The AI Researcher: From Information Overload to Active Knowledge Synthesis (Part 2)

Voiceover Artist | 00:00

Are you ready to manage your work and personal world better to live a more fulfilling, productive life? Then you’ve come to the right place. Welcome to ProductivityCast, the weekly show about all things personal productivity. Here are your hosts, Ray Sidney Smith and Augusto Pinault with Frances Wade and Art Gelwix.

Ray Sidney Smith | 00:19

Welcome back, everybody, to ProductivityCast, the weekly show about all things personal productivity. I’m Ray Sidney Smith.

Augusto Pinaud | 00:25

I am Augusto Pinaud.

Francis Wade | 00:26

And I’m Francis Wade.

Ray Sidney Smith | 00:28

Welcome, gentlemen, and welcome to our listeners to this continuation of our discussion on the AI-powered professional. In our last conversation, we were really defining the problem around information overload and many of the issues that the modern professional or knowledge worker really deals with as it relates to all of the information. In our lives today. And what we wanted to do in this episode is continue that conversation. And talk through really how to take the sometimes overwhelming amount of information, but the treasure trove of information that we have every day coming into our world and really utilizing it in productive ways. I think that today, Thanks to AI, we no longer need to think about the concept of a search engine. We need to really think about this from the perspective of it being a collaborative engine and there is this kind of reality that it could be considered an answer engine, a research engine, all of these kinds of ways in which we can coin it. There are lots of different use cases today. We’re particularly focusing in on the research And these more sophisticated AI tools can now perform tasks previously reserved for a research assistant or for you to take intensive manual effort to produce. And so let’s talk through some of the ways in which you’re utilizing AI for research purposes. And let’s think through perhaps some of the pitfalls that people fall into as they’re trying to use AI for research.

Francis Wade | 02:12

I’ve been in a whole different world as a result of deep research in the last year. I remember before It was available. I used to do… Research via looking for documents like ResearchGate, I can search for a PDF using Google. I could search Google Scholar. You could go to academia.edu and What it would give back to me, these different sources, is Stuff that was close to what I was looking for, but not exactly what I was looking for. Matter of fact, it was often not close at all because I would have a specific question. And I’m trying to get a specific question answered. But I have to find somebody who actually answered that question in a document. Or maybe a book or in something. And usually I’d be looking for an academic source. And usually I wouldn’t find anything. 

So that’s just, The game I would play was would be hunt and never find and that was 50%, 75% because I’d be looking for Esoteric stuff. Today, however, I have at my fingertips multiple A few different subscriptions to deep research and chat GPT does it for free up to a particular limit. And I can ask a very specific question. And to my shock, I can receive a plausible reply to my question Right. Pulls from credible sources for the most part. In the beginning, it When it first came out, they would pull from hallucinated sources, which was pain in the neck. But today… They’ve gotten to the point where They give credible… Specific answers to my very specific questions. 

So my research has just multiplied by, it’s hard to even compare what it was like No, Versal, what it was like before. Because I do so much of it now. It’s really been a game changer. 

So that’s at the high level. The game is completely different for me right now. See you next year.

Ray Sidney Smith | 04:15

And it will be different in a year from now even. More so. As the technology gets better.

Francis Wade | 04:21

– I’ve told people that different parts of my work. Have undergone more change in the last year than in the last decade. 30 years before that, 20 years? And this is certainly one era that is completely different.

Augusto Pinaud | 04:37

Sometimes digging and research in a topic and sometimes more than the papers, find the books. What is the book that, okay, I read this book. Now, What other… Go. Into this line and with books go on the opposite line. 

Sometimes it’s not only The papers, it’s the one to give a more… Book rented? What books? Hey, I’m dealing into… And sometimes once I want to deal or work or research into this particular idea, Bye. Where can I find those books? Because you think, okay, I want to get, how do you get granular and now fast? But then now how do you find those book, those authors, who are the authors who I’m researching this, the same areas that I’m research, it doesn’t matter if they’re agreeing or disagreeing with you, but how you find them, that was a labor Of love. A lot of times, to find those books and to find those authors. 

And then after that, then you needed to start Figure out which one was good, which one was bad. That job? One from weeks to hours. And you in hours can get a list that is better than what I was able to produce in months. This gets very interesting, the issue. Who’s this? The expectations that now the people have. Because for what you’re describing, similar to mine, it’s not only get the information, now that just you were able to get to the sources pass through. But the other part of the process is still, you need to still read it, still download them, still digest them, still trying to connect those dots. That is still takes the same amount of time, but then First part, it’s fantastic. The issue I see with this is I find a lot of people who think that find the sources is enough. And find the sources is just a step one of X number of steps to be able to get to the next conclusion.

Ray Sidney Smith | 06:47

So I think about AI in a research context, when I say this is an AI researcher, Bye. That AI can still hallucinate. I know Francis is a little more, maybe more trusting than I am when it comes to these tools. But I’ve found ways to revalidate information even after it has pulled research And again, I Preface this always with everything I do with AI, I presume to be a first draft when it puts it out. And so I’m reviewing everything as though an intern handed it to me and it’s an intern’s work product. 

So I need to make sure that it is correct. So we were all on the same page there. I think there are certain areas where AI is really good right now and where it will get better. I think that the deep research functions within all of the major tools that AI chat bots are pretty good right now. 

So you have this deep research function in Claude Gemini, and ChatGPT. Personally, I’ve found that Gemini’s does the best. I’m not sure why, but I just feel like it gets the most right when you prompt it correctly. And I don’t like the verbosity around the deep research that Google puts out, but it’s fine. It gets the data right, which is what I care about most. And that’s one piece, which is you have this complex question and you need it to go out there and scour lots of sources and come back to you with an answer. And you don’t know what the sources are. And I think in that sense, it can go ahead and find sources and then go ahead and do that analysis and synthesis that is really complex and therefore laborious and make it simpler. 

Though Concern I always have with folks is that We’re a little too trusting. So I’m going to, again, underscore the point that even after it does this research, look at the sources. It’s like statistics. You can lie with statistics. You can tell any story. With statistics. And so when it gives you all of the sources that it has found and synthesized into these deep research reports, you must look at the sources and figure out whether or not those sources say what it says it said and then making sure that the sources are valid, right? If it cites a source and it’s like, Cracker Jacks University, you know that you probably shouldn’t give as much weight to that response as, and again, the LLM doesn’t know to make those choices. It doesn’t have life experience in that sense. If it’s a well- research topic. Then lots of people have given opinions on the subject and therefore the LLM, the model underlying it, has the data to strengthen or reinforce that. The reality, though, is that popularity wins. 

So when you have something that biases popularity of a subject, so if the most people said that The world was flat on the internet. Which it is not. Then you get this problem where the consensus of the model could be different. Because it is presuming that the wisdom of the crowd is correct and it is not always correct. 

Like for, I think, hundreds of years, potentially thousands of years, humankind thought that there was an air sickness, right? Miasma, which doesn’t exist. I first learned about it actually in German language courses in primary and secondary school, where they talked about the idea of air sickness. You could be sick from the air. And so like these pieces of popular myth, these concepts can take hold in material and give you really wrong answers. And you want to make sure that you are getting closer and closer toward the truth, toward Veritas. 

So that’s one thing that I think is really important to keep in mind. We do have rag style stories. Pools like Google Notebook LM which grounds our research. In more truth. And I think this is probably a really good place 2. Spend time looking at sources that you have. You want to place those sources into a particular space. And you want highly accurate information relating to that. This is something that I do on a regular basis. I have a series of recurring meetings And those meetings are transcribed. And so I’m given the transcript for those meetings. And now when I place them into a notebook in Google Notebook LM, I can actually… Review what’s happened across many meetings, very quickly And with high accuracy, because notebook.lm is sourcing every point in the transcript where it is referencing what has happened. And so that’s been really Plus, And I’m not to just focus on notebook too much here, but it allows you to produce an audio overview, which is like a mini podcast of the content in it. It gives you the ability to create slide decks for really quick review of the content, all kinds of other. Content that it can produce based on what’s in that particular notebook. And again, it’s all grounded much more closely to the truth than not. Underline.

Francis Wade | 12:10

This, the comparisons that we can do today that we could never do before. Because it seems to be for me anyway, the big thing that’s available to me right now is that if I found one paper That answered a question that would have been amazing back in the past. Now I could take the wine, Peppa. Extract from it. 

And then go to a different LLM. See? What do you think of the output of the first LLM? 

So I could go to DeepSeek and say, what do you think of this output from chat GPT? And then it could say, no, this is wrong. Because it’s as if I’m talking to two different interns. 

So I’m not just trusting the wine. I’m not even trusting the source. I’m actually using the two and I would say poke holes in what this. Critically examine this conclusion. And sure enough, it comes back with a rigorous examination that I could not do. I wouldn’t have the time to do, let alone the background or the data. 

So doing this kind of rolling verification I found to be extremely useful. And again, I’m asking the second source. What sources are you using to refute the first set of conclusions? And it’s fine. It’s like sitting the interns on each other. And because they’re good, they’re obedient interns. They come back with a better product than the one would. Or the one source would. And they’re willing to look at different sources. Even if They don’t come back with something conclusive.

Ray Sidney Smith | 13:34

They’re not.

Francis Wade | 13:38

They would have done the initial thinking. That I just don’t have the time to do. The fact that they’re doing the thinking and sharing the conclusions is by itself tremendously valuable because at least you’re now following the argument before you had no clue. Now you could say, okay, there’s two schools of thought. There’s no clear winner, but I… Understand the nuance in a way that I couldn’t. And to pick up from what Augusto said, if I went and read the three books… I’ve read a book where I got to the end and I was like, This doesn’t answer the question.

Augusto Pinaud | 14:08

And pick them more.

Francis Wade | 14:09

Cups. Sometimes. I can’t believe I read the whole book and it got to the end and I’m like, where’s the rest?

Augusto Pinaud | 14:17

But this is not only that. You are page 30. Okay. It’s not answering the question, but okay. Okay. Probably is coming. Page 90. Okay. As a respondent, but can let me give you a little hope. Last two pages. Right.

Francis Wade | 14:29

Those are the ones where you feel like you’ve been misled. Because I’ve had to read the whole book. A huge chunk of your life has been spent doing that. 

So that’s the, let’s call that the intern versus the intern in the in terms of the LLM sort of one versus the other to cut on our hallucinations, but to understand the nuances of the topic and the pros and the cons is that it’s better if you have different sources and they’re competing for your attention. But you can do this As you said, Ray, in Notebook LM. Because You can feed Notebook LM the different sources And this is where books And I don’t know where the book industry is going to go because the way I’m reading books now, It has nothing to do with the way I used to do from the beginning to the end. No. “, and then draw conclusions from it, query the both of them, and then use the answers to do deeper queries. But Noble Kalam allows you to do that. One source versus another source. Or one source plus another source. It allows you to do something that I can only say it would have taken… Weeks of effort, amazing bandwidth, to be able to read three sources in semi-technical language, And then draw a conclusion from it. You have to remember all three at the same time. To draw the conclusion That has been, you know, notebook elements might go to Or it. Half the time, I would say. Just finding things and comparing them is so important.

Augusto Pinaud | 16:10

It is interesting. You mentioned something interesting. What is going to happen with the books? And that is a very interesting question. Outside of our topic today, but interesting question, because the readers change. Kyle, if you think on You and I as a reader, you weren’t used to read the whole book, try to understand the full thesis. When you get now to younger generations, they are, because they, we didn’t grow up, you wanted to go to the library, you started in library, in the index cards, okay? And go cryptic, okay? Try to make, okay? I’m sure this never happened to any of you, but it happened to me because my handwriting is awful. Okay, I finally got the index card which has the book that in theory I’m looking for. Write down the number, walk to the place and discover that I cannot read what I wrote and I need to go back and refine the index card. I’m sure that never happened to any of you, but it happened to me. 

That’s why people ask me, why do you love the Palm Pilot so much? Because I learned graffiti. At that time, many months back. And finally, if I write, That thing was occurring. I cannot explain the impact that accuracy in graffiti provide to my life. No more going back and forward. Okay. But then now you grab that book and as you were describing, you need to go read it, figure it out. When you look into younger generations. Okay. They’re looking. They want to say, okay, the sun is yellow. Okay. They’re looking for that. You were looking, what are the colors of the sun? Okay. They are looking for the sun is yellow. Okay. And they want, they don’t want to read the 200 pages because they don’t have the attention span for 200 pages. You were used to get a book, okay, and be able to see. The reason you didn’t read more hours straight was not because you didn’t have the ability, the skill, the training, or the passion is because you didn’t have Okay, more hours to put together. These generations, they work in a different way. They communicate in a different way. They communicate in much shorter things. Then I was a meeting recently and somebody was complaining about it. I said, hold on. This is completely, I remember when I needed to write a memo to so-and-so boss in a memo format to be able to communicate. It’s the same thing. The communication style just changed. But the reading, the research, they don’t… If I tell you an encyclopedia Okay. The people who are old enough to have a home, okay, went and think on the books and what the parents told you about the encyclopedia and all that, and how you spent hours to read half a paragraph. Wikipedia. This gets… In Google can get much faster than where we were able to. A lot more extensive and the issues are completely different. 

So. Is the reading going to change? Yes. Is the writing or the books going to change? No, the books will continue having the full thesis, good or bad, even if you and I continue finishing reading the book and say, where is the part that I wanted to read? Okay, but… What is going to change with the coming of And I’m really hoping soon It hasn’t happened. Okay. Where is the artificial intelligence Kindle where I can tell the Kindle, okay, Keep it at Amazon. That’s fine. Okay. 

Well, I can tell the Kindle, okay, I have now over 3,000 books. Give me a basis of my 3,000 books. Okay. Give me a recommendation of the next book. 

Reading, researching, always this artificial intelligence.

Francis Wade | 20:17

Then. It happened to me, what you described happened to me the other day. Where I interviewed someone for a podcast called and they had written a book. And I said, send me the book. Send me the book in PDF. The galleys. And I just said, let me put the notebook at him and see what happens next. And so what happened next was… And I did read the book. 

So I sat down for a couple of days and read the book from start to finish, just the way we describe, the old-fashioned way. But I have no need to read the book again. Because the experience I’m having of the book is not by reading the book. In the linear format. The experience I’m having is my notebook LM. 

So I know, okay, I’m not able to go in there and ask it. Ask and question and dice and slice and compare this and compare that and where is this said and where is that said. Who I did the podcast. I had accumulated all these… Insights from reading from We’re reading the book. Not from… Reading the book. And the way the book was written meant that it was difficult to Remember everything, all the different ideas. But they were very easy to parse using Notebook LLM. 

So the insights that I took from Notebook LM were what I carried into the podcast interview. And it’s what I then compared to other books, other sources. I found the sources that the author used. I brought them into Notebook LM. And I said, tell me more about this area in which the author went into. My whole experience of the book is no longer reading the book. It’s experiencing the book. Through this different But I happen to be writing a book and I’m thinking, What the heck do I do? Because… 

Someone who has the PDF, or has a version of it that they could put into Notebook LM. So same way you said, Kindle needs a… Notebook LM version that you can buy of the book. And Ethan. Give you that or sell you that or whatever. But I need to get my readers. The opportunity to Experience the book. Beyond reading the book. And I’m scratching my head and thinking, Okay, the technology, I don’t think the technology is quite there yet for what I’m in Is it coming?

Augusto Pinaud | 22:40

My mind. It’s just fairly where it is.

Francis Wade | 22:42

Is it? Commercially available tell me.

Augusto Pinaud | 22:45

It’s actually scary when I’m writing a book right now that I should be out. In May and I have the first draft. Okay. I did exactly the same exercise you were doing. Bye. Okay. And I wanted. Okay. And… It was a very From another perspective, a very interesting exercise because you are now experiencing what you are saying, but your own They want this, call the BS and call it what makes sense and bring me this and bring me the comparison and now read this text, okay, and tell me. Who has written about this in this way? Who has not? Who will write exactly on the opposite and make the writing of the book much different, much complaining.

Francis Wade | 23:34

Wow. I didn’t even think about that. I haven’t tried that. 

So writing the book, just using, having availability of a notebook LM allows you to slice and dash your own writing.

Ray Sidney Smith | 23:45

I’m going to be a little bit of a contrarian view here, which is that Not much is going to change about the book industry because of AI. I think we’re going to have more books. There’s going to be a lot more AI slop. But which is just basically low quality AI generated content. For anyone who doesn’t know. But… You need deep thought. To be able to source feature LLMs. The models need source material. They need us to do the thinking, to produce the content. If the Snake is eating its tail. We’re going to have a non-virtuous cycle for AI to continually stay up with itself. 

So there needs to be something that changes here. Because if the entire internet infrastructure, why we posted on the World Wide Web, goes away. Because You can’t be found. The same ways to generate traffic to a website or otherwise in the practical sense, which I think will be solved. It’s just people are going to go to their AI chat of choice as opposed to Google over the next few years. I still think that there’s a reason to go to Google, which is more of an environmental stance, right? Every time you do an AI chat, you’re using more power, which has an environmental impact. And so, Thinking through What just needs to be a Google search versus what needs to actually be an AI chat is a cognitive choice. And so that choice needs to be made each time. If I need to figure out how tall a celebrity is, I could just Google that. I don’t need to go to ChatGPT to burn down a tree to find out how tall Tom Cruise is. And so we need to be cognizant of that. But beyond that, The reality factor is that not many people buy books. Already. That’s always been the case. 

So the reality factor is, we are looking at probably the same people we’re going to by books are the readers of books and people who are just buying books for folks to help support their work. Are those people who are buying the books to support their work going to actually read the books? No. I know this as having published several books. People do not buy my books and read them. There’s, of course, a group of people who do. And there are lots of folks who have purchased my books and have given me really wonderful feedback. And I love that. The vast majority of people who buy books do not read them. They put them on their shelves, they sit on their digital shelves in their e-book readers, and That’s fine. 

So I don’t think it’s really going to change the book industry. As you both noted, it is changing the way in which people are potentially writing books and consuming those books. I think for the author, what I’m going to be doing is for folks who purchase my books, I want to give them access to a Google notebook, LM notebook. And so they will have access to a summary of the book. 

So not the whole book, but a summary of the book in which all the salient points and all the underlying research is there in the book. So they get the source materials and the summary of the book. 

So they’re able to go ahead and query the book based on that. And notebook LMs So it’ll work itself out, but you could create a custom GPT and share a custom GPT. In the ChatGPT space, you could do that in Qlod in some capacity. How about.

Francis Wade | 27:22

This as a way to onboard people to your book? Is that you do a YouTube short. With one idea from your book. 

And then you say to explore this one idea more. But within the book. Go to Notebook LM. 

So Notebook LM would have to, the interface would have to change, but If they could click a link from the YouTube short, No, you can’t do that, but they could follow a link. And then they come into some environment in which All you have is the one question. That they noticed or that they’re interested in answering or something provocative from the YouTube short. When they come into the notebook at them, it could be their point of entry into the bigger universe. 

So you give them what they want, which is you scratch the itch around this one thing that you’ve mentioned in the YouTube chart. And then they I’ll be able to query the book some more. 

So they go from someone who isn’t interested in reading the book but they are interested in answering one question. And they follow the one question all the way into the book. 

So to speak so they don’t have to buy it all but where i go through and i have been complaining They just need to follow the thread. Into the book, so to speak, with the one question. It’s like a little bridge. 

So if Notebook LM would make itself more a lighter version with just one kind of… You hide all of the mechanics behind there and you just… Give the person, here’s some curiosity around that. Here, scratch it over here. Then they could end up reading the whole book maybe by the end of it. But that’s the kind of research that It has to. It’s got to. It’s already here in the sense that we already do this. You pick up an idea, you go to the library, you open the book, you try and go to the index of the book, you try to find where, can I find this one person? Answer to this one question. You spend an hour, two hours, trying to flip through the book to find this one thing you’re interested in. And after you found it, or not found it you walk away because you can’t you don’t want to do anymore you don’t know how to We can capitalize on that human tendency. The curiosity that people naturally have. And you’re right, they’re not going to read the book, but they might be interested in one Sliver. And we can feed that. In the future, we should be able to feed that.

Ray Sidney Smith | 29:34

I think we’re talking about potentially different things here. One is someone trying to get using your book as an answer engine, with facilitated through AI. And I don’t know that I – Yeah, I have different opinions. I’m ambivalent about that subject as an author. And trying to get a deep understanding conveyed to another person a book is still the best vehicle for me to do that. And this comes down to a lot of the current research. This is an unpopular opinion, but I think all of the kind of handwriting over typing or fast capture into a tool being better than one over the other. And ultimately. Humans didn’t write for eons and we were digesting information by voice, right? We were capable of communicating things and people memorized ‘full’ Probes of information by rote memorization. And so this idea that somehow handwriting is better than typing has nothing to do with the physical modality of it. It has everything to do with the amount of time someone dwells on a subject. It matters if someone is corrected when they are wrong. When you make a mistake and are corrected, you actually correct. Learn something better than when you just I think you know it already and you get it right by happenstance. We need to remember that if you want to convey something deeply, you need to spend time with that person to communicate that deeply. And with a book, you have to struggle. There is work involved in the digestion. The process of comprehension creates communication. A pathway to retention. And so I don’t know that I want to facilitate Quick and easy. It’s like feeding sugar to people. That’s one part, the answer engine. 

Someone needs an answer, a quick answer to something. And then there is the deep knowledge piece that I just talked about. 

And then there is the promotional tactic of creating the YouTube short and getting someone to be aware of you, aware of your work and so forth. And that may lead to a sale. It may not lead to a sale. It may just lead to an answer. It may lead to just scratching an itch. I don’t want to try and understand all of the ways in which people are going to want to scratch itches. Or where they’re scratching them. But the goal for me is to be able to answer the questions legitimately for them where I’m able to. And want to be useful. And that’s the part where I feel like not much will change around what I’m doing. The way in which I might promote the work, the way in which I might provide different value adds. To what I’m working on. That’s just about it. 

Like Augusto, I use… The tools. To argue against myself now. And if I believe in a subject, very deeply, then I’m going to have that argument. And I have had this argument regarding different modalities of capture. I am a heavy note taker and I type a lot of my notes and I recall them just fine. I’m not stupid and I don’t think most people are stupid when they’re sitting in a classroom and they recognize that they are absentmindedly taking notes and/or just typing everything the teacher or professor says. Of course you’re not going to remember that better than if you actually were paying attention. 

So telling me that a bunch of children sat down and hand wrote something versus typed wrote something just tells me that they used a different part of their brain one that probably is underdeveloped and one that’s overdeveloped. And the one that’s underdeveloped forced them to have to spend more energy. And so the more energy slowed them down. And that’s the handwriting part. You force a kid to handwrite, Of course, they’re going to be forced to think about this thing longer. And therefore, their thoughts are going to be better. 

So what do we need to do? Teach kids to handwrite, Maybe. I have nothing against handwriting. I think it’s actually a legitimate mode of capture, but I’m unconvinced by the science that somehow handwriting is better than capturing by voice. As you just, as we noted earlier in this episode, we’re now capable of capturing the actual words in real time of what’s being said, putting those into a tool that can then analyze and synthesize all of that data across many of those conversations and find even deeper knowledge. Now, What’s our responsibility? To go memorize the transcripts? No, of course not. It’s to understand the outcomes that come along with the deep knowledge that we’re trying to absorb.

Francis Wade | 34:18

The value adds that you described I think that’s what I think is going to explode in the future. Because yes, one of the experiences is a linear reading of the book. But another of the experiences is Finding an idea that’s in the book. 

And then coming to the book through the idea. And if you, for example, say, You believe that the topic is an interesting one. That’s in your book, but it’s just one topic of many. And you decide as you’re creating the book. You say, let me argue against myself. And you create a two-sided argument around the topic. And that becomes a piece of content. And someone… Who’s interested in the topic. Comes to your whatever you have, whatever this object, this thing you’ve created of interest. And they say, hey, look, Ray is arguing against himself because you’re actually sharing it. And that little experience they have of you arguing against yourself is my God, this is so interesting. It doesn’t belong in a book. Because it’s not book content. But it’s something else that this add-on that you’ve created that adds to the experience and then makes them think, I got to go read the book. It’s interesting. Experience he’s added on at the front end, which I never would have even thought of. And he’s so generous in sharing it. And he’s even arguing against himself. That’s amazing. Who does that? 

So it’s a, promotion aside, it’s a way to explore the topic, a way that you know, you As Augusto said, you don’t want to read the whole book to find out, can I explore the topic that way? There it is, perhaps these smaller steps, these steps on the ladder. These mini experiences that we’ll provide in the future. That are Sort of. The entire linear reading. But they’re like a step stepping stone towards the ultimate reading.

Augusto Pinaud | 36:14

And as soon as I was writing the book, I went into the arguments. I went into the self-argument or arguing. What happened with that was unexpected for me because what happened is, okay, this topic is okay, but We started discussing and researching and this back and forward. And what came out of that, okay, was almost A second book? Okay, in the sense that allows me to… I was writing… Specifically about connecting invisible dots and what allowed me to look was The model… From a much higher level. If you look at it from a higher level, it’s not that the content I was writing was wrong. It is that there is another level to this, and that level, that second level, came out of a back and forward discussion, okay, between Claude Perfectly and judgey be based. And as you were saying, it was, they’ll be based on that and then get adjusted and then get to the next one in the tree of them. And that’s how it was built to this other concept that I’m Like now I’m working on it. 

So… It’s not perfect? No, it is not perfect. But probably will I have, will have get to that same place? Maybe not. Let’s assume yes. But it will have taken me a lot longer done what it took me to. Get that draft, then… If I will have done it without any of these models, and I will need to use just traditional Google or traditional research.

Ray Sidney Smith | 37:45

I think one of the next pieces here in AI is the value of us being able to have proactive information discovery. I think this is going to be really useful for folks. And The idea that if. We can have now like a Google Alerts. Type system that’s always monitoring for and providing to us information, whether they be emerging trends or counter arguments to things that we’re interested in, So say I’m writing a book on a subject. And I’m capable of. In real time, having something say, hey, by the way, new research study just came up on this subject that contravenes one of the points that you were making in the book. Now I can read it, I can digest it and say, okay, I think this is valid research. I think we need more research, but this is a good point to put in the book to say that some early research says that this point may be invalid, but I still think it’s true. Right. And so you can be a scientist, right? 

Well, not everything is. Fully fleshed out in science. 

That’s why we will always need scientists to be studying things on the frontier. And so having this kind of proactive AI information discovery engine, I think is going to be something that we start to see more and more of. Right now, I’m utilizing that in my own Google Workspace account, Google has created this ability to say, every day. Give me all of the news associated with these topics. And so it gives me a little brief. It shows up in my inbox and it’s the best sources that it can find. And if it doesn’t find anything, it says, by the way, no new sources. You don’t need to read anything. And that’s fantastic. 

Something that I’m using some of these tools for now today is being able to in essence, look at a group of information and to quickly vet whether or not I want to read that information. I use Feedly for most of my ingestion of news and other kinds of things that are coming in, whether it be podcasts or newsletters or blog articles. And as I see those things come up in Feedly, There could be thousand articles at any given moment that I can look at. And for it to be able to quickly process of the Feedly does a pretty good job. I have a Feedly Pro account, so it gives me the most important ones with its own algorithm. Once it’s done that, I can look at those few, which could be a few dozen articles, And it can now do the analysis on that. I add them all. To a particular space. And now the AI can say to me, you know what, these three articles are worth you reading. These three articles, here goes the bullet point summaries. And now I’m saving time by not wasting it on material that I wouldn’t otherwise need or want to read. 

And then the ones that I do want to do a deep read of, I will. And to be quite honest, I find that if someone spent 5,000, 10,000 words on a subject. I’m much more likely to want to read those today than I am wanting to read the 500, 900 word articles because they’re just… They don’t go deep enough into these subjects for my interest. That may just be me. And so I understand that I am a reader. I’ll read anything, including the sides of cereal boxes. I can’t help myself. It’s a compulsion. And that’s just my nature. Maybe other people work differently in that sense. But I like the idea of using the tool to do that kind of first pass analysis. And, of course, I’m still… Scanning the titles of articles. 

So I know pretty much what the article is going to be about if it has an appropriate headline. And I’m going to be able to get an idea of what the AI produced versus what I think it’s going to produce. And it’s getting pretty good. Once in a while, I’ll still pick an additional article here or there where I’m like, you know what? I still want to look at this, even though it may not. And those are usually like An article about a piece of technology that came out and all the specs are in it. I want to look at all the specs. I’m a tech geek. 

So I want to look at every little detail that came out for this particular phone or that particular laptop. And so the AI might think, that’s just that seems not within the scope of deep research. Concepts or whatever else, but They happen to be for me.

Francis Wade | 42:15

Is there a way to use your tool to score LinkedIn, for example?

Ray Sidney Smith | 42:20

So there are tools that can turn anything into an RSS feed. So in that sense, perhaps you’d still have to do the process of gathering the material. And placing it into your own AI tool of choice. In this case, it would probably be Google Notebook LM. 

So you’d have to automate the scraping of the content in that particular tool, which can be done. There are ways in which you can scrape all kinds of websites now. For better or for worse, there are ways in which you can scrape that data.

Francis Wade | 42:51

Right. I would definitely benefit from that. Most of the content that I read is on LinkedIn just because that’s where it happens to be around specific topics. And it could be scraped, aggregated and fed back to me every day. My life.

Ray Sidney Smith | 43:07

Yeah. So most of the things that are available are going to be Like when you have shared something on LinkedIn, as opposed to when someone has done something on LinkedIn. I’m looking very quickly here to see whether or not Zapier has triggers associated with LinkedIn and there aren’t any. There are only actions. 

So you can take action to post things onto LinkedIn, but the automation tools LinkedIn has blocked you from being able to look at when someone has done something versus when you want something to be done on it. And but then again, they’re more manual automation where you can have maybe Claude cowork or another tool, go to LinkedIn and literally copy and paste the content out of the feed. On a regular basis, a little bit more resource intensive. 

So you’d have to really want that maybe even doing some searches upfront. So it searches the data or searches particular things. If you are signed up to newsletters, LinkedIn newsletters, which are basically the old LinkedIn pulse. They’re basically blog posts slash newsletters, sub-stacking. And If those are coming into your inbox, then that’s a little bit easier because you can just like forward those emails into a tool that can look at them in one. Space like Evernote with the Evernote AI Assistant. 

So there are some ways in which you could do that.

Francis Wade | 44:34

In the future, they’ll have to get better. They’ll need to be aggregated and they need to be simplified and Because I yearn for that update that you I believe it exists for academia. But in the fields that I’m in, there’s not much coming out of academia that is of any interesting value The And I think that’s true for the productivity field as well. When’s the last time? 

Someone did some research. Academic research. That moved the needle. In our field. I struggled to And I use the Two hundred and something for them for my book. Back in 2016, because that was always there. Boy, I think that things have changed. That they’ve gotten less helpful in terms of academic research. It’s just gone the other way.

Ray Sidney Smith | 45:24

I think that’s more of a funding problem. Universities don’t want to fund interdisciplinary research, so that’s a funding problem. That’s not a lack of desire for people to understand it in the behavioral sciences. Because there’s lots of research being done on the productivity side, but it’s mostly technology-related as opposed to the specifically behavioral science side.

Francis Wade | 45:44

Exciting times that we’re in. These are amazing times. It’s an exciting time. Time to be alive. This is an exciting topic. And boy, the last year is any indication then. The years to come are just going to be, for the first time in my career, I feel as if I’m barely holding on for the ride. Holding on to the edges of the car and the car is like hurtling Up and down a roller coaster. But I’ve never had that experience in my life. Maybe very early in my career, Maybe. Before the internet was there. Maybe when the internet came out. Maybe. But this is really game changing and it’s exciting.

Augusto Pinaud | 46:21

Well, I’m going to cover it to you. This theme, and we will say this multiple times during this theory, AI. It’s impacting things. That We don’t know, but you need to be careful and we always need to be careful. Okay. Terminator will come and you need to make sure that you’re not Sark Hunter. Okay. That’s coming in a certain way. And… The same thing with the research, the same thing with what you want. Can it be An incredible tool? Yes. But is going to be a tool that is going to be used for manipulation, is going to be a tool that is going to be used for misdirection. And it’s a tool that by itself can misdirect you because the best definition I have heard Yeah, it is like having… Hundreds of teenagers trying to find your information. They can find the information, but they have no criteria about it. 

So you need to think. What you are receiving and adjusting to really get 100% of That benefits.

Ray Sidney Smith | 47:26

So I’ll bundle this up into three thoughts. Closing us out, which is that I think that we should always treat the AI. As A. Drafter So when they synthesize output for us in a research capacity, that’s a hypothesis that they’re providing or a draft. Of knowledge. And so the involved is charged with verifying resources and ensuring logical consistency. Because what I’ve found is that there many times can be illogical consistency and logical inconsistency. And those two things will trip you up if you are not doing the cognitive work. The fool cannot replace your thinking. And when it does, it can lead you astray in ways that could be incredibly terrible for you. 

So I just really think that it needs to be involved in more of these kinds of what you want to do, right? And if you want it to take over your thinking for what you want to do, In certain capacities, I think that makes sense. But in other capacities, when you want it to do the data synthesis for helping you think about the world. That’s when it becomes a problem. Because you should have your thoughts. You shouldn’t have something else, some thing, right? This is code, an algorithm. Choosing your thoughts for you. Next is use AI primarily to find the structure of the knowledge because it’s really good at doing that. You can talk out loud for hours on end and then have it do some research based on what you’ve talked about, and it can then go ahead and fill in gaps. It can verify specific critical details from original sources for you. 

So it’s a really good tool for that kind of thing. And then finally, you are collaborating with the AI to refine and restructure complex concepts. Those can be things that challenge your own thinking, but they can be challenging the sources that you have summoned, put together. But arguing against and for particular types of logic flows. I think that’s really helpful and something that the AI researcher role, that persona that you can create can be really useful. 

So with that, gentlemen, thank you very much for this conversation, and we will move on next. Episode to talking about AI as a learning partner. We’re going to talk some more about that. And just once, I’m not sure if I did this, I said this already, I’ve been writing this Substack AI series. And so by the time this episode comes out, they’ll Most of these articles will be out and done. 

So feel free to go check those out. I’ll put a link to those in the show notes for folks to contextualize what I’ve written about the subject, because each of these personas are ones that I think are the fundamentals. There are more, I’m sure, but the four fundamentals of the AI assistant researcher, the learning partner, and then the coach are the four fundamentals that we’ll be talking about in this first part. And so next episode, we’ll talk about the learning partner. While we are at the end of our discussion, the conversation doesn’t stop here. If you have a question or comment about what we’ve discussed during this cast, please visit our episode page on productivitycast.net. They’re on the podcast website. At the bottom of the page, feel free to leave a comment or question. We read and respond to comments and questions there. As well, you are invited to join our listeners group inside Personal Productivity Club, a digital community for personal productivity enthusiasts that I host, where you can interact with the ProductivityCast team directly. To join for free, visit productivitycast.net forward slash community. And you can get started there. By the way, to get to any ProductivityCast episode fast, simply add the three-digit episode number to the end of ProductivityCast.net forward slash. 

So episode 100, productivitycast.net forward slash 100. Episode 101 would be productivitycast.net forward slash 101. And so on productivitycast.net on each episode page, you’ll find the show notes. 

So links to anything we’ve discussed are easily jumped to from there, along with text transcripts to read and download. If this is your first time with us, please consider adding us to your favorite podcast app. If you click on the subscribe tab on productivitycast.net, you’ll see the instructions to subscribe and or follow us and get episodes downloaded for free every time a new one comes out. And if you enjoyed spending time listening and learning with us today, it’d be a great help to us if you added a rating or review in Apple Podcasts or your podcast app if it has a rating and or review feature. Your compliments motivate us and help us grow our personal productivity listening community. Thank you to those who have left reviews. We’ve seen them and appreciate all the feedback. Keep on coming. If you have a question about personal productivity you’d like us to discuss on a future cast, please visit productivitycast.net forward slash contact. You can leave a voice recorded message or type a message. Into the message box and maybe we’ll use it as a future episode topic. I want to express my thanks to Augusto Pinaud, Francis Wade, and Art Gelwix for joining me here on ProductivityCast each week. You can learn more about them and their work by visiting ProductivityCast.net and visiting the About page. I’m Ray Sidney Smith, and on behalf of all of us here at ProductivityCast, here’s to your productive life.

Voiceover Artist | 52:50

Sit this episode of Productivity Cast, the weekly show about all things personal productivity, with your hosts, Ray Sidney Smith, Augusto Pino, with Frances Wade, Art.