --> 00:06:07.759 topics as well right but this can this can be 00:06:07.759 --> 00:06:11.620 your essential like your additional um you know 00:06:12.569 --> 00:06:15.110 supplemental learning tool and that's something 00:06:15.110 --> 00:06:21.529 that's something really cool so my vision is 00:06:21.529 --> 00:06:24.790 just this basically just I know a lot of people 00:06:24.790 --> 00:06:27.350 are talking about using notebook LM for podcasting 00:06:27.350 --> 00:06:31.769 and that is amazing but what we should be also 00:06:31.769 --> 00:06:36.589 thinking about is using it for learning learn 00:06:36.589 --> 00:06:40.089 complex topics So any kind of topic that you 00:06:40.089 --> 00:06:43.110 read online, for example, you read a research 00:06:43.110 --> 00:06:47.910 paper and that is going above your head. It happens 00:06:47.910 --> 00:06:50.410 a lot with me. I try to read research papers 00:06:50.410 --> 00:06:54.250 to keep up to date in my field. But in your case, 00:06:54.550 --> 00:06:57.069 if you're kind of similar like me, maybe you're 00:06:57.069 --> 00:06:58.730 reading research papers, maybe you're reading 00:06:58.730 --> 00:07:03.250 blog articles or something else online, which 00:07:03.250 --> 00:07:07.939 you want to understand. take the content from 00:07:07.939 --> 00:07:12.339 that page and just upload it to Notebook LM. 00:07:12.439 --> 00:07:15.540 And you will be amazed at how it can convert 00:07:15.540 --> 00:07:19.579 that kind of complex information into easily 00:07:19.579 --> 00:07:24.459 digestible data. So I think that's really cool. 00:07:25.100 --> 00:07:28.300 And it would be very educational for you. And 00:07:28.300 --> 00:07:30.639 speaking for personal experience, it was for 00:07:30.639 --> 00:07:34.899 me. So I hope you can feel the same about it. 00:07:36.149 --> 00:07:40.029 Yeah, I mean, just using it for podcasting, yeah, 00:07:40.110 --> 00:07:42.170 I'm sure we'll have a world where we just have 00:07:42.170 --> 00:07:44.889 podcasts where just two people are talking instead 00:07:44.889 --> 00:07:48.029 of all the amazing content creators around there. 00:07:48.209 --> 00:07:51.029 So I hope we don't get to that place where we 00:07:51.029 --> 00:07:54.870 just have AI -generated podcasts. I still want 00:07:54.870 --> 00:07:58.910 to listen to actual humans, but maybe that's 00:07:58.910 --> 00:08:04.529 just me. But I would say this, that you know 00:08:04.529 --> 00:08:09.250 use it to benefit you just learn something new 00:08:09.250 --> 00:08:11.970 so when you're on the go you can use this to 00:08:11.970 --> 00:08:14.810 learn something so like i said since you're learning 00:08:14.810 --> 00:08:18.370 uh you know complex topics maybe you can generate 00:08:18.370 --> 00:08:22.329 a podcast but that's just for yourself um so 00:08:22.329 --> 00:08:25.069 just play it while you're on on the go somewhere 00:08:25.069 --> 00:08:27.730 maybe you're in your car just have it playing 00:08:27.730 --> 00:08:30.410 on your phone or you know it's connected to your 00:08:30.410 --> 00:08:33.610 multimedia player on your car, your Apple CarPlay 00:08:33.610 --> 00:08:36.850 or your Android Auto or whatever. So you are 00:08:36.850 --> 00:08:40.049 connected to that and you're listening to this 00:08:40.049 --> 00:08:43.470 podcast generated through Notebook LM. I'm not 00:08:43.470 --> 00:08:45.570 sure if Notebook LM is completely there yet, 00:08:45.610 --> 00:08:49.169 but maybe they'll have a product. Right now, 00:08:49.190 --> 00:08:53.250 I think it's in beta phase or experimental phase, 00:08:53.350 --> 00:08:58.210 whichever. But I think when you want to learn 00:08:58.210 --> 00:09:00.190 something, you should use Notebook LM in the 00:09:00.190 --> 00:09:05.250 future. So that's all from me today and I look 00:09:05.250 --> 00:09:06.950 forward to seeing you in the next one. Thanks 00:09:06.950 --> 00:09:07.190 everyone.