In this episode, Robert and Todd discuss what the recent rise in AI could mean for your GTD® system.
watch time: 31 mins
Subscribe to the Podcast
0:00:05.4 Robert Peake: So welcome everyone to another Change Your Game with GTD Podcast. My name is Robert Peake, and I’m here with Todd Brown.
0:00:13.6 Todd Brown: Hi, Robert.
0:00:15.6 RP: Hey, Todd. So in this podcast series, our goal is to help you find ways to get more done with less stress, and in particular, do that by applying the getting things done or GTD methodology, which is a systematic way to get on top of all the things going on in your world to increase well-being, to also increase your effectiveness in doing the right things, the things that matter most to you in your work and in your life. And so today, we were thinking about what should we discuss, and the thing that’s kind of at the fore of many people’s minds right now is AI, and in particular, some of the very recent advances in generative AI that have created a lot of excitement and enthusiasm for AI in general, as well as some, I think, anxiety and concerns about what it all might mean. And we’re in the business of thinking about how the real brain works, [chuckle] and how we can treat it better, how we can get more out of it, and how we can get, in a sense, more out of ourselves with less effort and less stress as whole beings, right? As not just brains or computers.
0:01:35.0 RP: And so it’s early days. I think we need to preface things in that sense that we don’t exactly know what all is coming, but based on some of the recent developments and based on some of what we’re seeing, we thought it might be interesting to talk about the Getting Things Done methodology, and how it might work and how it might support us in the coming days of what seems to be a bit of a little mini revolution in the use of artificial intelligence in our everyday lives. So Todd, with that, does that kind of tee up or frame the basics of what we were thinking we could address, and if so, what are your initial kind of thoughts on all this?
0:02:19.1 TB: Yeah, yeah, I think you teed up well there Robert, I think… And it is absolutely early days. So the technology is new, how people are using it is new, I think a lot of people are playing with things and seeing what works. And I guess my initial, one initial thought about all of this is, you used the word support, right? How would AI support us? And I think that’s an important idea. I had a… Some of our listeners will probably know that Microsoft has started to integrate AI in addition to their investment in OpenAI and the ChatGPT for the latest version of that. They’re also starting to implement in their product set, they’re starting to implement AI elements. So if you go online to Office 365 and you fire up Word in the browser version, has more AI elements, and I’m assuming they will over time, that they will continue with this trend, and we’ll see it in more and more places. So I think as a framing idea, what I find helpful for now is, yeah, what kind of support would we like from an AI in order to help us to be more productive and less stressed? And as a quick little experiment, something that I did over the weekend was I actually…
0:03:50.7 TB: Well, the first question that I asked ChatGPT-4 was, could it look through the elements of my system, which are basically kept in Microsoft tasks in the Microsoft to do, in Office 365, Microsoft 365, and present me with a summary? And what I found out very quickly was, well, no, ChatGPT-4 doesn’t have access to that. So the next thing that I said was, what if I provided you with a, basically the contents of my system, right? So in essence, an extract of all of my lists. Could you summarize that for me? And it was very open to that. It said, “Yes, I could probably do that for you.” And that just got me, and no conclusions here, just sort of early days impressions, but that got me thinking about if we… And by the way, I have not done that yet, right? That’s still yet to be done in that particular conversation with ChatGPT, but I think one of the things that occurred to me was, it’s a really interesting question, what would I want from… Let’s assume that I’ve got a good external GTD system.
0:05:03.3 TB: What kind of support would I like from AI in interacting with it, in creating it, in modifying it, in making decisions about it? I think that’s an incredibly rich question, I don’t have… Right as we’re sitting here right now, I’ve got some thoughts, but I don’t have an awful lot of answers to those questions. I know you’ve been doing some pretty deep thinking about this on your own, what have you come up with? What does that trigger for you?
0:05:32.9 RP: Yeah, no, I think the possibilities are really, really interesting. And I think one of the questions on a lot of people’s mind is, will this help us think less? Will this help us get more done? And if so, as a result, do we really need productivity methodologies and approaches? Do we need GTD anymore? Is AI going to… And this idea of generalized artificial intelligence, superintelligence, is this gonna potentially supplant our need to think so much? And I think for me having seen the Getting Things Done methodology grow up over time, even as I’ve grown up over time, from Paper Planners to PalmPilots, to PCs, to the smartphones, consistently what rings in my head is this quote from David Allen where he says, “If you don’t know what you want, any tool will do.”
0:06:33.1 RP: And so I think it’s very interesting that in, more so in Computer Science, there’s this idea of levels of abstraction, and what that means is, how close to the ones and zeros are you? Are you working at a level where you can almost describe things in human-like languages, or are you working at a much deeper level where things are more like Math? And I think what AI may do for us is help us work at higher levels of abstraction, so that we’re describing what we want into a tool that’s helping give us back feedback that we can’t just take wholesale and go off and use, but we can then start to shape. And I think the key thing there is, we need to shape it toward our desired outcomes. And so still keeping track of and still defining clearly what our outcomes are I think is gonna be even more essential in an era where we have the opportunities to some extent offload some of our cognition into systems that can help us do a bit of grunt work.
0:07:35.9 RP: But frankly, computers have always been that way from my first batch script that could run through the same thing over and over and save me a bunch of time and find and replace or whatever it was, till now where I can get the outline of a script or a book or something without having to do the initial brainstorm out of nothing. But I think the key is, that our human concerns are kind of perennial and our intentions and getting clear on what those are, I think are gonna be ever more important in relation to having even more information, even more tools, even more opportunities, even more possibilities, that we’re just gonna have to get actually more and more clear, more and more sharp. So I see this as a time where GTD can help us navigate what potentially is gonna be some really interesting roads ahead in terms of possibility. So I don’t know, what does that… Does that make sense to you? Does that kind of mesh with where your head’s been about all this?
0:08:40.6 TB: Absolutely, and I think one of the really interesting things that’s going to happen over the next little while is that, if you think about, digging into the question that you’ve asked there about our desired outcomes, I think our desired outcomes will become clearer as not only we maybe get in touch with our deeper needs as it were, but also as we see examples appear in the world of how other people are making use of this technology. I was mentioning to you earlier this article that I saw over the weekend on 35 different things that people are doing with AI right now. And it was absolutely fascinating, the range of things from the sublime to the mundane to the trivial, there were all kinds of things that people were already doing.
0:09:23.7 TB: So I think… And I’m very much looking forward to this. I think we’re going to hear from a lot of people, from each other, of course, as we, and I don’t just mean you and me, but everybody in the broader certified trainer and coach community, we’re gonna hear from them about things that they’re starting to do, and we’re gonna spark off each other. And I very much, as I say, I’m really looking forward to that. Because I think in a sense, the desired outcomes you’re talking about are the desired outcomes about my system and how I use it and the results that I get from it, and some really interesting food for thought there will be, what have other people managed to do with all of that? So yeah, I think that’s going to be quite important as we go forward.
0:10:15.1 RP: Definitely, so we’ll be keeping our eye on all of this. And it’s kind of really right in our bailiwick in that a lot of it is about distributed cognition, is about externalization as well, it’s about how do we offload some of the stuff in smart ways so that we can be more effective, but also less stressed? And I think there’s a bit of stress that’s been induced, introduced into society at the moment by the uncertainty of what this is, what this means, where this is headed both in terms of optimizations of jobs and also being on top of and staying on top of what all of this is. And I think it’s gonna shake out over time, I think we’re gonna figure out obviously what this is in the course of time. There’s a sense of people jumping on things right now, but ultimately, I think when you have some of the solid fundamental principles of GTD in place, and ultimately you have this goal or this intention to be able to be present, to be able to focus on one thing at a time, trusting that the things that you’re not focusing on are okay, and that your system is helping you to run your life and be that second brain.
0:11:33.9 RP: We talk about the GTD trusted system as a kind of second brain. Well, in a sense, we’re getting potentially a third brain with all of this influx of AI tools that can help us to, I think the phrase is become sort of cyborgs, right? Sort of part person, part machine-assisted person in going about our daily lives and activities. But I think there’s some caveats to be aware of that I think GTDers in particular, or anyone who has a real clear focus on stress-free productivity rather than frenetic productivity wanna be aware of, right? And one is that, so I’ve done, this is not exactly my whole field, but I’ve done a fair bit of work in AI, I’ve written neural networks and done a lot of different kinds of the AI analysis that you can do, including natural language processing and even some generative stuff.
0:12:36.5 RP: And they are what are called stochastic systems, they are statistical-based, right? So that’s why you can get refresh on a stable diffusion image and get a different one every time or plug in different parameters. And there’s a little bit of dice rolling going on in all of this, which is why when people say, “Well, one day AI is just gonna run my life for me,” I go, “Good luck,” right? There’s a lot of things in my life I won’t trust to a dice roll, essentially. And also these things are trained on us in a broad, big generic way, right? Right now, the quality of a report you would get is sort of Wikipedia grade by default because that’s, in a sense, what it’s been trained on is the internet that’s full of truth, half truth and outright lies, because that’s what human beings or our society as a whole is full of, so it’s reflecting us.
0:13:30.4 RP: Which means it’s not necessarily reflecting the best of us, I think. So there’s still, for now at least, gonna be quite a bit of double-checking, quite a bit of needing to treat this with a grain of salt, needing to treat this as potentially a slightly unreliable narrator or a slightly dodgy personal assistant maybe that every once in a while goes rogue and goes off the rails, or what have you. And so I think all the more reason that when we have solid practices about keeping track of our own desired outcomes and what we’ve committed to do and where we’re headed, it’s gonna help us kind of separate the wheat from the chaff in terms of tools, in terms of output from individual tools, and then in a way, we need our own systems to manage a highly system-focused world more than ever before.
0:14:23.7 TB: Yeah, I think… Don’t disagree with any of that. And it’s this whole idea of, wouldn’t it be great if I had the technology at some level that said, okay, what you should do next is X, and in essence decide for me. And we of course, in the work that we do, we’ve heard this for years. If you’ve got the right tool, you don’t need to think anymore is sort of the underlying, to paraphrase something you just said. That said though, I think it’d be really fun and helpful probably for folks, if you and I just kicked around some thoughts about things that we think could be in the future, could be made better, more efficient, more enjoyable maybe with the help of AI. One thing that came to mind for me is I think about… I was thinking about what goes on in a weekly review, right? And amongst many other things, one of the things that we’re doing in essence is that we are bringing our system up to date in essence, we’re sort of figuring out, okay, one of the reminders here in my system that are no longer helpful, and an AI that was pretty aware of my life, what…
0:15:44.9 TB: All these waiting for reminders, which of those are still relevant, it can make suggestions about things that could get ticked off because I’m not waiting for those anymore, getting my system up-to-date, and I go back to what you said, I probably wouldn’t want all of that to happen automatically, I probably wanna have the ability to sort of just go, okay, agree, agree, agree, right? As it was making suggestions about changes to make to my system, but I think that is… That’s quite an interesting one. And again, thinking about the importance of the weekly review, would if it were tuned in the right way, and again, if it were really in tune with my world and my preferences, then I think that could be helpful. What’s your take on that?
0:16:32.0 RP: Absolutely right. Yeah. And I think a lot of what people are seeing is that this very large generic model of human language, which basically understands words and phrases in relation to each other, that’s basically all it does. Alright? It’s auto-complete on steroids, times a billion training on the internet, essentially, to some extent. There’s a whole debate about whether that constitutes reasoning and sentience, and I think it does constitute reasoning in the definition that AI has of reasoning. It’s certainly better than things like propositional logic and expert systems and some of the old school stuff we used to do at what computer science calls reasoning, it’s not the same thing as sentience, and it’s not quite the same thing as what humans do. But these things can then be re-trained on different data sets. So the idea that you’re gonna have… We already all have generated, if you’re reasonably along in your life, a pretty good significant corpus of text, and so there are things we can start to train these models on, and it can be very, very interesting to us in a sense, be able to explore the domain of your own brain a bit more through whether it’s emails or things you’ve written or notes you’ve taken, or what have you, to be able to mine that back in a sense better than a Google search, but a Google search of your specific stuff.
0:17:55.6 RP: So talk about a different way to look at reference information, this is information that you never filed anywhere as a conscious decision that could be coming back to you through different lenses and different ways of querying it in natural language, not having to query it like a database search anymore. So I think that’s fascinating, for reference, for example. And in general, anything you would be doing research for on Google or any search engine, and then sort of pasting in and manipulating and mushing around, I think a lot of the generative AI can do a better job of the initial mushing around for you. So in a sense, it’s a bit more, again, like an assistant. You can say, look, you know what? I need a better trigger list for my mind sweeps, go off and find what are the areas of life that people generally have concerns and considerations about and give me back a list of those things that I can then tweak and edit. Or I want an areas of focus list, what are the typical areas of responsibility and job rules and so forth for our job areas for this particular role? What would be a good job description for this particular role? And then take that as a starting point for your professional areas of focus.
0:19:11.0 RP: There’s a lot of other things, so… Yeah, where are you seeing, Todd?
0:19:14.7 TB: Yeah, well, no, no, I think those are great examples. Another example that came to mind for me was an AI that would be aware of my current context, where I am, what tools I have to hand, what’s my energy level like, typically, at this point in the day, could make some suggestions about next actions that would be appropriate in the moment, and I find that… And again, it’s not the kind of thing that I would ever take an AI’s word for what I should do next, but I think it could be an interesting… If it came up with a shortlist, I think that would be really fun and interesting to play with and see what kind of results that generated, but… Yeah, the examples that you just come up with there, I think are fascinating examples of how we could use AI to create that larger, that higher level structure of our system, which is, again, absolutely fascinating and I think promising, I guess, is what I would say. And the thing for me, and I think we’ve said this a couple of times in different ways, the thing for me that’s lacking at the moment is that the AI doesn’t know me, it can carry on a conversation and it can learn from…
0:20:35.9 TB: And I’ve been very impressed with having a conversation with an AI that it learns as I respond to things that it’s come up with and then spits back things that seem to be relevant, which is great, but it doesn’t have any real knowledge of me just yet, and that’s, as I say, I think that’s an interesting… That could be an interesting evolution of all of this.
0:21:00.6 RP: Yeah, and talking of evolution, I think one of the things we’re probably going to see is that all of the sudden, enthusiasm and popular appeal that’s put AI in the spotlight due to large language models and generative AI, like Stable Diffusion, is creating just general interest. So I think things that have already been around for a while are gonna get funding, are gonna get interest, are gonna evolve and going to improve. So one such thing is just classifiers, the old school classifiers and already, for example, in some of the Microsoft Teams Suite, there’s some degree of trying to detect actionability in emails and summarise and say, “Hey, it looks like you said you’d do something here,” or “Hey, you attended that meeting, is there any follow-up?”
0:21:50.8 RP: So that, I think, is gonna get better. And along the lines of exactly what you’re saying, learning the individual’s preferences, because I think these systems are clunky right now, and to some extent, a little bit inflexible, it just wants to prompt you for stuff in a kind of general way, and it wants to sort of throw things onto a task list, and I think you need to actually be wary of that at the moment. The example of a friend of mine was saying is like ChatGPT and Bar, there was a meme, and they were spouting out these amazing things and writing sonnets and doing all this stuff, and then Amazon Alexa was there saying, “I see you ordered a toilet seat last week, would you like to order a toilet seat again this week?”
0:22:39.2 RP: No, I only needed the one, thanks a lot, Alexa. So that kind of AI that… Or the intentions that were underneath that sort of more clunky technology, I think are going to be better served, and I welcome the time when a truly flexible system will highlight, for example, the key parts of an email and allow me to take those key parts that do designate some degree of commitment on my part or someone else’s part, and then prompt me to get that into a clear next action for me or for someone else, or a desired outcome, or anything like that. I think the tools are gonna better and better at detecting what’s in the contents of a message and helping us to actually do something about that. So far, very few of those really do a good job of how to support you in getting that systematised, but still, again, if you have the underlying thought process, if you understand about where commitments need to go and how you need to track them, you really just need the classifier and the filter. That’s exciting.
0:23:52.7 TB: Yeah, yeah, but as you I think quite rightly say there, that’s not to say… You’re just, I think, making the very important point that we, as human beings, we’re going to be working with these technologies, these technologies are not going to be making some of the really key decisions for us; they may be filtering, they may be providing us with more contextualised information, which is representative of us as individuals and kind of the preferences and the roles, and all of the things that are important to us, but at the end of the day… Well, it’s not for nothing that Microsoft is starting to talk about the next generation of their… Or one of their big pushes anyway, I sort of lost track of all of the various threads of the things that are going on there, but they’re calling it Copilot, and I think that’s a great term, in the sense that it’s not the pilot, it’s not making decisions for you, but the Copilot is gonna be there to support. And I think… And interestingly, I think as well, that’s a kind of a… From a marketing point of view, a brilliant name, simply because it goes back to what you were saying at the beginning that a lot of people, quite rightly, I think are suspicious and hesitant about AI, and if you give it… If you give a name to something that it’s a copilot, it sounds less threatening than the all-seeing, all-knowing AI that’s gonna replace you in your job, that probably wouldn’t sell quite so well.
0:25:34.1 RP: Yeah, absolutely, and I think with the concerns about, is this a copilot or an autopilot coming our way, I think it’s important to remember, I think, that human concerns are perennial, and GTD therefore also has been a pretty perennial way of addressing those fundamental human concerns. So even if our jobs change in that we’re working at higher levels of abstraction, higher levels of thinking about things and directing things, and AI is doing a bit more of, again, the grunt work, even though we haven’t so far seen things like drafting the outline of a book or creating sketches of some images for a game you’re gonna create as grunt work, ultimately that’s where that’s gonna be sort of relegated to. And so our jobs are, again, gonna need to be thinking at higher levels about how to utilise and direct and engage all of that potential toward where we wanna go or where we wanna go specifically. So I think that’s kind of a big key… One of the most exciting things, I think, is that large language models are going to start to allow us more and more to interact in the way that we’ve been trained since our mothers need to do, which is using our native tongue, our own language.
0:27:00.2 RP: And interestingly, there are so many AI tools, so many models, there’s this site called Hugging Face, weird name, but it has an enormous number of language models and increasingly people are looking at how can we make language models our operating system? So Jarvis, one example, there’s a lot of others, but the basic idea would be to say, for example, “Make me a video of a dancing bear,” and Jarvis would go out and say, “Well, I don’t know how to do that as a language model but I do know over here, how to generate an image of a dancing bear with this model, and over here, I know how to then extrude that into a 3D model over here, and then over here, I know how to rig that with bones so it does motion in a realistic way, and this other fourth model knows how to then take that and apply a motion capture of a bear dancing to the rigged model, and then another one over here knows how to render all that down in a way with interesting camera angles and atmospheric effects,” and suddenly you’ve got a video, just having spoken it.
0:28:01.8 RP: So the idea of AI as an operating system is a very interesting idea for task execution, but I think the point that I think we wanna make is that you need an operating system for the executive function of your brain as well, and the executor needs to know and keep track of what the outcomes are, what the steps are, what the state of play is, how things are moving, and continually be able to re-evaluate that in light of new information, new priorities to keep the ship steered, if you like. So in a way, my hope is that we’re not getting put out of a job, we’re getting promoted, right? From sort of captain to admiral, we’ve got more ships in our command and in our fleet, and that having really good practice is gonna be all the more important to be able to be effective in a world where we have an opportunity to be a lot more effective thanks to technology. I don’t know. Am I mad, Todd?
0:29:03.2 TB: You’re not mad, Robert. You’re not mad. No, you’re not mad. And it just feels like… It feels like… I think, I hope that what we’ve done today is we’ve given people some food for thought about the current state of play, which again, at this moment… I have used AI to do things like summarise books for me and do some really practical and helpful things. I’ve generated some really helpful results, but it’s not yet to the point where I feel like it’s fundamentally plugged into my productivity OS, we’re not there yet, and I’m very much looking forward to what the next months and years bring, because I’m sure the picture is gonna change and there are gonna be some, on the one hand, some really exciting things going on, and on the other hand, as you quite rightly say, we need to have our skeptical antennae up as well.
0:30:02.1 RP: Great stuff. Well, thank you, Todd, I think a really another rich and topical conversation. If you found this at all useful, we’d love to hear from you, and you can do that through [email protected]. It’s exciting times, it’s interesting things going on, but as always, the key is to be kind to your mind, be kind to your future self and focus on the things that matter to you. So if this podcast series is one way of supporting you in doing that, if you found this useful and would like to hear more, do hit subscribe and like, and meanwhile, from Todd, from me, go enjoy exciting times and be kind to yourself in the process. And we’ll see you next time.