Is NotebookLM—Google's Research Assistant—the Ultimate Tool For Thought? - Ep.22 with Steven Johnson

[00:00] I want to find a scientific or technological idea that is central to the Apollo 1 fire. Here's the crazy thing, Dan. Maybe there's kind of a version of the Apollo 1 story that connects to that history of oxygen that I've already written about. Now we're getting exciting because we can take this and go back to your Readwise notes and try to pull some sources, right? [00:30] of an early enclosed environment that relied on a pure oxygen supply, similar to the Apollo 1 spacecraft. That's so cool. That's the opening. [00:52] Stephen, welcome to the show. Thanks very much, Sam. It's great to be here. [00:56] I honestly, I'm so excited to have you. You're such a great writer. I love your writing. And you're also kind of in this like interesting intersection where you're doing incredible writing and you're also working at Google. So for people who don't know, I feel like I'm jumping the gun a little bit. But for people who don't know, you are the best selling author of How We Got to Now and Where Good Ideas Come From and a bunch of other books, I think 13 other books. You just wrote a new book, which I have right here called The Infernal Machine. And if that wasn't enough, [01:26] LM and Google Labs, which is Notebook LM is Google's AI research product. And I'm so excited to have you here. This is really great. You know, we actually have some kind of news to break as well. So it's going to be really fun.

[01:40] Yeah, so by the time this podcast comes out, a new version of Notebook LM will be out. And we will be on the show today. We will be demoing all those new features. We'll talk about how you used it to write your last book and then talk about how we can use the new features to maybe write a new book or find a new book concept for you. So I'm curious, before we get started, talk to us a little bit about what Notebook LM is. Yeah, it's a tool that we've been developing for about the last two years. [02:09] at Google Labs, which is this wonderful new division inside of Google, that is really designed to build your ideal research and writing software, knowing that you have, from the very beginning, a language model at the core of the product. We really began this in the age of large language models, knowing that there was going to be a possibility to build entirely new kinds of [02:39] about like adding some ai to you know a word processor adding some ai to a photo editor this is about like what can you build that's like a genuinely new kind of surface to work with and this had been a like the the we can talk about a little bit but like the the obsession in my life with using software to help me with with organizing my ideas organizing things that i read and turning those ultimately into books or into tv shows or whatever uh it's just been a [03:05] obsession of mine for like 30 years or longer even really. And so Google's got very interested in this process of co-creating with people. So not just kind of sitting around and building products and then handing them over to writers or musicians, but actually having a writer or musician in the room from the beginning with the project. It's a big part of the labs ethos in particular. And so they called me up on day, you know, two years ago, and they were like,

[03:35] helping us build this thing you've always wanted. And that was an easy yes. And so, yeah, we kind of built a very early prototype and... [03:48] I think because I've been in part because the the technology was suddenly, you know, available with language models. [03:56] And because I've been sitting on these ideas for so long, we were able to build. [04:00] you know, a pretty quick prototype and that got a bunch of interest internally. And we launched Notebook LM in the U.S. in December of last year. And today we are announcing that we are rolling out to over 200 countries around the world. So we're really excited about that. That's amazing. And I just want to go back, like the thing that like really sparked my attention [04:30] and using those to produce produce work. Cause like, that's, I feel that same thing. I like, I'm a nerd for that stuff. And I just want to understand, like, tell us about some of those, what the, what those, uh, what that obsession is or what some of those ideas are, like, what are those things that you've always wanted that are kind of starting to come through here? Yeah. I mean, it actually dates back. I'm going to really date myself here. Um, which is that one of the things that changed my life truly changed my life was when I was a sophomore in [05:00] Apple released something called HyperCard. [05:02] which was this like crazy app. I always say it was a little bit like the Velvet Underground of software. Like it never really had a hit, but it like influenced all these other people. And it was basically a kind of prototype app.

[05:19] almost web-like hypertext kind of system where you could organize information pretty much any way you wanted and kind of make links between it. And I just got obsessed with the idea that I could use this tool as a place to kind of keep all my notes from my classes. And I kind of built this little application that I called Curriculum that was kind of a way of like taking notes for classes. And I spent way more time like building the tool than actually like using it to take notes for classes. And I kind of stopped going to the classes for a while because I just wanted to [05:49] It was one of those things. But it just gave me like a taste. [05:52] it wasn't ready you know in any way for kind of prime time use but i got a sense of the possibility and then obviously when the web came along i was i kind of jumped on that maybe a little bit earlier than some people because i'd [06:04] lived briefly in the world of HyperGuard and early hypertext. And then in the, about 20 years ago, there was a, there's a program that I wrote about a lot, actually, a wonderful program, way ahead of its time called DevonThink. [06:16] And that was it's still around, actually. It's a really cool application. And it enabled me to keep all these quotes from books that I'd read. [06:26] I would originally kind of type them in and then when the Kindle came out I could you know [06:30] get the quotes digitally and you could kind of make connections between [06:35] quotes or you could type something and say what quotes in my research library are related to this. We had this kind of associative, really kind of semantic search. [06:46] And I use that quite a bit on a lot of my books, like the ghost map, use that. And I would have these moments where.

[06:53] The software would recommend a quote that I had forgotten from an earlier book. [06:59] And it would make a new connection in my mind that I hadn't thought of before. And I thought... [07:04] This almost feels like a partnership with the software. Like I'm kind of I'm curating these quotes. So it's me and I'm and I know how to turn them into a chapter or a paragraph in a book. [07:16] That's my intelligence. But the connection was really made by the software. That seems kind of new and tantalizing and weird, but also maybe very powerful. And so that was another taste that kind of got me along that way. And then, you know, when I first started, [07:34] started experiencing what was possible with language models starting with gpt3 um uh you know before i came to google i i thought oh wait now it's all really going to be possible like all this stuff is it's going to get very serious and very real um so that's that's the prehistory and i'm curious like um like why do you think that's every like nerd's dream is like to have [08:04] Why is that so appealing? Yeah. Yeah. Well, and, and, and they're different, you know, they're different. This is true on so many levels, but they're different kinds of nerds. That's true. Like I do, I don't, the thing I've always felt like, and this is generally, this is true of the way I organized my like email as well, which is to say I don't. [08:20] spend zero time organizing my email like i am not a my principle has always been like create one place where you dump everything and then use you know smart tools like search and and now all these language models um

[08:34] to find what you need. Don't spend any time organizing anything. Just throw it all in one place and focus on having the ideas and stuff like that. And so, and I think Notebook, I mean, probably to a fault, Notebook LM has been kind of, [08:49] Deside a little bit with that principle. Like you can't like tag your notes, for instance. And, you know, we probably should have. People do like to tag things. And I just am always like, I'm not going to spend a second tagging anything because I want the software to understand what categories they are. I don't want to put things in advance into buckets because I want it to be an open-ended connective system where I can make new associations or create new kind of clusters on the fly. [09:19] that I can do. So, [09:21] So I'm on the side of emergent chaos, right? Don't organize it and just let things bubble up and figure out tools that will let that bubbling up happen. But then there's a whole other set of folks who really like to organize it and systematize it and have it all in these categories and things like that. And so hopefully we can ultimately make Notebook LM play well with both those groups. I think it's certainly within our power. Yeah, it's the top down versus bottom up folks. [09:51] So I'm excited to see Notebook. Let's roll into that. So give us a little bit of a tour of Notebook. Tell us about, because we're going to talk about how you used it for your latest book. So tell us a little bit about how Notebook works and then how you use it for the book that you wrote.

[10:06] Yeah, I'll just give you the kind of basics. [10:10] The idea is... [10:13] Everything inside of Notebook LM, [10:16] is grounded in the documents you provide. [10:21] We may open this up a little bit over time. I think it's probably a logical thing to do, but unless you provide notebook with what we call sources, kind of underlying the documents that are the source of truth for your project, the things that you're working on, and it could be everything from your journals to work documents or to research materials or quotes from books that you've read, you begin each document. [10:47] project by opening up a notebook and uploading sources. And at that point, once they're uploaded, the model [10:56] in a sense, kind of becomes an expert in the information you've shared. Now, this has become increasingly common. It was kind of a radical idea when we were first toying around with it two years ago. But the idea of having documents attached to a model like Gemini or ChatGBT has become increasingly common. But our use is like you're always working with documents and the whole interface is designed to let you basically load a lot of different documents, [11:26] potentially read those documents while you're working and not get into that kind of flow that so many people I think are finding these days where they have like 12 tabs open and they're like grabbing some text from one tab and then pasting it into the chat bot and another thing and then they're getting the output and they're saving it in another document like we want to have a single integrated surface where you can do all that work so in a sense designed to not interrupt your flow state like if you're thinking or writing or reading you should just have one

[11:56] that I had open where I had that quote that I wanted to use and that other thing. So that's the underlying model. And if those of you who are watching this, I have open here a notebook that this is kind of cool. This is my crazy notebook. This is where I have [12:17] Um, [12:19] all of the quotes that I have collected over the years, um, [12:24] for books dating back to something like 1999. I think it goes back to this. It's about 7,000 quotes. Oh my God. That I've collected. So it's really like my reading history, um, like the things that were important to my books in the past. And, uh, and so, uh, they're, they're lined up here as a bunch of different sources. This is, this is kind of leftover from the fact that, um, we used to have a kind of a cap on how long the sources could be. Um, so now you can have this, [12:54] a couple of weeks ago. In each notebook, you can have up to 50 sources and each source can be up to 500,000 words. So you can effectively be talking with, [13:06] 25 million words, um, in a single notebook, which is just kind of mind blowing. So these quotes are like, um, [13:14] They are, here I'll just open up one of them. They are, I think there's something like 2 million words total, 2 million, 7,000 quotes, 2 million words, something like that. This is crazy. Yeah. Okay, so these are all of the quotes from all of the books that you've read since 1999. Well, yeah. I mean, not all of the books, but yes, a significant amount. And so. This is very valuable. Like, can you just like send me a Stripe link?

[13:44] you want for this yeah yeah and by the way i should point out one thing that's really important here so these are these are quotes these are all books that i've purchased right and these are quotes that i've you know clipped increasingly using the kindle um using the kind of like limits that are built into kindles and the amount that you can quote and use and we are not it's really important to stress this and i this is important for me as an author we're not training the model on this information so this information we're just loading the information from these quotes into the [14:14] of short-term memory of the model and we're using that to answer questions or be intelligent about it um so there's no chance that this information which is under copyright is going to be used to train the model or be shown to anybody else um and so you have this freedom to work with material if you if you have the [14:31] right to use it under copyright you can use it inside of notebook on we spent a lot of time ensuring that that works so yeah so this is this incredible this is just one of these [14:40] I'm just scrolling through this, for those of you who are listening. It's just like an endless list of quotes, and I'm just scratching the surface. And each source that we put in, we create a source guide that summarizes the [14:52] the source. Now, this is normally extremely useful because generally a source is on a single topic and you can get a kind of high level thing. It's crazy when you give it like whatever, this is probably like 800 quotes in this one source on all these different topics. And so it has to, you know, it creates a summary like this source explores the intersection of scientific advancement, societal impact and ethical considerations and how, you know, it's doing a very good job of trying to make some kind of pattern out of all this. But source guides for just quotes are not quite as

[15:22] So now at this point, basically, [15:25] I can ask any question. And so actually, I preloaded a question. I'm going to close this source. I preloaded a question if we view the chat. And the question will be asked of all of these sources together. Yes, I can. You'll note that they're checked off here. So you can always tell, this is a really subtle thing, but it's really important about it. It'll say down at the bottom at the chat box, it'll say 15 sources. [15:55] to all 15 of your sources here. If I... [15:58] actually deselect this one for some reason. Now I'm talking to 14 sources. And so you sometimes have moments where you're like, actually the information I want is only in this one document, I don't want to ignore all the other information. So you can actually shift, it's like you're able to shift the focus of the model to various different things really easily. So, I asked this question, what are the most interesting facts about ant colonies here? [16:22] um, mention authors and specific books because I actually wrote a book a million years ago called Emergence. We talked about, you know, the emergent approach to these things and there was a big riff about ant colonies in that book. Um, so, uh, and, and the model is smart enough to understand this concept of like interestingness too and, and surprise. Like I often say like, what's the most surprising idea here? Cause so you think about that as an author and the idea that that can be like effectively a search query is just totally bonkers. So, I mean, so I'll just read this for

[16:52] So it comes back with interesting facts about ant colonies. Ants use pheromones to communicate in various messages, such as danger, food location, and nest mate recognition. This complex chemical communication system allows ant colonies to function as a single unit or superorganism. Yes, fantastic. That is a great answer. And now this is new, by the way. So this is a brand new feature rolling out today. We're incredibly excited about it. We now have these inline citations. [17:22] Thank you. [17:23] from [17:24] the my reading notes that it used to generate this answer and so you can just roll over them and you can see [17:33] where the model came up with this. This one is from Norbert Wiener. That's pretty interesting. And you can see there are citations all over the place. And what's even cooler is-- [17:46] You can, although a little bit in a way less useful for this project, we'll show it in another thing, but I can always click on those. And it takes me exactly to the point in my documents. [17:57] where the original quote came from. So you have this ability to kind of ask the model to help you get the lay of the land, like what's in here? I'm interested in this topic. What's there? And then because the sources are integrated into your notebook itself, you can then dive right in and start reading. [18:17] You can go through all these things. [18:22] And basically, and then you, and we also suggest questions.

[18:25] based on what you just asked. So there's always, you know, there's always something to just click on. [18:31] Can I ask you a question? Yes. Go right ahead. And you tell me whether or not this is a good question to ask it. But one of the things I'm interested in just looking at this and knowing how many things are in here that you've collected over the years is what are the types of things that Stephen Johnson is likely to save? What are the patterns in texts or books that are likely to make you want to put them in there? Do you think it would be good at that? Would Notebook be good at finding those things? [19:01] Yeah, it's a great question. So there are two... [19:04] different versions of why I save things. [19:09] The first is pretty easy, and yes, to this... [19:12] particular use case notebook will be great at this, which is I have a very specific project in mind. [19:18] And I'm in the middle of a book and I'm writing, you know, in Infernal Machine, the new book, you know, there are a bunch of themes. You know, you've read some of it. I think it's a you know, it has the history of anarchism, the history of forensic science, the history of the birth of the FBI. [19:34] You know, the kind of thread, the invention of dynamite all kind of woven together in a single plot. So when I'm in research mode for that, it's like, you know, I could very easily tell Notebook LM. [19:46] You know, these are the key themes of the book. Help me find passages that are relevant to those themes. And, you know, a little later in this conversation, I'll show you that in practice. That's very easy to do. And it's just incredibly good at that.

[20:02] And the trickier question is like, I often save things without knowing where they're going to go. You know, there's not there's not a specific project that they're attached to. There are often things that I often save things that I don't understand. Like, you know, when I read something, I understand. I'm like, well, I already know this on some level, so I don't need to save it. [20:32] I worked on enough projects, you get a little glimmer of like, that could be something. [20:37] And so you save those things, even though they don't have a slot to go in or don't have a chapter to go in or even a book to go in yet. And that's the question, like, over time, will... [20:48] software like Notebook LM get to know my general sensibility enough [20:53] To be able to detect those things, I would have said two years ago, absolutely not. And now I think it probably will be able to do that. And then you start really getting into the agent mode where you're like, please go scour the Internet for things that could be useful for me. And clearly that's in our future, I think. [21:14] I guess that's what I'm asking is, do you think we could ask Notebook LM what your sensibility is based on these sources and then have it return something that might tell us about the way your brain works that might be surprising and interesting? I don't know. Yeah, let's try. I'm Steven Johnson.

[21:33] the author who has... Maybe we shouldn't say who you are, just so that it doesn't have that to go. Yeah, yeah, yeah, that's right. That's right, because it's clearly... I'm so... [21:43] Well represented in the training data. I'm sure it knows exactly. 13 books will do that. [21:50] Okay. So these sources are my reading notes, notes from books I've read over the last [22:02] 20 years. Please describe-- what would you say? [22:08] Interest and sensibility and general sensibility. Sensibility. [22:14] What do I care about most? These quotes. Yeah. Oh, sorry. Sorry. [22:21] Thank you. [22:23] Based on these quotes, what do I care about most? And what can you tell about me? And what can you tell me? It's going to get very intimate, Dan. I know. I'm excited for this. We're going right for it. [22:42] Yeah, I mean, this is so much of what our existence is like now with these models. Talk about emergence where you're just like, what will it do in this situation? [22:53] do if I... That's a little bounce of the show. Okay, okay. So this is interesting. It does a little hemming and hawing. So it says, it is difficult to draw concrete conclusions about your interests and sensibilities solely from the provided source material. However, the content suggests an interest in a diverse range of topics, including literature. You included reading notes from novels like authors, just like Jonathan Franzen, history and politics, philosophy and

[23:23] right to [23:24] but it's not going to win any profilers. It doesn't want to go too far. I wonder if we did the same prompt, but tell it not to hem and haw, and assume that if I saved it, it's because it's interesting to me. [23:40] Give it a little extra juice. [23:45] Um... [23:46] I wonder if it does have some memory. I'm just like, can you please try to speculate on my personality based on my decisions to clip these quotes? I know you don't want to, but I'm asking you nicely. [24:16] Okay. It's so funny to like, to just like learn how to cajole these things. [24:24] But, but, you know, the truth is that this is harder to do with Notebook, Ellen, because for instance, you know, if you try and ask Notebook about something that is not in these sources, it will decline to answer. And so, yeah, so it's just, yeah, it's not going to do it. That's funny. Like we definitely, like we, we have tried to build this model and I think eventually we're [24:46] different levels of openness and closeness. But we really started with the idea, like, let's build a model that is fully grounded and that will really just stick to the facts. And so what it's straining against here is that the sources don't have any psychological explanation of who Stephen Johnson is. You know, and that's attention. Like I and I sometimes definitely use it. And I'm like, oh, OK, I would like you to be a little bit more open ended. But I think in the research mode, we're erring on the side of groundedness.

[25:15] more than we are on the open-ended side. [25:17] I have another one and you tell me whether you think it'd be good for this because it's less about you and just more about the sources. And that's like, what are the two sources that I've saved that most disagree with each other? And what do they disagree about? [25:30] Okay, that'll be interesting because it's so, there's so many, right? So can it, but let's try. I mean, it'd be interesting. Like of all these quotes... [25:43] What are the... [25:45] to... [25:46] to let's say authors. [25:48] authors who... [25:51] most [25:52] disagree. [25:54] with each other. [25:57] Whose positions are... [26:01] the most opposed. This is kind of weird grammar, but it'll probably understand. It's just so much, Dan. I'd be surprised, it'd be interesting to see what it says. But it may have an odd, like, "Hey dude, there's 7,000 quotes here. I can't possibly..." Okay, here we go. Here we go. That's interesting. Well, it's something from [26:23] it's from the recent stuff from Infernal Machine. So these are my notes. This is from, so this is [26:31] disagreeing about the use of violence. This is interesting because they didn't disagree at the beginning. But Goldman, Goldman, [26:37] Um, [26:38] it gets in this kind of like major fight with him. So it's not a, uh, yeah, she ends up, okay, here's an argument for why this is a good answer. They had a philosophical disagreement about the use of violence in politics that got so intense that most got up in New York City and gave a speech.

[26:54] Uh, and, uh, Goldman jumped on stage and, uh, attacked him with a horse whip. [27:02] So, so actually, this is actually a pretty astonishing answer because like, it really, like it really found, I mean, this is the only probably may arguably the only place in the notes where two figures who were mentioned in the quotes physically attacked one another. [27:20] So, okay. That's great. I have one more followup. [27:24] And then I'd love to move on. But so the follow-up is something like, [27:30] Uh, [27:31] Of the other sources in this corpus, who is the author or what is the set of ideas that these two should have come in contact with that would have helped to mediate their dispute? Okay. It'd be interesting to see. I'm just going to make sure it keeps... [27:53] the dispute in its mind so in the dispute between goldman and most over violence which [28:06] Other author... [28:08] in these quotes would have been most helpful in kind of resolving their dispute. Yeah. Resolving and mediating their dispute. Resolving and mediating their dispute. And why? [28:24] And why you're pushing the boundaries of this. I love it.

[28:32] And we shall see. [28:33] Yeah, yeah, yeah. That's what I would have said. That's so funny. So, Peter Kropotkin is the anarchist philosopher who... [28:48] Goldman was heavily inspired by, who had a kind of middle ground position on the use of violence. [28:57] And was, yeah, I mean, it was like, that is exactly the right answer as well. Okay, good. Good. Well done. Well done. I love it. I love it. This is very cool. I would love to get into, I know we have, you know, another segment, which is what I really love doing as part of these shows is to go on a little exploration. [29:18] together and use these tools to do something new that neither of us kind of know where it's going. And so I think what you have prepared is another instance of notebook with a bunch of preloaded information, documents that we've never really gone through. And we're going to use that information to help us find what could be a new concept for maybe a book or a documentary. Can you sort of set the scene for us here? What are we looking at? [29:45] Yeah. Okay. So this is a new notebook that I've created. One of the things that notebook LM is incredibly good with is a, [29:55] which is useful across a lot of different domains is interview transcripts. So, [30:00] So many workflows involve people like if you're a reporter, if you're a documentary filmmaker, or if you're a user researcher or market researcher, you do all these interviews and you're trying to discern what the patterns are, figure what the lessons are, and you dump a huge number of words of interviews. It's very disorganized.

[30:30] legitimately, which is there's a great oral history project that NASA created on the history of the Apollo missions and others. There are like thousands of interviews. And I've gone in and I've taken about seven of them from people like John Glenn and Gene Kranz and some famous people. And it's I think it's about 200000 words worth of transcripts that are in this particular notebook. [31:00] new as of today. We're very excited about it. Up until recently, we've supported Google Docs, PDFs, text files, and so on. And now we have Google Slides. [31:11] for the first time and we have true image understanding. [31:15] built into notebook as well so you can have slides in there as well and in a sense kind of talk to your images and do queries and it will [31:26] understand the images. So it's amazing. It will do handwriting analysis and things like that. It's pretty powerful. So we're really excited about that. That's a brand new feature as well. So [31:36] In this, what I was thinking we could walk through is this idea of I'm thinking about the project that... [31:47] uh might involve the history of the tragic apollo one fire that killed three astronauts in 1967 and um thinking about that maybe is like could you make a documentary about that what would that look like and and and the kind of the question is like i've dumped a bunch of sources here what do i have that i could use to kind of build the beginnings of a script maybe for for a documentary um and i want to show you one other new thing um that is related to this and in a sense

[32:17] and I need to get my bearings. [32:19] just like what is in what are in these sources you know it's like this is not a situation like my reading notes where i've already read it it's like somebody gives you a bunch of files you're trying to make sense of it you've downloaded all this stuff you don't know you haven't listened to the interviews yet and so we've added this this is a really cool thing we previewed this at ios um in sunder's keynote a couple of weeks ago um there's this new feature called notebook guide and notebook guide basically gives you a summary across all your sources in the notebook um gives [32:49] how do I really understand everything here? But we also have these templates that are kind of pre-created, FAQ, study guide, table of contents, timeline, and briefing doc. [33:03] And so those are ways to get kind of like the big picture view of what's in the documents. And they take a little bit of time to generate. So I preloaded a couple of them. So they're back here. [33:13] By the way, I forgot to mention this before, you can save anything as a note in this kind of noteboard area. So you have this area can be filled with notes, you can write your own notes. If the model says something interesting, you can pin it to this noteboard area. So you're able to kind of capture this stuff as you're having conversations, as you're reading, you can capture everything to this noteboard. I'll show you a little bit more about that as we go on. But here's the FAQ it generated. [33:39] based on these sources. [33:40] So went through... [33:42] 200,000 words worth of material and [33:45] decided, you know, figured out, okay, here's some good questions. What motivated individuals to join NASA? What kind of rigorous training and selection processes did aspiring astronauts endure? How did NASA manage the immense technical challenges and risks and their answers for all of these? And the other thing that I love to do, because this is something as a writer, you're constantly dealing with is like, it'll create timelines, which is just incredibly useful. Like creating a timeline is like the most laborious thing, but it's really something you need if you're writing a book,

[34:15] like what is the sequence and it just will go through all these disparate transcripts and pull out this is not based on it's like training data this is based on like the information that's in all these transcripts and it goes through and it it breaks it up into like the early pre-nasa era the mercury program the gemini program the apollo program their bullet points you can see for each of these um uh that break it down um so and then we also have it do a cast of characters whoa so [34:45] mentioned, who are important, gives them a brief description. It's just like, it's so useful to understand what's in the material. So that's a start. Now, [34:55] I pre-wrote, those are, this is, [34:59] The point here is that these transcripts are not at all focused on the Apollo 1 fire. So it's not a needle in a haystack, but it's like there's a giant haystack of NASA-related information in this notebook. And then there's something, I don't know, like something the size of a shoe, not a needle, about the Apollo 1 fire. But it's spread out throughout all the documents. [35:23] And so what I'm trying to do in... [35:25] putting together ideas for this documentary is figure out what's there that's relevant to this particular topic. And so I wrote, I pre-wrote this prompt and, you know, again, I'm trying to give it a little bit of context about the kinds of things I'm interested in. So I'm the author and TV creator, Steven Johnson. I'm interested in making a TV documentary about the Apollo 1 fire in the multidisciplinary style of my books and shows like The Ghost Map and How We Got to Now with a focus on surprising scientific explanations and compelling narratives. Give me a reader's guide

[35:55] interviews that I should read in getting started with this project. And actually, I think it's still in the chat, actually, if I bring it up. Yeah. So this is the reader's guide that I created. And it just goes through [36:08] interview by interview and pulls out the most relevant sections. Um, and so, um, [36:15] It talks about the Kraft interview and explains who he is. And then he talks about the Apollo 1 fire as a turning point in the lunar program. And then it moves on to this Yardley guy and then Frank Borman. [36:26] And so let's say I'm interested in the Borman quote. So I think it's this one. [36:34] uh, [36:36] Um... [36:37] Let me see. [36:39] Yeah, yeah. Okay. So I click on that. So there's a citation next to this quote, and I can click on that citation. It takes me straight to this passage. And the passage is talking about the trouble they were having the spacecraft before the fire. And so let's say I'm thinking this is an interesting quote, I'm working on the documentary. I like that. And so I can just select this quote. [37:03] And then I can say, add to note. [37:06] And so it'll take that quote and just add it. [37:09] right there to my noteboard. So I've got this saved response from this interview. So then I can kind of go through, I can go back to the chat, and I can see if there's-- OK, here's a quote from Neil Armstrong. [37:22] Um, and he's talking about, uh,

[37:25] Let's see, I can flip through these ones. [37:28] Um... [37:30] I think the last one is probably the best. [37:32] Yeah. [37:34] Okay, so he's talking about the Apollo 1 fire. Some very traumatic times. I suppose you're much more likely to accept the loss of a friend in flight, but it really hurt to lose them in a ground test. [37:46] so i'm taking that quote that's doubly doubly traumatic so i could also save that i can add [37:52] to another note. So I'm kind of starting to [37:56] collect ideas here that could be useful. And then I can, obviously I can ask anything else. Do you want to [38:03] Anything you want to... [38:04] I have a lot of questions. I'm full of questions. So... [38:08] I guess one of the things that strikes me a lot about your books is you find like a little like a pivotal moment where everything sort of like changed. And then you sort of trace how different technological innovations or different ideas like led to that moment and created it. And I'm wondering if there's a way for us to find some of those in here in ways that like we might not know already. [38:38] in this narrative. I don't know how you would frame that exactly, but that's sort of what I'm getting at. Yeah, no, I love that. Okay. So let me show you one little kind of low-tech notebook LM hack that I often have a written note with prompts because sometimes you're reusing your prompts, right? So inside the notebook, I'll have just a prompt note. And so for instance, like I actually want to continue this idea of, you know, I'm interested in the Apollo 1 fire. I don't want to have to rewrite that. So, you know, we can just take that and then create a

[39:08] with that kind of introduction and then say, I, I, [39:14] I want to find... [39:17] a scientific idea or scientific or technological idea [39:26] that [39:28] is central to the Apollo 1 fire. [39:33] that I could develop... [39:38] into a major set piece for this project. [39:44] What? [39:46] Would you recommend based on these? You don't really have to say based on these sources, but I sometimes like to do that. Can we add a little bit more? Yep. Ideally, the idea is seemingly unrelated and surprising, but in hindsight, inevitably led to the... [40:08] to the fire? It's, yes. It's tricky again because it has this, it's limited to the sources. This is, you know, this is, but let's try it. Like the... You modify it in a way that you think will work best. Yeah. [40:23] Ideally, the scientific concept will be... [40:26] surprising and involve an unusual connection that the viewer might not have originally thought of. Okay, so now I'm just going to copy and paste that so we have it. We can reuse it too. And let us see what it comes up with.

[40:50] It came back with a note called Pure Oxygen Environments. And it says, the use of a pure oxygen environment in the Apollo command module, while seemingly counterintuitive, played a significant role in the Apollo Fire. [41:02] The unusual choice stemmed from a desire for simplicity and weight reduction in the early spacecraft designs. [41:09] This is what made it so flammable. [41:13] And so here's the crazy thing, Dan. When I... [41:17] was thinking about this [41:19] a little while ago just kind of as an early idea um [41:25] independent i mean i was kind of researching with notebook but i i saw this stuff about pure oxygen and i notebook doesn't know this so this is [41:34] This is just fortuitous on some level, but I wrote a book called The Invention of Air that was about the discovery of oxygen. Joseph Priestley, the... [41:41] 18th century chemists who kind of isolated and kind of named, didn't quite name, but first identified oxygen. And I thought, oh, maybe there's kind of a version of the Apollo 1 story that connects to that history of oxygen that I've already written about. And that could somehow like be set up here. So like the fact that it like pulled that out. [42:11] that it knows that I've written about oxygen? I mean-- - Well, now we're getting, now we're getting exciting because we can take this and go back to your Readwise notes and try to pull some sources, right? Do you have oxygen? - Yeah, yeah, I could. I could, yeah, let's see.

[42:27] Uh... [42:29] So... [42:32] That would be interesting. I can't remember what's in there from that book because that was a long time ago, but it probably has some stuff in there. Let's try it. We'll go back over to that other notebook and we'll say I'm writing about the use of a pure oxygen environment. [42:51] Thank you. [42:51] that caused the Apollo... [42:55] One fire. [42:58] What... [42:59] sources in these, what, um, [43:04] Thank you. [43:05] What quotes in these sources could be relevant to the use of oxygen and its history explain how I could use those ideas? Should we ask it to be surprising or do you think we should just do straight up first? [43:28] Let's just... [43:30] It's not overtax it, Dennis. It's a very sensitive model. [43:35] I'm not going easy on it. It can handle it. [43:40] Yeah, and this does raise an interesting question. Like, I have to jump to the other notebook to do this. And, you know, certainly there's this idea that perhaps you would want to have, okay, let's start it right away.

[44:00] Oh, that's really cool. It brought up Picard. [44:04] Yeah, this is great. This is another idea I actually never wrote about. [44:09] I think we're building something good here, Dan. I love this. Are you going to want a piece of this project? I'll pick 10%. So this guy, Picard, I wrote about this because... [44:21] I wrote about the... [44:23] discovery of the ozone layer, because I wrote about the guy who, [44:26] invented the Freon [44:28] CFCs that caused the hole in the ozone layer. And so there's this explorer, August Picard, who went up to the kind of stratosphere for the first time. And so like, I'm [44:41] This product just blows me away so many times. So look at what it says here. So it reminds me of this story, which I thought was fascinating, but have never used. And it briefly describes it. I've got a link back to the original citation so I can go and read more about it. But this is what notebook says. This source provides an example of an early enclosed environment that relied on a pure oxygen supply similar to the Apollo 1 spacecraft. [45:11] environment. [45:12] Oh, and it gives this quote, as the professor remarked, when you face the possibility of shutting two men up in an airtight space of such small dimensions, you must study very carefully the problem of their respiration. [45:23] That's so cool. That's the opening to the book or the documentary or whatever. That's amazing. That's so cool. I'm actually going to just.

[45:31] go and copy that into the other notebook. I'm shifting tabs here, but we got to, we got to add that. So I'll close the chat briefly. And so here I'm just like creating a new note that it's like a written note so I can paste that in there. Okay. Do you have another one? Well, so now we've, now we've got this, this sort of August Picard idea. And I'm curious if there are any other like parallels between the two stories. So we've got the like pure, the, the, the oxygen and the [46:01] search through, you know, given this as context, this note that you just put in there, do you think we could search through the sources to see if there are other parallels? [46:09] um in other parallels to august picard in the in in the nasa transcripts yeah yeah so this is a one [46:20] thing that's actually annoying to me about notebook ellen right now which is the um the contents of your notes and the contents of your sources [46:28] are kind of distinct. So I can-- this is a key feature that people actually-- most notebook users, I think, probably don't fully understand. But if I select all my notes, I'm now effectively talking-- I focus the model. You see down at the bottom it says six notes. I'm now focusing the model on the notes and not on my sources. And so I can do things. I can summarize. I can suggest related ideas. That's what I could do, actually. I could-- actually, this will work. But what we really want to be able to do-- we just haven't built it yet-- is [46:57] have the option of talking to all of your notes and all of your sources at the same time. It doesn't quite work that way, but there is a hack that will work actually. So what we're going to do is... Couldn't we just paste the note into the prompt? Yes, we could, but I have an even better version. Okay, great. We have a dedicated...

[47:12] Um, [47:13] prompt that we've built called Suggest Related Ideas. And so I've selected the note about Auguste Picard, right, or Auguste, I don't know how to pronounce it. And now, based on that note, I'm going to click on Suggest Related Ideas. And basically, what it is saying is, take this note and find things in my sources that are related to this note. [47:34] um and you know eventually you can imagine where this is going you could be writing a note [47:38] and be in the middle of a paragraph and you're just like, okay, I just wrote this paragraph, tell me other things that are like that. And it's just... [47:44] an extension of your memory. Like that's, [47:47] obviously where we're headed um but right now you have to kind of select the note okay so here we go we suggest related ideas um [47:54] I actually haven't even tried this one. I wonder if it comes in with citations. This is kind of an older feature that I wonder if we've updated for citations. I'm learning about the product as we use it here. We're doing it live, folks. Yeah, yeah, yeah, yeah. This is the beauty of launch day. Okay, yeah, I'm sure this is going to be a great answer, but it didn't have citations because of that. I got a note to the team here. Okay, so it says the original passage focuses on his innovative use of a sealed code. [48:22] uh gondola um it it makes a connection to the perils of a [48:28] pure oxygen environment, which is great, but it actually quotes from Gene Kranz. So it gives, it gives you a quote from Gene Kranz, the famous flight director. We had become very complacent about working in a pure oxygen environment. We all knew this was dangerous. So that's going to be helpful. Then it talks about human factors in spacecraft design and says a Jonathan Yardley's oral history. And he gives a quote. And then it says this quote, while seemingly humorous,

[48:58] of the challenges of human spaceflight. [49:02] Yardley also discusses the challenges of atmospheric control, which is really important for Picard's flight. And he highlights an incident where a test engineer passed out due to unexpected nitrogen buildup. And then he talks about the [49:15] Notebook brings up the importance of redundancy as another idea. And it quotes Gene Lunney talking about redundancy, which was really important in the [49:24] Apollo Fire that they didn't have enough redundancy built into the system. So you see what it's doing here is that [49:30] It's basically been told the prompt is basically saying, find three related ideas and, you know, explain what they are and explain how they connect to the original thing you've given it. And so this is like that. It was such a like high level form of intelligence of like making those conceptual leaps. Yeah. And again, it gets back to what we were talking about at the beginning. Like there's. [49:55] This is a fusion of so many different... [49:59] separate intelligences coming together here. So you have the original sources, whoever wrote about, you know, Picard that I quoted from, [50:10] um you know that is an author somewhere who's come up with this idea or written about this person or biographer um there's steven who has curated these quotes [50:20] and put together this idea of the NASA project and, you know, gathered all these transcripts. They're the individual astronauts or flight directors who are talking and their expertise, whatever. And then,

[50:31] in the middle of it all, there's [50:33] you know, Gemini Pro and Notebook LM synthesizing all these things and making these connections possible. So it's like, I used to talk about it like a duet between, you know, human and computer, but this is like a full chorus, right? It's so, it's so extraordinary. So let's say we're going to pin. So, so let me show you this. I haven't showed this before. This is a key part of it. So this is, [50:54] where you save interesting responses, you just pin them. And so that becomes part of your note board here. - Got it. [51:02] And so we've got some stuff about Picard. We've got some quotes from Borman and Neil Armstrong. So we've got the beginnings. [51:13] of [51:14] uh you know maybe a way of framing a documentary about this so [51:19] Let's we're going to select these specific notes that we've curated. [51:23] And you could imagine this would be a much larger collection, but this gives you a sense of the new workflow that's possible. And this is the kind of thing I really think that 1% of Notebook users actually understand that you can do this. So once I've selected these notes, not only can I summarize or suggest related ideas or create study guides or create an outline based on those specific notes, I can do whatever I want with them by typing into the chat. And so what I'm going to do is say, I'm going to take that... [51:51] original quote about the thing I'm interested in making. So I'm interested in making this documentary. I'm Steven Johnson and I'm going to add, based on these notes, suggest an opening script for a

[52:10] documentary episode about Apollo 11. [52:18] one and the prehistory of spaceflight. Just adding that for the card. I will try in a little extra twist here, suggest images from these sources that could be relevant. [52:34] Let's see what it does. [52:37] So now, oh, shoot. Actually, it's not going to work because it's not going to be able to do the images because it's focused on the notes right now. [52:45] Um, but, uh, [52:47] We'll see what it does, but it should be able to generate a pretty interesting script. Yeah, there we go. Okay. So opening script. Visual. Open on black and white footage of Picard's high-altitude balloon gondola from the 1930s. It looks surprisingly modern. Narrator. In the 1930s, Swiss physicist Auguste Picard looked at the skies and at the fantastical novels of Jules Verne to imagine a new way to explore the world. [53:17] in January of 1967, tragedy struck on the launch pad at Cape Canaveral. [53:23] And look at the transition it does between these two. Narrator says, decades later, a new generation of explorers would borrow from Picard's vision, sealing themselves into metal capsules, breathing pure oxygen as they slipped the bonds of Earth. But the dream of space travel would soon be met by a devastating truth. [53:39] That's so good. I love, I love this whole journey of like Apollo on fire, Picard, and then we get a little script. Like it's, it's a beautiful idea and it's a beautiful parallel. It's so cool. Yeah. This will be coming out in two years, folks. I hope you all, a Dan Schiffer, Stephen Johnson production. Do you heard it? Dan only gets 10% though. He says, that's what he said.

[54:09] idea guy. Um, yeah, this is, this is wonderful. Um, I really appreciate you taking the time to show us this, um, this product, taking the time to go on this intellectual journey with me. Um, I would love to have you back anytime, uh, for people who are looking to, uh, read your book, uh, your latest book, or to find you on Twitter or anywhere else on the internet. Uh, how, how can, how can they find you and tell us about your latest book? Where can they find that? [54:39] true story of dynamite terror and the rise of the modern detective. So it is very much in the spirit of what you've just seen being created here, weaving together a bunch of different stories about the history of science and technology, but also... [54:51] uh, political violence. And it's kind of a thriller in a strange way in the second half of it. Um, it's really fun. Just, just in bookstores now. Um, and I'm Stephen B. Johnson on Twitter and, uh, I write the newsletter, um, [55:05] adjacent possible on Substack where a lot of [55:09] Information about Notebook LM will be there. And of course, Notebook LM is notebooklm.google.com and now available in over 200 countries around the world as of today. So please do check it out. Incredible. Thanks, Stephen. Thanks, Dan. [55:32] Oh my gosh, folks. You absolutely, positively have to smash that like button and subscribe to AI and I. Why? Because this show is the epitome of awesomeness. It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure, unadulterated knowledge bombs about chat GPT.

[55:54] on the edge of your seat. [55:56] craving for more. It's not just a show. It's a journey into the future with Dan Shipper as the captain of the spaceship. [56:03] So do yourself a favor, hit like, smash subscribe, and strap in for the ride of your life. [56:09] And now, without any further ado, let me just say, Dan, I'm absolutely hopelessly in love with you.