#585 AI Co-Thinking: How Leaders Think Better with AI, with Prof. Nathan Mauck | Niels Brabandt
AI Co-Thinking: How Leaders Think Better with AI, with Prof. Nathan Mauck
By Niels Brabandt
Niels Brabandt is an expert in Sustainable Leadership with more than 20 years of experience in practice and research.
Everyone has an opinion on AI. But how many leaders are actually thinking better because of it? In this week's interview, I spoke with someone who researches exactly that question: Professor Nathan Mauck, Professor of Finance at the Henry W. Bloch School of Management at the University of Missouri-Kansas City and co-author, with Bob Greene, of the new book "AI Co-Thinking: A Framework for Working with AI".
Why most organisations are dabbling
Nathan Mauck started with a practical question. He asked organisations how they use AI, because he wanted to make sure that what he teaches in the classroom matches what employers expect. What he found was a lack of strategy. Many organisations were taking a dabbling or wait-and-see approach: a few pilots here, a group experimenting there, and a great deal of hesitation, driven by the pace of change.
Together with Bob Greene, who comes from consulting, he developed a framework for using AI to think better and make better decisions, rather than focusing only on automation or prompt training.
Why prompt engineering is the wrong place to start
When companies begin with AI, the first step is almost always a prompt engineering workshop. I asked Nathan Mauck why that is the wrong starting point.
His answer: you do not need to be a technical expert. You can simply talk to AI in plain language. Precisely because it is so easy to create output, you can produce a great deal of material that improves nothing, unless you are clear about what you are trying to build and why. The starting point should therefore be the problem to solve: What exactly are you trying to address, and why does it matter? Learning AI by solving real problems is also more effective than learning it through hypotheticals.
Using AI versus thinking with AI
Most people use AI like a super search engine, or to automate workflows. Both are useful, and Nathan Mauck does both. But neither is thinking with AI. In fact, he sees a real risk in leaving the thinking to the machine.
Co-thinking starts with our own thinking process. What problem are we addressing? What are the inputs? Which questions do we need to answer? What are our assumptions, and what logic and evidence do we bring? Only then do we bring in AI, for example to find additional evidence, to challenge our assumptions, or to check whether we are applying our own decision process consistently. Used this way, AI acts like a trusted colleague rather than an answer machine. Many people want a button that delivers the answer. That is exactly what co-thinking is not.
The real risk: high stakes and low expertise
I shared an example from a recent assessment for a job applicant. The task allowed any tool. The candidate asked an AI to analyse population data in Excel and pulled figures from a so-called think tank whose data was anything but neutral. Smart use of tools, poor judgement on sources.
Nathan Mauck sees this as the bigger risk of AI, larger than the cheating debate in academia: How do you know whether the output is good? He and Bob Greene use a two-by-two matrix: the stakes of the decision, low or high, against your expertise, low or high. With high expertise you can quickly judge whether the output is reasonable. With high stakes and low expertise, you cannot. In that quadrant, the right use of AI is to build your expertise first: What questions should I ask about this data, its sources and its validity, and what would a good answer look like?
A practical book for busy leaders
Leaders often ask whether such books are written only for academics. Nathan Mauck describes "AI Co-Thinking" as short, accessible and practical, readable in one or two sittings. Neither author is an AI technologist. They describe themselves as power users. The longest chapter consists entirely of real use cases, from personal coaching to strategic planning and launching a new business. As he puts it, seeing is believing.
Where to start
For leaders who feel they are already behind, his advice is reassuring. Do not try to answer the big macro questions first. Pick one tool, take a real and preferably small problem you are working on, and describe it as you would to a colleague. Check whether the AI has understood you, correct it and iterate. Aim to move your thinking forward in about 20 minutes, rather than solving everything with one prompt. Treat AI like a new analyst: you would give context and expect questions, not hand over a spreadsheet and walk away.
What Leaders Should Do Now
1. Start every AI initiative with a clearly defined business problem, not with prompt training.
2. Distinguish in your organisation between using AI and thinking with AI, and reward better thinking, not activity.
3. Apply the stakes-and-expertise matrix before you trust any AI output on an important decision.
4. Check the sources AI draws on, especially when you lack expertise in the field.
5. Pick one real problem this week and work through it with AI as you would with a trusted colleague.
Apply, apply, apply.
Niels Brabandt
---
More on this topic in this week's videocast and podcast with Niels Brabandt: Videocast / Apple Podcasts / Spotify
For the videocast’s and podcast’s transcript, read below this article.
Want your leaders to deliver better results, sustainably?
Excellent leadership makes the difference.
Let's talk: NB@NB-Networks.com
Answer guaranteed within 24 hours.
Contact: Niels Brabandt on LinkedIn - https://www.linkedin.com/in/nielsbrabandt
Website: www.NB-Networks.biz
Podcast and Videocast Transcript
Niels Brabandt EMBA MBA MSc
AI is here, it came here to stay. The question is: how do you use AI? And of course you wonder, well, anyone has an opinion on AI. However, we have someone who's researching on the matter with us here today. Hello and welcome, Professor Nathan Mauck.
Nathan Mauck
Hey, thanks for having me. Excited to talk all things AI here.
Niels Brabandt EMBA MBA MSc
Thank you very much. Thank you very much for taking the time, sir. You're a professor for finance, so you definitely know numbers. And probably that's pretty important also when you deal with AI.
Niels Brabandt EMBA MBA MSc
You now wrote the book "AI Co-thinking." What was your motivation at this point in time to write and publish the book right now, at this point in time? Because I think we are at a quite critical point with AI exactly in this moment in time.
Nathan Mauck
We really are. And my co-author Bob Greene and I were working together on AI. And so I come from the finance background, and Bob comes from the consulting background. And we were seeing a lot of the same things in AI adoption in organizations.
Nathan Mauck
And I always say, like, for me, one of the things that was interesting is I really started off to just ask organizations, "How are you using AI?" And I was really coming at it from a perspective of, I want to make sure that what I'm giving my students in the classroom is relevant to what employers are expecting them to be able to do. And then also in my own research, how am I going to leverage AI and how might AI be impacting the industry? These were kind of the questions I had in my mind.
Nathan Mauck
And then as I talked to more and more people, it became clear that there was not really a clear strategy. And a lot of people were kind of taking a dabbling or a wait-and-see approach. So they were either, "Hey, we've got these kind of few pilot type of programs, and this group's trying this with AI," or they were hesitant to jump in to AI. And I assume because of the pace of change in the technology was causing that.
Nathan Mauck
And so what we decided was, you know, Bob and I were really developing our own framework for how we want to work with AI and what we found to be useful. And so that was really the vision there, is to really get a framework for how we can use AI to help us think better, to make better decisions, instead of just automation or getting, like, more prompt training and things like that, which there's a lot of that content out there, which is great, but, you know, a little more strategic thinking with AI.
Niels Brabandt EMBA MBA MSc
Excellent, because that's my next question. Because usually we're, okay, AI, what's the starting point? And the starting point always is, let's do a workshop on prompt engineering. That is basically the starting point. So why would you say that prompt engineering is the wrong place to start from?
Nathan Mauck
Yeah, so first of all, just one of my favorite things about working with AI is you don't need to be a technical expert. And while I think prompting strategies can help, you can really just talk with AI. I mean, the conversational nature of it, the fact that you can talk in plain language and get relevant output is really valuable.
Nathan Mauck
But partly because of that, it's so easy to create output and to build things that if you're not really clear on what it is you're trying to build, and maybe even more importantly, why you're trying to build it, then you can get a lot of output that may not actually improve anything. And so we think the place to start is the problem to solve.
Nathan Mauck
So being really clear about what problem are you actually trying to address, why does it matter, and then you can get into, you know, working with AI. And also, to me, it's just a better way to learn how to use AI and how to better leverage it is instead of on hypotheticals or more of a traditional training, it's like, "Hey, let's use it to solve problems and kind of build our own personal philosophy that works for us."
Niels Brabandt EMBA MBA MSc
Absolutely. And often people say, "Hey, I'm using AI," and you make the very clear distinguishment between using AI or thinking with AI.
Niels Brabandt EMBA MBA MSc
So what, in your opinion, is using AI? What is thinking with AI? And how to distinguish between it?
Niels Brabandt EMBA MBA MSc
And of course, how to actually think with AI? Because let's face it, when my AI tells me, because of course they know, these companies know which terms people like to see.
Niels Brabandt EMBA MBA MSc
So any AI says, "Oh, I am thinking, I am reasoning."
Niels Brabandt EMBA MBA MSc
And let's face it, when you follow algorithms, thinking and reasoning is not really led by algorithms, because otherwise we would have different human behavior on this planet.
Niels Brabandt EMBA MBA MSc
So quite a complex question here. So what is using AI for you? What is thinking AI for you? And how do I actually think with AI?
Nathan Mauck
Yeah, great question. And on the using AI, I always say, I think a lot of people use AI like a super search engine, like a super Google, you know, they prompt it, they ask, which can be really helpful. Or people think about using AI to automate. So I've got this workflow and I can work with AI to automate. Again, very, very useful. And I do a lot of this type of work with it, but it's not really thinking with the AI.
Nathan Mauck
And in fact, you know, one of the things that has kind of been on my radar is I think there's a lot of risk to leaving the thinking to the AI. I think, you know, we need to provide the thinking and then work with the AI. And so that, you know, how do you think with AI? I think it starts with how do we think in general, which we can all get better at, let's say, as you're pointing out, like humans are not always the best about following a rational process.
Niels Brabandt EMBA MBA MSc
Let's put it that way, yeah.
Nathan Mauck
So it's actually kind of an invitation, I think, to revisit our own thinking process of let's get as clear as we can ourselves and kind of start with this, what problem are we trying to address? What are the inputs? What questions do we think we need to answer? What are the assumptions? What kind of logic and evidence do we bring? And then you start bringing in AI to help wherever you might need support.
Nathan Mauck
So if you need additional evidence from outside, maybe AI can assist with that. If you need kind of like brainstorming on assumptions of, you know, help me think through my assumptions, what am I not considering? Or even just laying out your own decision, kind of like matrix or algorithm or process and asking AI to kind of check you of, "Hey, am I applying? I say this is how I think. Am I actually doing that here?"
Nathan Mauck
This kind of mirroring that AI can do can be really helpful and actually make us better thinkers rather than replacing our thinking. It's the, that's kind of the vision of this co-part is it makes the AI more useful. It makes us better thinkers if we kind of treat it like we would, let's say, a conversation with a trusted colleague rather than an answer machine. Or we're talking to one company and they said, a lot of people want like a button to push with AI. You know, give me the button that gives me the answer. I just want to push this.
Niels Brabandt EMBA MBA MSc
Yeah. And exactly that's my next point here. When we talk about creative and critical thinking, I give you something which I experienced yesterday. So someone applied for a job and they were given a certain task. The task was supervised, but they were allowed to use absolutely anything, whatever they think. And there were quite simple tasks, but it was an analysis about a certain population, a certain area. And I have never seen someone in that situation as that person did yesterday, suddenly sitting in front of Microsoft Excel, opening a certain AI and said, "Oh, AI, please do an age lookup on that data. Hey, can you get me a diagram out of that?" And pulling data from a think tank.
Niels Brabandt EMBA MBA MSc
And the term think tank, of course, sounds fantastic. However, that think tank was far, far away from delivering neutral, not politically intended data. And then he presented the result and we had to say, "Hey, look, smart solution. If you don't know Excel, use the AI perfectly fine." But you gave us data, which is clearly biased in a certain, "Oh, I didn't know." So how can AI actually make me a better creative thinker, a better critical thinker, when many people at the moment now fear we're going the different way at the moment?
Nathan Mauck
You really hit on, to me, the bigger risk of AI use, which in academia, a lot of folks have been focused on cheating. You know, our students aren't going to do the work, they're just going to use AI. And that's an interesting discussion. But to me, the bigger problem is, let's say, okay, you are allowed to use AI, you use it, you get the output. How do you know if it's good output? You know, how do you know you can trust it? And especially like, let's say you don't really know the field all that well.
Nathan Mauck
So one thing that Bob and I have been working on is this kind of just this two-by-two matrix where you can kind of think about the stakes of the decision. So low stakes, high stakes, and then your expertise, low and high. Right? And so if you've got high expertise on a high stakes decision and you ask AI for almost any kind of support, you can probably pretty quickly judge whether or not that output is reasonable, if it's kind of pulling the right context. But if it's a high stakes decision, you have low expertise and you're asking for an answer, how would you even know if the answer you're getting is good?
Nathan Mauck
And so we think you can really use AI in any of those quadrants. It's just the way that you use it and the way you think with it is very different. And so, you know, like in this example of, let's say, high stakes, low expertise, you know, it makes sense to help maybe AI build your expertise of what kind of questions should I be asking? You know, before I dive in, in your example, before I dive in and start running Excel, what do I need to know about this data?
Nathan Mauck
What questions should I ask about the data, the sources, the validity, the, you know, how do I, how would I know what a good answer looks like and kind of, you know, use it to help build expertise? Whereas if you were the expert, maybe you could jump in or maybe you'd already know, just your human judgment would tell you this data has got issues and I can't just jump into, can't just jump into Excel. So I think there's like different vulnerabilities to, I'd say, like getting unhelpful output depending on your expertise and how important the decision is. You know, so yeah.
Niels Brabandt EMBA MBA MSc
Absolutely. And of course, one question I have to ask because when I prepared for the interview, and of course I asked a couple of people, one question which came up when people said, "Look, I look at the people, CVs who wrote this and they're very good at what they're doing. However, what kind of read am I going to expect? Is it very academic, high level for people from basically Harvard, Princeton, Yale, and below that you better don't open the book? Or what do I have to expect?" Because when people say people love frameworks, a framework for working with AI, what can I expect when I read the book? What kind of read is it going to be?
Nathan Mauck
Yeah, so it's, we've set it out to make it very readable, very accessible. It's pretty short. We want people to get through it. Depending on your average, you know, read length, maybe one sitting or two. It's supposed to be really practical, partly because, you know, both Bob and I are not AI technologists. You know, we're not technical experts. Our expertise is in other areas. We're more like power users who are, let's say, excited about working with AI, but also see some of the potential drawbacks. And so it's really practical.
Nathan Mauck
To me, my favorite part of the book is we've got a fairly long, the longest chapter is nothing but use cases. So it's examples of Bob and I actually working on problems, everything from kind of personal development and personal coaching to more like strategic planning, launching a new product slash business, you know, repetitive processes, but very different use cases that span not just finance or strategy or anything else, but from personal to organizational.
Nathan Mauck
And I think I've got this, this comes from my, you know, my classroom experience, but I always have this like seeing is believing philosophy, which is I can talk about AI co-thinking, but when you see it in action, you're like, "Oh, okay." And actually, like when we do some workshops on this co-thinking, and usually when we work through an example, people are kind of like, "Oh, yeah, it's like, it's just, you know, kind of taking how I problem solve and, you know, leveraging this new tool and in a conversational way, like I would if I had a new colleague sitting at my desk next to me that I, you know, trusted and respected, but maybe didn't have all the context of, you know, the problem and I'm working with them." And there's kind of associated pros and cons of that.
Nathan Mauck
And so it's, we also call it, it's kind of like a podcast in writing. And that the way that Bob and I wrote it was we just had, we call them just these riff sessions of what are we working on, what are the problems, who are we talking to, what are we seeing? And then us just making sense of that and then aggregating it into what we were seeing in ourselves and others. And so that's the approach we took.
Niels Brabandt EMBA MBA MSc
Yeah. So it's a very practical book for anyone to use immediately. So that's an excellent, that's of course excellent news here.
Niels Brabandt EMBA MBA MSc
And of course, one question I have to ask as the final one, when people now say, "Hey, I'm about to buy this book." And of course, anyone listening here will buy the book. We all agree on that one.
Niels Brabandt EMBA MBA MSc
So when people now say, "Hey, I saw this AI and I already feel quite far behind. I just don't know where to start." There's an abundance of tools and it looks like even if I try to catch up, the development is 10 times faster than I will ever be able to process.
Niels Brabandt EMBA MBA MSc
If someone now says, "Hey, I think I need to onboard this for my personal skill set at work, where to start?" What would be your top one, two, three tips here as the final question for the interview?
Nathan Mauck
Yeah, this is, I think, a huge issue right now. And one thing is there's so much discussion about AI, right? It's everywhere. It's in the headlines. Everybody talks about it. You get kind of really macro level, what's the impact going to be on labor? And, you know, very micro, you know, model details. And I think it causes people to just freeze and not jump in. And so maybe the obvious, but important thing is if you haven't started, get started. And, you know, it's time to jump in.
Nathan Mauck
And I think rather than worrying about answering these really big macro questions, the much more practical issue is how are you going to use AI now, right? With the tools the way they are now, and people worry about this tool or that tool, just pick one and take a problem you're actually thinking about and working on and just describe your problem just like you would to a friend or anybody else of, "Here's a problem I've been thinking about. Here's what I'm working on." And then at first just check and see if it's understanding. You know, just ask for like, "What's your read on this scenario? You know, what are you seeing here? What are you hearing from me?" And then if it doesn't understand, you can kind of iterate and correct and work back and forth.
Nathan Mauck
But, you know, I would say pick a real problem and probably pick a smallish problem to start with. You know, maybe something that you could move your thinking forward in 20 minutes. And, but I think if you take a view rather than like, "I need to complete a task or get some output," and instead have the view of, "I want to move my thinking forward on some subject," right? I don't need to solve the problem in one sitting or in one prompt, but I just want to get a little clear in my own thinking about this. And that's, let's say, the goal of the session.
Nathan Mauck
Pick a real problem and very much treat it like a conversation, you know, a back and forth instead of, "I'm going to make one prompt and it's going to solve all my problems." I was thinking about it's like, I try not to get too finance-y with my examples, but it's like you hire a new analyst to think you could just give them, "Here's a spreadsheet, go do this," and they're not going to have questions. There's not going to be some back and forth, or you might not want to give them some context on why am I having you do, you know? You know, I think we, if we treat AI much more like we would a colleague, we'll be much happier with the output that we get.
Niels Brabandt EMBA MBA MSc
Yeah. I think these are the perfect final words. AI co-thinking, a framework for working with AI. At the end of this interview, there's only one thing left for me to say. Professor Nathan Mauck, thank you very much for your time.
Nathan Mauck
Thank you so much for having me.