How to Subtract: The Most Underrated Skill of the AI Era - with Leidy Klotz

18:27Jeremy:It's called every human cognitive bias. Right. Yeah. How do we even think about that? So anyway, I just, I put that out there. I'd love to hear what you think. Yeah, no, I love, I think that, well, as Henrik was describing the, this, you know, default to stop things.

18:42Henrik:I mean, that also works against the status quo bias, right? So you can almost think of the, it's not as if the, I mean, the interventions that you would make for your Odysseus tied to the mask could deal with multiple biases, right? Yes. So it's not like you need to have an intervention for each one. You could just think about like, okay, here's, here are the default patterns that are going to be the ones that most get in my way. And it's not, it also, you don't really care about dealing with every single bias that has had an article written about it in the academic literature. You care about the ones that are, you think your people are most susceptible to in this situation.

19:20Henrik:And then you would design your protection accordingly.

19:23Jeremy:I like that.

19:24Henrik:I like the idea of designing that, though, deliberately, because it would also force you to think about like what's going on here. Right. Because like that just awareness. Goes a long way when people are thinking about it.

19:35Jeremy:Well, and that's the beauty of system prompts, right? With AI, it has these system prompts, which are the way I like to think about system prompts, like custom instructions, for example. Is there a parenthetical instructions to the model that precede your explicit instruction? Right. Right. But so these parenthetical instructions, you can take this Odysseus contract and you can tell Claude, hey, I know my tendency is always going to be to add. Please protect me against that tendency. And the system instruction can actually act as, you know, the, you know, Odysseus's sailors or whatever, which I think reduces the cognitive load or the burden on the human to have to remember that in the moment.

20:15Jeremy:Right.

20:15Leidy Klotz:I think the, we were talking about this the other day that the app and getting the startup land, the about us page is kind of good because that is something that you can emotionally connect to as a customer. Right. But increasingly this about me page that you have to give to your agents are going to be important, right? Because all these things are not necessarily default in the models. And so your first principles where you have strong biases, you probably need to write these about me pages now for just how the models want to interact with us.

20:46Henrik:Yeah, that's really interesting. I mean, you do kind of put in what you want for the project, but it's like the about me is a different thing. Also, of course, like these models have the same biases that humans do, right? All of them. All of them.

21:01Jeremy:Yeah.

21:01Henrik:Yeah. And which makes sense. I mean, but we just saw a paper came out like a couple of months ago that was basically replicating our stuff that we did five years ago on subtraction in AI and saying like, yeah, the AI neglects subtraction the same way that humans do. And that was validating for our research, but also just a clear illustration that all of these things are baked into the AI, which makes that even more important to put about me.

21:27Leidy Klotz:On your new work, I think all of us are trying to figure out what is the architecture of our lives in this world where we increasingly sit home alone, where we increasingly talk to bots instead of people. There is definitely this, I feel, emerging culture of everybody wants to get out a little bit more and kind of feel something. I wonder what your perspective is in what are the things that people have not designed, and that could be a physical design or kind of like a conceptual design into their lives that they should be aware of, or just start to kind of plan for.

22:06Henrik:Yeah, I was listening to your podcast. It's Nicholas Thompson, right? That's the Atlantic CEO. So, and he, I think it was him talking about the kid in the bathtub trying to compete with the phone. It might've been one of you.

22:19Jeremy:Yeah, it was one of our friends who had mentioned that.

22:22Henrik:Okay, yeah. So the idea being that, like, okay, my kid's in the bathtub, that's giving me a certain level of dopamine, but then my phone has a notification and that overrides it and my kid's having to compete with this thing that's really hard. And I don't know like the dopamine stats, but I kind of think about our surroundings the opposite way of that, which is like, there is no freaking way that a screen can do what the world can do. Right. I mean, maybe dopamine is one, one point, but we have evolved, survived or died based on being able to interact with the physical world around us. And we're wired to, to need to do that.

23:00Henrik:And so even just if you look at it, you know, a screen is whatever colors you put on there, it's still two dimensional and basically going to one sense. It's our dominant sense, but it's still just one sense. And our surroundings are multidimensional there. They hit all of our senses. They are constantly changing and they're like hitting these deep needs that we have that have evolved, you know, forever. So that's a mindset shift that I think is really important that it's like, hey, if we really want to enjoy life, if we really want to be happy, we have to like re-engage in this relationship with our surroundings.

23:40Jeremy:I agree. I agree philosophically. However, having children like you do, I know the best way to trigger an absolute meltdown is to take away the device. And what is true of children, I believe is actually true of adults, right? I mean, people are totally addicted. So while I think philosophically, we probably agree that it feels like the built and physical world is better. There's a lot of data. I mean, there are people who are literally whose lives are consumed with games or social media or whatever.

24:08Henrik:Right, right. But that's addiction, right? It's like addiction. The definition of it is it's taking you away from something better that you could be doing. And so I'm not saying I'm not saying it's not powerful at all. I'm not and I'm not saying that, you know. So, but I just like, that's not where the happiness is coming from. It's the same as alcohol, right? It's like, this is like an addiction to alcohol is not making the alcoholic happy or making them thrive. And so just remembering all the great things that are out there in that relationship with our surroundings. And, you know, Henrik, your original question was about how do you design this?

24:45Henrik:And I think some of it is just going back to paying more attention to it, right? I mean, a lot of the ways that we've made it easier to navigate in the world. So like GPS is an example, right? Now you don't have to know the no directions as well. And again, I'm not saying don't use GPS. I'm just saying that some of these technological advances have taken us away from this relationship. Same with even just air conditioning, right? Or sealing off buildings so that they can heat and cool to a specific temperature. And now you don't need to know about opening windows and getting airflow through and which cuts you off from the external environment.

25:22Henrik:So just like engaging with your physical surroundings more, I think, is a big, big kind of first step towards designing it because it's not so much like the exact design. It's more that you're doing it. It's like that you're now designing.

25:37Leidy Klotz:But I wonder if you have to design things. you know one the problem with the real world is outside of my six-year-old who goes daddy daddy daddy which is kind of like its own notification system right you know going for that walk with my dog or going to have a cup of coffee with my wife whatever like all these things that i really enjoy what they don't have is a little red number in front of it's right what they don't have is something that flash in front of my eyes 100 and so i almost feel that and maybe there's a question statement manager's question, but I wonder what there are of systems out there where you can almost like, it sounds pathetic, but reverse engineer a little bit like the notification universe of the screens that makes us so addicted and then make them available.

26:25Leidy Klotz:Yeah.

26:26Jeremy:One, can I...

26:28Leidy Klotz:My wife would go around saying ding!

26:30Jeremy:Well, let me point to one paper because I think it's, I've been thinking about this paper a lot. I think it's relevant and maybe it will even abstract beyond the kind of call it digital world a little bit. Clay Christensen, you know, the innovator's dilemma, Harvard Business School professor, when the graduating class of 2010 was leaving Harvard Business School, it was a very different world than they entered into. In 2008, the market crashed and they're leaving. And they wanted Christensen to talk to them not about disruption, but about personal lives. How do they make decisions that matter? And Christensen delivered what became one of the most famous speeches at Harvard Business School that later became one of Harvard Business Review's most famous articles, which later became a book as well, called How Will You Measure Your Life?

27:18Jeremy:And what Christensen basically observed is what's true for organizations is true for individuals. The marginal cost decision that organizations make that lead them to make wrong strategic decisions are just as true for humans. And he basically observed that we tend to over rotate. I'm very, I'm obviously simplifying here, but as humans, we tend to be gratified by immediate feedback and rewards. And so we tend to invest the marginal unit of effort towards that, which gives us immediate feedback. And so he said, for example, at my job, I get a quarterly performance review or I close the deal or I make the I write the book or whatever.

27:56Jeremy:Right. And so I'm getting feedback. He said, whereas with my kids, I don't learn if I'm a good dad. I don't put my hips on my on my hands on my hips for like 20 years and say, honey, we did it. Right. And so he said what he's observed is people, therefore, when they have an additional 30 minutes, they say I'm going to invest it in work because I get the immediate feedback. This predates certainly AI, but this observation largely predates the social media revolution. And to me, it speaks to something about our hard wiring. And I have been thinking a lot about this in the age of AI, because now the competition isn't between kind of say dinner time with my family and one unit of work, you know, make it incremental progress on the book.

28:40Jeremy:look, now with these Ironman suits, it's like 10 units of work or 100 units of work. And I was already struggling as an unwired human to make the decision for my family, you know, per Christensen's framework. Now I feel, what hope do I have to choose between kind of the unwired one unit of time and 100 units of output that my competitors are doing? Does that resonate? And if so, how can we start to design, given that the economic incentives are being stacked even harder against us?

29:14Henrik:Yeah, I agree. This concept of bike shedding, that's something I talk about in my next book. But it's basically the classic illustration of this is there's a committee convened to talk about a nuclear power plant, the construction of a nuclear power plant. And they have a meeting, get all the important stakeholders there. and they're talking about the nuclear power plant and the engineers and architects present and you know people ask some softball questions because they don't really understand and you get past this discussion of the nuclear power plant in like 15 minutes and then there's time left in the meeting and so they're like well what should we do at this time they start talking about the bike shed that's going next to the nuclear power plant and everybody has an opinion um because it's just easier to grasp right so it's like okay what color should it be painted what kind of system should we have?

29:59Henrik:And everybody can talk about the bike shed. And I think it's a, that's where the name comes from, but it's basically this tendency to prioritize small and knowable things over big, meaningful, unknowable things like, you know, what's going to happen to your kid? Because it's like, yeah, it's 20 years down the road. Plus you're never going to be able to connect the specific thing that you did in that moment to a positive or negative outcome is just messy. You know, you're moving the needle a little bit. You hope you are. I think part of that is you know kind of relentlessly focusing on what matters right for yourself um but then also for the people that you're working with or that are working for you is like continually like at the nuclear plan example if you brought to them the actual things that were important that they could talk about about the nuclear power plant you know instead of maybe the engineering details you talk about okay these are the three basic safety approaches and here's how we should talk about them and i really want your opinion on them so breaking it down into like important but manageable but yeah that's a fundamental way of thinking that predates predates ii to come back to i think the

31:06Leidy Klotz:premise of your book of we have this other pre-hybridism that we call structures to find outcomes and that seemed to be even more important as you're just describing it i have date night every friday with my wife right and we we have at least once a week where we walk the dog and it's like a religious thing, right? Like we always, you know, nobody messes with date night. And I really think that it is so easy to cancel the date night, right? Because, you know, like there's stuff that comes up and hopefully we'll be married for 15 years and hopefully we'll be married forever. And so it does seem to me that when you have the issue that Jeremy outlined, we almost have to get these things into some kind of like architectural system for ourself because otherwise, of course, dinner time will lose out when you can otherwise do a hundred people's work.

31:57Jeremy:It's almost like Stephen Covey's, have you seen this illustration of the sand and the rocks? You know, if you put the sand in the jar first, you can't fit any rocks in, there's no room. But if you put the rocks in first, this is like, you know, seven habits of highly effective people, right? Yeah, yeah. Find your big rocks, put them in first, and then you can pour the sand in and sand will fill all the way around it, right? Right. And to me, the phrase that came to my mind was, I was thinking about this as big rocks before bots. Like you have to be, you have to know what are your big rocks. And then, I mean, Henrik, to your point, right?

32:33Jeremy:You're, I mean, you may not know this lady. Henrik is helping thousands of entrepreneurs start new companies. It sounds like Henrik, if I put this framework in your life, Henrik has premeditated, not on Friday night. I'm not helping thousands of companies, right? And he's decided that my time with my wife is so important that the sand's got to fit around that. It highlights the importance of these very human decisions and establishing, as Hendrik, as you're saying, architecting systems to protect us from our baser impulses, right? If you're wondering on Friday afternoon whether you should keep helping a thousand entrepreneurs or do dinner with your wife, you'll probably go.

33:14Leidy Klotz:I'll add another one. I'll add another one that I use, OpenClaw, the new bot for some stuff. And one I've asked it now is that I work New York hours, but I'm often in Europe. And so the night tends to get long and I want to get sleep and my kids wake up in the morning. But so I've now gave the bots a command saying, I aspire to basically be sleeping by 11 o 'clock. And so now at 1030, all the bots are just going like, hey, I'll, you know, I'll ask them, like, you know, can you install this new thing? And they're like, I can do that, but it's coming up to 11 o 'clock. So do you, should we push it to some R dude?

33:50Henrik:Yeah.

33:51Leidy Klotz:And it's kind of interesting. Like I have to hack myself. Otherwise I'm just so easily seduced by these things that, that I'll, I'll, I'll stay up all night.

34:00Henrik:Yeah. I think going back to your question about the design prompts to like re-engage with the world too. Right. I mean, you can use something as simple as like when I'm looking at my phone, like space before screen. So I look at my phone as like, wait, have I checked out this place that I am right? Like really taken it in. Um, and so you can, knowing that the addictive power of the phone, you could certainly use that to help you engage with your surroundings more. I'm sure there are things that you could, yeah, like build in ways that the, the technology could augment our regular experience in the world, right?

34:34Henrik:Like, Hey, now look over here or whatever. And as simple as like just getting the phones out of the system, right. Or getting the screens out of the system. So it's change the default, right?

34:45Jeremy:That's behavior design while I want. Just eliminate the thought.

34:48Leidy Klotz:I saw somebody who had made a shortcut where before you open one of the social media apps, it basically asks you to breathe three times. And then it goes like, do you still want to open it? Yeah. And so it's going to lie a little high.

35:01Jeremy:That's so good. There's something about mantras. I mean, I'm just riffing with you all now, but I just heard, I've heard a bunch of structures define outcomes. Default dead. Dead. Sorry, default dead. That's different. Light are you saying space before screen? I wonder if there's something about encoding these truths in catchy, almost like you could think about it as choruses, right? There's a reason earworms tend to stay in your head, right? I wonder if one way, I love, by the way, Henrik, your statement to hack yourself. I wonder if one way to hack ourselves is to define what are these mantras in kind of a memorable slogan so that our kids can repeat it, right?

35:40Yeah.

35:41Leidy Klotz:Yeah, I agree. You guys into that? You guys into that? There's a weird silence there. No, no, no. But I just feel I've interrupted and spoken so much. And I feel guilty when we have such an incredible guest on. No, no, no. And I'm like grabbing the microphone all the time. But I will give you one more of these sermons. So a few ones. I have this one that I call Making is Thinking, which is where often I feel like instead of getting stuck, I basically start to kind of do something, which I really enjoy too. I stole it for somebody, but now it's mine.

36:10Henrik:Well, that's a, I mean, that was one thing that I wanted to bring up on in this conversation about AI, right? Because I see students and I see how I work. And it's definitely like making is thinking, right? And if the AI takes out the making process, you're not doing the thinking. And you've just got to, again, it's like, of course, there are times when you don't want to do the thinking. It's very rote and boring. But there are other times where it's like, no, the thing here is the thinking. And that's where I'm getting the benefit. And I can't subtract that out. But I guess there's a little bit of that because you do subtract that.

36:42Leidy Klotz:That is often what you subtract now to Jeremy's laziness point, right? We do become these cognitive diabetics where we've kind of like have basically said yes so many times to the bot that we just forget that sometimes we have to stop and go like, hey, should I just think like it sounds compelling, but is this really the best argument? And so even the trick saying, don't suggest me one thing, suggest me three white leaf different things. Yeah. At least then you're self-prompted to kind of have a decision about them instead of just taking the, what is the other one? The devil is in the default.

37:24Henrik:David Epstein's written some interesting stuff about this on his sub stack about just like the cognitive ease and like the deficits there and the risks there. And we were thinking about it as people who have like, we've learned what we've learned without AI, right? So for us, it's like, yeah, there's a little bit of a harm because we're not doing our peak thinking. But when I see college students do this, or when I think about my 11 year old doing this, it's like, not only are they not getting the ideal outcome, but they're not even learning how to do the thing, right? And so they don't, they're not developing this capability that makes them better in some way than AI, which is super scary.

38:05Jeremy:And concerningly, I think actually this is maybe it's too big. It's a societal thing, but concerningly the things that we as experienced individuals and managers and leaders have delegated to younger people that give them the ability to do it. We're now able to prompt AI to do. And now young folks are no longer given the at bats, like the failures required to get to success are being eliminated because now we can prompt an AI to do it and manage it. it's an interesting challenge societally to think about the opportunity for learning by doing and learning by making is being eliminated perhaps for a class of people, at least in the conventional workplace kind of apprenticeship model of on-the-job training, cognitively especially.

38:55Henrik:Well, you all have to turn this into a mantra, but yeah, I think the thing that I see, and the last thing I would want is people not to be using AI. Like when I talk to my students, you know, you guys, you need to understand what this can do, but you also need to understand what you can do that it can't do. And so it's like this engaging but also being deliberate about how we're splitting up the tasks.

39:16Leidy Klotz:So on the cheat sheet of your new book, one is to, per the bike shed argument, try to atomize some of these bigger kind of decisions or problems that you have to solve so they are easier, more palatable. The rocks are smaller, if you like. What are some of the other things that as you're talking to your students and you think they might not read your book, what is the thing that they that they should make sure they think of?

39:44Henrik:With a new book, I mean, I think the space before screen is the first part and you're just getting engaging, right? Whatever it's going to take to engage with your surroundings. I think the next one is, you know, it's not so much the exact design. It's like having a say in the design. Like it feels good to have some level of control over your environment. And that is something that they all have the opportunity to do.

40:12Leidy Klotz:Could you give any sample of that?

40:14Henrik:Well, so like open office versus closed office, right? There's like every 10 years, there's a new article in the New Yorker that's arguing the opposite of what the one 10 years before about which is better. It's like, of course, it depends on the details and the company and so on and so forth. But I'd argue that what's more important than the office layout is whether people feel like they have a say in the office. Do people feel like they have some control over being in this space and over how this space works? Right. And so it's things like, OK, yeah, you can give them a if they have an office that they can design.

40:46Henrik:That's one thing. If they have a cubicle that they're allowed to modify. So that's flexibility of the space, but also like the legibility of the space. You think about these offices where people come in and they're, you know, if they don't know how to get to the shared rec room or they feel like the shared rec room is off limits, then it's not serving its purpose. So like making it flexible, making it legible so they can see how to use it. I think that is a, that's an example of where like agency is this thing that is so fundamental and should be the first consideration that we have in designing and making spaces.

41:19Henrik:And yet we go to paint colors, right? and because it's easier to think about.

41:23Leidy Klotz:When we did BarkBox, one of the things that I think we did well was I designed them as homes

41:30Jeremy:where you could work from instead of basically the offices.

41:34Leidy Klotz:In the offices? Because I wanted... Yeah, so in each... On the floors, there was a huge kitchen. And that was because that's always where the cool kids hang. But also, it's such a natural way, place to kind of relax and kind of like turn off your way of thinking. And what we observed was that people really failed at home and it became a place where we didn't have to bribe people for, you know, for the drinks because it was just a nice place to be. And so I always said it's easier to make a office, you know, a kitchen or a home efficient than it is to make kind of an office cozy. And obviously in Denmark, we have this notion of hygge, which is this kind of like.

42:20Jeremy:I love hygge. it artificial created uh intimacy and it really worked same thing like people could bring their

42:27Leidy Klotz:dogs to the office but you a normal office didn't have a place for dog and human to be together and so we created these kind of sofa seating areas that was not sofa but they were more like these boxes that kind of like sat around different places where you could cuddle up with your dog and there'll be like a place where the leash could be and stuff like that yeah and so i mean like all the stuff you're saying super resonate because like i've seen the the actual result of how you can affect culture by creating a space that promotes a different type of feeling yeah and i

43:03Henrik:mean that's amazing i'm writing an article for the wall street journal now about that and i'll have

43:08Leidy Klotz:to look are there pictures online of the yeah you just serve bark box bark box office then they'll come up.

43:15Henrik:And also if you think about what's nice about a home, it's like, yeah, I control this. I can do what I want in this space. And so that's really cool. And one of the other ways to kind of give them that agency is to force the agency opportunities, right? So maybe you don't feel like moving the furniture around, but if you go into a kitchen, now all of a sudden you have to like search around, find the stuff. Maybe you decide where to set your food and all of a sudden you're interacting with the surroundings, which is, I mean, that's the ultimate like hacking where you're, you're not just giving them a reminder, you're actually having them physically do it.

43:49Henrik:So that's, that's really neat. I feel like there's always three things. Oh, but I want more.

43:56Jeremy:You want three things?

43:57Henrik:You want three, you want three growth. I think the growth hack is like, we don't do a great job of moving amongst the spaces to which we have access. um and so i think how can you explore spaces that are like beyond the edge of your comfort zone whether that's you know some place that you don't go in your office some new way to use your house there's an example in my book that everybody seems to love is like one of my friends the lees they have four kids and at the time they did this they were all four of the kids were like under seven so it was insane and they um it's like your life jeremy but they had a so what they had I asked them how it was going.

44:35Henrik:Like I asked you at the beginning of this conversation, Eric, and they said, it's going great. We've taken to eat dinner outside. I'm like, what do you mean? And they're like, no, like outside is just the default for dinner all throughout the year. And they're like, it's like expanded our square footage. It's made it so that when the kids spill, we don't even care. It's made it so that when they get done eating, they can like run around and not disrupt everybody's dinner. It's quieter, blah, blah, blah. It's so much better. And this was like a space that they had available to them and it just required them to like go a little bit outside of their boundaries and kind of fundamentally changed how they were living.

45:09Henrik:And so if you can think about analogs to that, whether it's how you move through your day or whether it's like your personal space, just using spaces that you have available to you in different ways.

45:19Leidy Klotz:And I guess you put an AI spin to it, as Jeremy were talking about, and you were mentioning too, like this ability now that to talk to your agent is probably allow you to say, hey, I have 30 minutes where I have to think through something. I'll go for a walk and basically just put on Grogle, ChattyBT or whatever, and then have a conversation as I'm going through the forest is probably useful.

45:40Henrik:Yeah, that's awesome. That's certainly one of these amazing new capabilities that we have that we just totally take advantage of. And we process things differently based on the spaces that we're in, right? So if you think about the things that you remember, they're often associated with a place, right? So it's like maybe it's a field trip from middle school. Just by taking you out of this familiar environment into a new environment, so you could even align the walk with what you're hoping to do on the task, right? So if you want to think of new ideas or you want to remember this thing, then go somewhere new.

46:15Henrik:I do this with parenting, with my son. If it's like an important conversation, I will talk to him about it in a neat spot, just hoping that -

46:25Leidy Klotz:By now he's going to be freaked out. Like you go like, dad, we haven't been here before. What are we going to talk about?

46:30Henrik:Here comes the birds and the bees talk.

46:32Jeremy:So Lighty, I'll give you a gift in closing. At the d.school, we tell folks to take a wonder-wander, which is basically what you're saying. It's like, take a walk with your problem. yeah you take a walk with your problem i borrowed this the actual physical idea from sister carita kent who is the nun in charge of the school of the immaculate heart this like radical art program and she would teach art students to make a finder which is basically what it what's written on it i'll tell you in a second but basically it's a piece of cardboard with a hole through it and they had to as they walked around they had to look at the world through the whole and it basically constrains your vision.

47:11Jeremy:And then she would give art assignments where, you know, like if I say, say I looked at the screen like this, my assignment is to draw only what I see here, but now make it 10 times bigger. Right. And then take this and, and actually look at my drawing and make another drawing, but only of what I see through this of my drawing. So she would do stuff like that. Well, I basically borrowed that idea and said, Hey, if you have a problem you're trying to solve. One way to think outside of the box is to get outside of the box. So literally stand up, take your problem and you can just say, how would insert the blank do blank or solve blank?

47:49Jeremy:And so you go, Hey, we're trying to build trust. And you go on a walk around the neighborhood. You're like, and you see a UPS truck and you're like, how does UPS build trust? Oh, Brown. They have a color that signifies, you know, deliverability. Right. And then you, and then you look at the playground and you go, how does a playground build trust? Oh, the ground is soft. Okay. How, how, what bearing does that, but all of a sudden by having a problem in mind and by constraining your vision, the world starts serving up solutions to you that when everything's available to you, actually you can't find, right?

48:23Jeremy:Yeah. It's a fun one.

48:26Henrik:It is a fun, I, I, it's, I'm not, not saying anything because it's not good. I do with my students in class we uh i'd have a class on like sustainability in the built environment it'll go outside and be like okay look at how would this leaf solve this problem i don't have the box that would be a better way to do it because it would the artificial constraint actually

48:43Jeremy:makes it better in that case yeah yeah because you can't you can't say how would the world solve this problem that's too big and actually that's a really interesting uh uh analog to ai when you ask AI question, you are saying, how would the internet answer? Right. When you, when you provide constraints, you're actually focusing. I don't want the internet's answer. I want the leaf's answer or whatever it is, which is kind of a cool full circle connection. I hadn't thought of.

49:11Leidy Klotz:That's awesome.

49:12Jeremy:Thank you so much for spending time. Well done, lady. Well done. Jeremy Utley. Henrik Kavartalan, what do you think?

49:20Leidy Klotz:Like, as we talked about, like, I had a good friend who passed recently. And I think, you know, as we were talking off air on this and I'm listening to this, I am really a victim of my fascination of AI and its seductive nature. And so what I think really hit home for me was what I think is the task for at least myself, but I would imagine for a lot of other people, of figuring out how do we create an architectural environment around ourselves that allows us to have a balanced relationship with this magical technology, but still not have it suck us in for time, not have us suck us in for laziness of thought, not having sucked in for biases that we have in the world that is encoded into the technology and all these different ones so i mean like a he was inspiring and had incredible energy and i always kind of like pretty humble by being able to have a conversation with somebody like that but for me actually more profound because it suddenly was an hour spend and the vocalization of this kind of feeling that I haven't expressed, which is, hey, I need to probably start to create God rails for my own behavior with AI.

50:43Leidy Klotz:So.

50:44Jeremy:Yeah. The notion of protecting ourselves against our, you know, baser impulses perhaps, or are at least not to say base, but a very natural impulse to doing more. And I love your examples around having your agents check in with you. Hey, it's kind of getting close to bedtime. Are you sure you want me to do this? To me, just to pay you a compliment, it sounds like you're already beginning to architect some of that environment. And I realized with your friends passing, there's the additional kind of weight of thinking about this question, are we living right? And I think it's actually funeral or no funeral, memorial or no memorial, we're all going to face that day.

51:24Jeremy:Our lives will come to an end and thinking, are we spending our time in a way that we will be proud of? I think is a really important question. And I thought the example of, you know, your date night ritual with your wife, it's, could you be more productive? Probably. Could the world get more startups out of Henrik if he spent his Friday nights supporting entrepreneurs? Probably. But it almost makes me wonder, what is Henrik for? You know, just to put it to you, right? And one thing is, yeah, sure, to facilitate and to unleash entrepreneurship. But I think another is to be a husband to your wife and you have to, uh, in an age where it's so easy to keep commissioning agents to build more startups to say one thing that an agent cannot do is be husband to my wife.

52:16Jeremy:How do I protect that which is irreducibly mine? I think is a question on everybody's minds.

52:22Leidy Klotz:And that resonates a lot. I mean, like we, I was making the joke with my wife the other day. Cause like I had installed open claw and I was excited about it and I called my agent Iggy and then I was lying in bed before bedtime and I was like you know hamming away on my phone talking to Iggy and she goes you'd rather want to talk to Iggy right now than you want to talk to me and I was like kind of and you know it was a joke right but like also horrible because obviously this kind of synthetic thing that I could talk to at any time so i i do think it is very important and i do also remind it of um so at autos the startup that i do that help entrepreneurs we are building thousands and thousands of of startups and the other day i had a journalist from the economist beyond and we went through um this whole process and we built his startup and we started to run ads and have the first potential customers come in and in the middle of it he goes yeah i can see how this could be a business, but I don't want to make this.

53:21Leidy Klotz:And so I think, you know, and I alluded to it also in the conversation, I do think that we are this interesting point where we have to articulate our principles and our narratives because we have to be able to tell it to the bot so it can help us become a better version of what we would, you know, the version that we'd like to be. I had this phrase you asked for phrase called um your narrative is your soul's code and i do think that we as humans have not necessarily had to define our soul's code our being our values because we never had anybody that could help us as fast and could basically move us forward as fast as this technology.

54:06Leidy Klotz:And I really think that without taking stock of, would I want to build this startup? Is this a startup for me? Or would I like to write this email in this way? Or would I like to push forward this idea, whatever it is, then we'll go wherever the model will take us instead of where we would like it to go. Yeah.

54:27Jeremy:Knowing oneself has never been more important. And I love actually the mechanism of a mantra or a motto, something that's, you know, pithy, repeatable, almost sing-songy in quality. I think as we think about being parents, having mantras like the space before screen. I love that as a mantra. I love default dead. I love structures to find outcomes. I think it's important that we codify these and they can't be universal values or else they're already in the model. They have to be particular. they have to have a unique resonance with you personally, your family. And that's what in the entrepreneurial space, you'd probably call founder market fit or something.

55:08Jeremy:Right. But personally, it's actually, it's about alignment and growing in alignment with our values, which I, I love. I, one other thing I want to highlight from the conversation is this idea of system instructions or custom instructions to tell the model or to, to, to invite the model's assistance in helping you think in the way you want to think. If you don't want to fall subject to the addition bias, why not tell the model that you actually want to look for opportunities to subtract? We have an amazing thought partner now that can help us stay honest to, it's too late to be deploying another agent right now, Henrik.

55:50Jeremy:And if we'll take the time to actually codify that and instruct the model, we actually get help being more of who we want, which I think is quite cool.

55:59Leidy Klotz:It's kind of a subtle thing, right? But I think when you talk about custom instructions, I think our mental model go to, we should instruct the AI to be in a specific way. It's where the object is the AI. But as we're realizing, of course, AI is more like electricity. It's a foundational technology. It's not like the web. So it's not like I used to have a magazine. Now I'm going to make it online. And so I think the reframe is really, The instruction is ironically not about the bot. It's about me. And so I have to basically look at it as a mirror and saying, when I instruct you to do stuff, what do I look like?

56:39Leidy Klotz:And I don't know, for me, it sounds like a pretty profound reframe. I know it's subtle, but I do actually think that most of us have thought in this way.

56:47Jeremy:Yeah, I like that Odysseus framing of knowing my own self. Therefore, how am I instructing you to behave? I think is kind of, what's interesting is think about the Iliad. I mean, is it the Iliad or the Odyssey? One of the two, I don't even know, but it's a, it's kind of an ancient story. And you go of all the paradigms to be drawing upon. Now we're drawing upon these ancient stories is because the things that have been true about us are even more true and even more amplified now that we have this mirror of all of human knowledge can be reflected back to us. What do I want of that? What do I want to be protected from?

57:24Jeremy:What's my unique, you know, constraint that I'm putting on it, so to speak.

57:30Leidy Klotz:Anything else we should add from this conversation? Thoughts?

57:35Jeremy:I liked hearing about voice again. I really liked Leidy's characterization of Danish, or actually you said this, we talked about subtract at the beginning. You said done is when you can't take more away. Yeah. And I think that's a principle that could be widely relevant to many listeners. Done isn't when you can't add anything more. It's when you can't take anything away.

57:59Leidy Klotz:There's a link I should show in the show note. It's a clip from the TV show Bear, the chef show. And there's this one scene that I send to my design teams all the time, which is basically where he comes in and he's being taught something. and he basically gets yelled at and then he writes on a note that's just subtract and then he puts it down there and it is such a beautiful scene. I'll share it with you and with the listeners also.

58:29Jeremy:Please do hit the link there if you enjoy it. And if you enjoyed this episode, hit the other link, which is write a review or leave five stars. If six stars are available, go ahead and hit that six star. We appreciate it.

58:41Leidy Klotz:And with that, bye-bye. Bye-bye.