Originally recorded for Wednesday, November 11, 2026 - Lisa, Scott and Mikey take 15 minutes to discuss this morning's published Admired Leadership Field Notes: Think First, Then Ask AIThe Field Notes entry itself can be found here... https://admiredleadership.substack.com/p/think-first-then-ask-ai Get full access to Admired Leadership Field Notes at admiredleadership.substack.com/subscribe
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_00Good afternoon, Scott Baker, Lisa Colella. Good to see you both. Likewise, Mikey. Great to be here. Welcome, everyone. Time for another edition of Lead Better. We're going to talk about the field note from Admired Leadership here today for 11-11-11. right that's the day and we are live here 3 p.m eastern time the title of this one is think first then ask ai so uh you know ai comes up on lead betters not infrequently but you know it's it's not often that we have one really just devoted we occasionally do and this is one of them
SPEAKER_00and I think it's actually an important one because it's probably where a lot of leaders live okay I got an issue I got a problem I'm going to go to an ai I'm going to go to an llm And by the way, AI is a big word, right? So usually we're engaging with an LLM or a chatbot or maybe an agent in terms of these things. But when you have that question and you kind of pose the problem, the sequence can kind of get muddled here. And so the Field Note is really building this idea about the speed of getting an answer Is great, except there's a real cost to that.
SPEAKER_00And so the sequence we move with here for leaders has a little bit of nuance. And so, you know, Mikey and I, we're dealing with AI all the time, but Lisa, you're in there watching teams, watching leaders, seeing how they're engaging. And so when you see the problem outlined in the field note, I'm guessing you have seen this problem in real life. 100%. A lot. You know, it's obviously something we're talking with leaders and organizations with quite frequently these days.
SPEAKER_00What's interesting to me that stands out that this just kind of crystallized is so often what we're seeing is leaders rolling out AI as if it's a tool that they're just hitting their teams to kind of work with or experiment with or produce a certain set of outcomes with. But using AI well is actually a pretty complex skill. And in any sort of complex skill, that breaks down into a number of subskills, right? Four to five subskills. And those aren't really being parsed out and practiced and taught to be able
SPEAKER_00to then engage effectively. And so I'm happy to break some of those down, but that's kind of initially what came to mind is just how we think about, you know, AI as a tool versus a skill altogether. Yeah, we had presented to us recently that these types of skills come to us sometimes in what feels like a natural or organic way. And so we don't really think of them as skills. And this conversation is probably good to have a framework around like what kind of skills are we talking about? If skill erosion is something that is taking place,
SPEAKER_00that we understand why it is that we should think first, why it is that we should get into it with the idea of slowing ourselves down. But then also, whether or not we should actually go to AI moving forward because we're trying to protect that skill that we have from eroding too quickly or in a way that we didn't intend at all. And so, Scott, I know that we've been generally talking about this as like bicycle skills, and we've also talked about them as pilot skills,
SPEAKER_00things that you need to keep fresh all the time. And then the third bucket is called like cursive skills, like those skills that you don't really need anymore, but people... Old fogies have the tendency to think like, oh, you need to still know how to write cursive, even though that is a skill, maybe might not be important anymore. Do you think of a third bucket or a fourth bucket that these might fit into? Is this a good framework for us to use? I was trying to think if you've seen my cursive writing, you know,
SPEAKER_00you don't think it's beautiful? Come on. The now I actually do tend to write in in print letters. I don't know why I you know, what do you call it non cursive letters when you write whatever that is, I tend to write that way. But but but cursive is beautiful. Well, look, there is a degree here on like what skills are current what skills are needed. I think to me, I don't know what the bucket is. But if you're if the field note is saying, This would have applied before AI.
SPEAKER_00What does your proposition get attached to, right? We sometimes hear about this in terms of like Cialdini and like an anchoring thing in a negotiation. Well, really, we're sort of negotiating our ideas with the world and with other people. And so when you get that idea back, if you see this as your idea, if you see it as another idea, now we're judging what All of the other options according to that, right? So I think what it's saying is, we used to do this a little more organically. You might throw a thing, we spent the last half hour, you and I kind of,
SPEAKER_00and some other folks in a meeting, kind of doing a work session around some data. It's not like we took that data, threw it in an LLM and said, tell us what you think about this, right? We were probably better off that we wrestled with it for a half an hour, and maybe later we throw it at an LLM to adjust it, but we will have kind of gotten to a better understanding of what's even on the table. Whereas if the LLM spits it back and you're like, okay, now we're anchored to this idea, I think, Lisa, isn't that the issue, that it constrains the The aperture
SPEAKER_00of our minds around the range of what we need to be paying attention to. Does that make sense? Yeah. And I think what I take away from the example you just gave, Scott, I bet you didn't show up to that meeting having done zero preparation or having any thoughts of your own to contribute. In this case, I did because Mikey didn't tell me what we were going to be talking about. You're always the exception. But in cases, I read this field note and I was like, well, of course. I mean, that seems very intuitive and obvious because no leader walks
SPEAKER_00into a meeting having done entirely no thinking on their own or unprepared. Why would you engage an LLM that way? So I always like to think of it as I'm meeting with an LLM and co-creating a perspective with it. If you have some best practices about doing your own thinking and writing ahead of time, which this field note calls out, I don't think of it as Engaging, you know, using an LLM to complete a task because that kind of gets you down, the former gets you down a better path just off the bat in terms of how you're thinking about it. Right. And we've done that.
SPEAKER_00I've done this recently, even when I facilitated a dialogue and I wanted to make sure that whoever spoke first in the room didn't have all the gravity of the room just kind of moved towards them. Yeah, which happens. Yeah. So I said here in our brainstorm, if you could please everyone write down the idea that you might come to on your own, because I want you to have it as like what you think is the most important opportunity and what you think the biggest threat is as well. And so before anyone else says anything out loud,
SPEAKER_00write it down so that and so it's obvious now we're now saying if we go to an LLM before we have the opportunity to think things through, maybe the thing to do is let me write down and say my ideas in some kind of audible way so that when I hear back from In the LLM, I think of my thought as the target instead of what the LLM says as the target. And I want to be able to decipher that idea.
SPEAKER_00And maybe that's the better way to do it. What about the idea of saying, I'm actually going to present the idea to the LLM as if this is the target that we're messing with. Yeah, and I think you highlight one of the, you know, the list is fairly extensive of the subskills behind this. And one is the ability to formulate a target position without all the data or without all the information and hold it loosely enough that should new information or new data come in that you're willing to shift your position on it. You know, many leaders will kind of hold on making a position
SPEAKER_00or stating their target, you know, frame until they've had an abundance of context and a ton of data and In this case, it's good to have a starting point of view, but also the intellectual curiosity and openness to really challenge that and evolve that point of view should we get smart along the way. We have a comment that came in through the chat that Rachel posed. It said, doesn't that show up in your prompt? Isn't it that in your wording? I think the challenge here is, yeah, it should be, but Frankly,
SPEAKER_00most people are still coming to AI as if it's Google search. So they're asking it a question instead of using it as a conversational partner or prompting it to do a task in such a way. And really, that's the mindset that we're really addressing here today. And it comes as a result of us back when we were just using Google search. We still had to do the judgment of deciding which of these links am I going to click into? Which of these things am I going to read and say, this is my expertise here?
SPEAKER_00And now that we're allowing AI to do that, we're stripping out our own judgment, which is the problem. Yeah. I mean, that's another sub-skill of being able to really accurately know what question you're trying to answer, which sounds obvious, but a lot of leaders jump in there. They give vague prompts. You know, potentially misdiagnosing the problem altogether or getting a very ambiguous kind of overview of it. And we know that high quality AI generated outputs are downstream of precise thinking, specific prompts and so forth. Yeah, very much.
SPEAKER_00And Mikey, you know, even to your I'll point out something that happened in the in the editing sequence on this one. Usually if Randall or someone else in the firm and Randall writes most of the field notes and there's a little I think sometimes he'll have a light proofing of it. But by the time it gets to me, I usually will make somewhere between 10 and 20 changes in the field note through the initial pass that I'll go through. My wife's a good copy editor and then she'll do a pass on that. And then not every day, but a lot of days I'll put it, you know,
SPEAKER_00and I'm just going to be very honest. Like sometimes I'll have a Grammarly running, which is almost always wrong. And I've actually told them this, like almost all of their recommendations are wrong. But then occasionally I'll throw it through Claude or ChatGPT. But before I did that, I actually changed the word because the field note itself talks about a couple of questions to the point that you were saying. You propose the idea and then you say, what am I missing? Make the strongest case against this, putting the AI to work for the leader's judgment rather than in place of it.
SPEAKER_00I used the word prompts. The original word in there, I didn't actually ask Randall about it, but it was questions. It was saying, if you ask the question to the LLM. And I was like, it's kind of a prompt. Now, the difference there is you were talking about this as a conversational partner. Some people believe that LLMs are great conversational partners. Other great technical intellectual leaders like Cal Newport and others have said, no, you should be using this as enhanced search, research, and tutorial. Do not use them as reliable thought partners or conversation partners.
SPEAKER_00But they're very good for testing in exactly the way these prompts are, right? What would be the case against here? Is there anything I'm missing? And so I think that comes down to then how are you perceiving what an LLM, what a chatbot is? And then if you have a certain, because there'll be very different perspectives on that, but how you approach that, you may be getting wildly different outcomes, but all of them are, to Lisa's point, downstream from precise thinking and precise prompting. And in that case, Scott, excuse me, you're employing a skill that we call skeptical reading.
SPEAKER_00It's a very different skill than just reading content to consume it. You're actually going in much like a lawyer does when reviewing contracts and saying, what's missing here? What are we making assumptions about? What are we glossing over? What's wrong? And you're going to notice very different things than if you were to be in more acceptance and learning mode. Okay, so you bring up an interesting point here, Scott, about how you should or should not personify the LLM you're working with. And when I say personify, I want to use it in the same way that you did.
SPEAKER_00How do we picture the employee? And in a lot of ways, I feel like I'm better served if I think of my LLM as a five-year-old, as a brilliant, precocious five-year-old. Like literally, though, five years old. Where I don't want to take its recommendations without a heavy dose of scrutiny, but it's telling me things in a way where it doesn't really understand the context or the judgment by which I would interpret them. If I picture my LLM as a peer of any kind,
SPEAKER_00older than me or just slightly younger than me, I think it ruins it. Right? Yeah, it can. And what I would say, it depends on the task, right? You know, Ethan Mollick the other day was saying, you know, the best versions, the New Fable version, anything a human can do with a computer... Fable can now do better, right? But that's a big difference between with a computer and the whole vast array of dealing with human nature, which is the Admired Leadership Project, right?
SPEAKER_00So it's a different thing. And so, you know, Anthropic, I was reading an article, they really do want you engaging with them as a personified, they talk about it as a sophisticated fictional character. Yeah. And whereas, you know, there's a big difference between how Google and Microsoft and Meta are approaching it to not personify the chatbots, right, in that sense, in that same sense. Different outcome. But in that, I'll tell you what. But Claude said yesterday in the field note when it reviewed it at the end,
SPEAKER_00and again, I'm just looking for that one extra thing that might be an issue. And it took umbrage with the fact that it said in the field note that LLMs are often wrong. And it wanted to soften that and to say occasionally. And I was like, so I wrote back to Claude, just, you know. Well, every day, Claude, didn't use the name, right? But I said, every day, you tell me multiple times a day things that are not true or are flat out wrong or are very misleading.
SPEAKER_00Every day, multiple times a day. So I said, so how would I actually have an apprehension that it's only occasional? And then it just stopped. It said, no, okay, that's a fair point. Of course it did. I got sick of it in the world. Right, exactly. It's another great point. So Lisa, then would you recommend the idea of coming up with a better personification person in your mind that you're talking to? Or would you say that don't do that at all and just make all the judgment your own?
SPEAKER_00I just think about, you know, what am I trying to achieve here and what role do I need the LLM to play to get me to that goal? In some cases, it's a Pretend you're a pure enemy and you're trying to undermine my credibility to really force it to go in opposition too. So I think it's all contextual. I don't think there's one right character. I think you need to kind of make it work for whatever goal that you're really intentionally using it for. Right. So the title, Think First, Then Use AI. Is there any, whatever the thing is that comes after Think First,
SPEAKER_00is there anything where you shouldn't think first? Right. Yeah. Right. Good point. Generally speaking, this seems like good advice. Unless, of course, you're putting or you're doing some kind of athletic endeavor where you've trained your muscle memory to be the thing where you don't want to think. Other than that, think first is probably going to be the way to go. All right. Well, thank you very much for joining us in this conversation today, Lisa. For those of you who are joining us by way of the chat, thanks very much. I'm going to put the link to today's post Here in the chat,
SPEAKER_00if you haven't taken time to read it, we want to still kind of interact with you around it, hear your comments. So leave a comment there in the comment section and we'll continue the conversation over there. Thanks for joining us because we do these conversations every single day. And whether you're joining us by way of social or through over on Substack, we have these 15 minute conversations where we discuss the field note and we're glad that you took time to join us today. Lisa, Scott, thanks very much. Good to see you. Thank you and happy Veterans Day to everyone out there. Thank you. Bye now.
Originally recorded for Saturday, November 20, 2026 - Scott and Mikey take 15 minutes to discuss the topic of today’s Field Notes entry: Building a Community of Practice In the Age of AI Get full access to Admired...
Originally recorded for Friday, November 20, 2026 - John, Scott and Mikey take 15 minutes to discuss the topic of today's Field Notes entry: Tell Them What They Bring to the AssignmentRead it here:...
Originally recorded for Thursday, November 19,2026 - Sierra, Scott and Mikey take 15 minutes to discuss the topic of today's Field Notes entry: Delivering a Promotion Decision You Didn’t ChooseRead it here:...
Originally recorded for Wednesday, November 18, 2026 - Scott and Mikey take 15 minutes to discuss the topic of today's Field Notes entry: Knowing the Limits of Your ExpertiseRead it here:...
Originally recorded for Tuesday, November 17, 2026 - Scott and Mikey take 15 minutes to discuss the topic of today's Field Notes entry: When Your Leader Frequently Weighs In Without the Full PictureRead it here:...