How Leaders Decide What Humans Do And What Machines Do

AI is accelerating every business cycle, but the real advantage is not adopting more tools, it is making better decisions than competitors who have the same tools. 

We sit down with Dr David Feavearyear, procurement leader and author of *Organizational Decision Making in the Age of AI*, to sort hype from clarity and build a practical way to decide what should be done by machines and what must stay human. If you have ever watched a boardroom go quiet because “the AI recommended it”, this conversation gives you language and structure to push back 

TL;DR / At A Glance

• The shift from human-only decisions to shared human machine decision rights
• Where machines excel through high data sufficiency and repeatable logic and where humans still win through creativity, cultural judgement, and followership
• Bias versus experience and why language changes how we judge decisions
• Why boards become risk averse around AI and how to challenge AI outputs
• Curiosity, EQ, communication, and culture as leadership differentiators
• Responsible AI and ethical deployment as a cross-disciplinary challenge

We talk about decision pressure and the end of the default human-only world. David explains how “data sufficiency” and repeatability shape the right decision owner, why deterministic automation can beat probability-based AI in many cases, and how bias can be reframed as experience depending on outcomes and context. 

We also dig into the human edge: creativity that imagines a different future, and followership that earns buy-in for ideas others cannot yet see. As AI becomes ubiquitous, those soft skills become a genuine strategic moat. 
 
 Procurement becomes our real-world test case for AI and automation, from invoice matching and transactional workflows to autonomous negotiation in tail spend. We also explore what could change next, including contract automation for routine agreements like NDAs, and how responsible AI and ethical governance will shape leadership expectations. 

You will leave with a clear starting point: map the decisions that matter, build your data, technology, and talent strategies from that map, and invest in curiosity and upskilling so teams feel excited rather than threatened. 
 
 If this helps, subscribe, share it with a leader who needs it, and leave a review so more people can find the show. 

What decision in your organisation should never be automated?

LinkedIn: Dr David Feavearyear

David's Book on Organisational Decision Making: https://www.amazon.co.uk/stores/Dr-David-James-Feavearyear/author/B0F895QZQD? 

Subscribe for more conversations like this, share the episode with a leader who is wrestling with AI driven change, and leave a review with the one decision you think should always stay human.

Transcript

The New Executive Decision Pressure

Kieran Gilmurray 0:03

Welcome everyone. Today we're stepping into one of the biggest leadership questions of our time. How do executives make smarter decisions when AI is accelerating everything around us? I'm joined by Dr. David Febrier, procurement leader, academic, and author of organizational decision making in the age of AI. His work focuses on the intersection of human talent, organizational culture, technology, and strategic judgment. This session, though, is not about hype, it's about clarity. How do leaders cut through the noise and make decisions that actually create value? And what does leadership look like in organizations where humans and machines work side by side? Let's get into it and waste no time. David, welcome, sir. How are you today?

SPEAKER_00 0:51

Thank you very much. I'm well, although I can uh honestly say I've never been described as an academic before. But uh and I'm not sure how I feel about it. So let's uh let's get into this.

Kieran Gilmurray 1:00

Well, it was meant as a compliment. So we'll work up from there. Well, let's see how academic you are. David, your book argues executives are facing a new type of decision pressure. So what's changed?

Where Machines Make Better Calls

SPEAKER_00 1:16

Well, listen, I think um I think for most of our evolution, the reality is that we've had to be reliant on human beings to make decisions. Um and actually, when you look at um a lot of academic theory about organizational behaviour, corporate governance, all those kind of good things, most of it's orientated around helping organizations ensure that the people that are making decisions on their behalf, as it were, are making decisions that are in the interests of the organization and its shareholders and stakeholders, as opposed to decisions that might be in their personal interest, if you like. And alongside that, you've got the reality that there are lots of things that are brilliant about human beings. Um, and we can talk a bit more about what makes us great decision makers in certain circumstances. But the reality is we are all bound by our cognitive ability, by time, the amount of information available to us, um, and we're all driven by things like ego, emotion, intuition. And therefore, when we make decisions, there's an awful lot going on, um, which in some instances is brilliant, in other instances is a challenge for us. And corporate governance, as I say, historically, it's been about how you wrap parameters around that to give us freedom within a framework so that we can make decisions that are hopefully orientated towards the organization's best interests and makes us the best version of ourselves. Now, what I think has changed in the last, I don't know, decade, but certainly accelerated in the last couple, is there's now an alternative, right? So, as opposed to having to default automatically to humans in the absence of choice, you now have some pretty cool tech that's emerging that actually enables us to think about machines as decision makers. Um, and so one of the things I think is fascinating and I think is a is a challenge for leaders in organizations is where should decisions be made by the likes of you and I? And where might we be better served to use machines and how do we strike that balance?

Kieran Gilmurray 3:19

Yeah, I could there was you wondering whether you're an academic and you led with academia as well. And when it comes to that decision making, I think it's down to my pay rise. That's when the real decision should be left to me and not someone else. So, where does human judgment then still outperform AI and why will that matter more this year in 2026, not less?

SPEAKER_00 3:42

Yes, I might actually take this the other way around uh if that suits. So let's talk about where I think machines are really useful in decision making. My view is where you have lots of data and lots of volume, and you have what I would call high data sufficiency, right? So decisions that actually require highly repeatable logic, where if you had sufficient data, you could make really well-informed, rational decisions on a repeatable basis, right? In those circumstances, machines are brilliant, right? They're tireless, they're impartial, assuming they're built properly. Um, and actually they can crunch across structured, unstructured, and semi-structured data in a way that just is beyond most human beings, right? So where you've got high volume and higher data sufficiency, you really want to be using tech. Um, and we can talk about that more in due course. At the other extreme, where you've actually got modest data sufficiency and actually not necessarily less volume of data, but just different types of data, so softer data, for example. So, for example, if you're making a hiring decision, right, that's a situation where most of us would not hire a human being purely on the basis of what the data told us. I might be able to read your CV, but actually within 30 seconds of meeting you, I'm going to have formed a whole bunch of qualitative judgments about you that are actually far more important in terms of cultural fit and all of that kind of good stuff. Now, the reality is in those instances, machines can help us to augment, they can help shortcut that decision-making process. But there's something innately human about that. Um I think anything that requires creativity, establishing followership, those types of decisions, they're things that actually are uniquely human. So there are situations where all the data can be telling us one thing, but our instinct, our experience, our judgment tells us to do something different. Now, how you quantify that, I think, is you know something we could talk about as well. But fundamentally, they're the two extremes. You've got instances where you've got higher data sufficiency, you want to be using tech, and where you've got much more modest data sufficiency and softer factors come into play, that's I think where humans really excel. And the reality is organizations merge those two things, right? Day-to-day, there are highly rational decisions, and there are things that by definition are going to be much softer.

Kieran Gilmurray 6:13

It's interesting though, even that when we you know tear off the plaster a little bit more, because when we apply our human judgment to it, arguably, not only do we do the softer skills, you know, but do we apply our own biases? And when we talk about you know data models where there's uh a quantity of data, you could argue our large language models, which are evidential of having collected more data at any point in time than probably any computer system to date, have tended to hoover up digital evidences of our human bias as well. So I'm kind of wondering here, you know, how do we actually combine the best of both? You mentioned, you know, creativity and followership as strategic differentiators in your last. Where does that become valuable as AI itself actually gets stronger? And we're able to see now and understand that we have biases that we weren't aware of. Now that we built these systems and everybody goes, but the AI is biased. No, my friends, it's a it's a it's a testament or a judgment to where we're at. So you explain that creativity and followership as a strategic differentiator within the context of all of the madness that I've just mentioned.

SPEAKER_00 7:30

Yeah, so I mean I so if you take bias, right? Bias, um I'm a big believer in language, right? So you can describe something as bias, and it's framed real relatively negatively, as it were. I've seen this thing in the past, I expect it to continue in the future, therefore there's a bias inherent in my decision making. But you could describe that as experience, right? And experience is framed positively, right? So I've seen this thing in the past, I expect it to continue in the future, therefore, my experience is valuable, and I use that experience to inform my judgments. Right? You're basically describing the same thing using language in two different ways, and depending on how you're framing the conversation, it can be framed as either a negative or a positive thing, right? So that would be point number one. Point number two on in terms of creativity and that kind of human decision making. If you take someone like uh Steve Jobs, and I know he gets used to death as an example, but I think it's with uh with good justification. If you were looking at the Walkman, right, you'd have never arrived at the iPod, right? Because he effectively started with a challenge or an opportunity and worked backwards, as opposed to thinking about what we have today and linearly improving it, as it were. Now, in the process of doing that, there's a creative imagination, if you like. You creatively imagine the way the world could be at a future state that's very different to what you do today, right? And that's an inherent part of decision making. But even if you've done that, right, you then have to be able to walk into a boardroom or a management team and you have to explain that you've got this brilliant idea for something that's far beyond their current understanding of the way that that outcome is performed. And you have to engender followership, you have to get people to buy in to that idea, to that purpose. So there's an element of decision making, there's an element of influencing, and to drive actual outcomes, you have to be able to do both. Now, if Steve Jobs didn't have both of those things in spade, we wouldn't have the iPod, we wouldn't have the App Store, because both of those things required that creative imagination of the future. I think at the moment machines are pretty linear, right? They look at what's come in the past, they extrapolate forward from that point. And if you've ever read The Innovator's Dilemma, that's the point at which you actually overpitch against the actual requirement. And those as you use technology fundamentally differently and reimagine the world, that's where you get supplanted. So again, you know, it's quite a balance working out where this is a superpower and where it causes you problems. But the best examples of human decision making are normally that ability to creatively imagine a future state that's different to the one we're in today.

Stop Defaulting To AI For Everything

Kieran Gilmurray 10:22

Yeah. I I go back over a couple of things you said there. You know, what's the difference between bias and experience? Usually the outcome, so if you've got it wrong, it's bias. If you've got it right, that's called great judgment and great experience. You know, that's first. I think the other one there, it's interesting because a lot of people point to Steve Jobs, you know, quite rightly, uh, gifted, no doubt. But if every human was gifted, wouldn't we have a multitude of examples where we point to the few, not the many? And to a degree, uh, and you have met many of these out there in your career, we will not name them at the moment. They were badly human-prompted. In other words, whoever was the idiot who taught the idiot who taught the idiot is now making judgments based on past data patterns to do it. Uh, so I do worry about humanity, you know, as well, because we keep going to AIs, not AI's prompted, AI is based on the on the world of data. But if you ever do an insights profile, you've got your green and your blue who look backwards to look forwards, and you have very few red and yellows who look forwards and then can't work out how they got there. And gorgeously, I mean that one as well, you know, as opposed to anything else. So we we live in interesting times. So uh what can we do then? Because you mentioned, you know, going into a boardroom, being able to emotionally tell a story convince many leaders can't do that either. And when they do get into that room, a lot of boards are very risk averse because now the dynamic is saying, you know, make a mistake and you're out as a corporate leader? It's worse than the times. I think I think the time cycle of a premiership league manager is nine months. The time cycle for a really high prominent CEO is probably aiming in that direction. There's with a parallel ends. So what habits do leaders need to have to make better decisions in AI heavy environments when everybody has access to the same tech? But not only that, when you see some things going to boards and you say, well, AI did it, everybody clams up and goes, Well, it must be right because AI did it. So you're almost running two parallel things. I've got all the tech, I can make all decisions, and you know, therefore we're we're all going to end up in the same place, but we're still ending up in front of a risk-averse board who half the time are going, I've heard that it's AI, and therefore I'm not competent yet to understand it. But I'm certainly not going to argue with that black box in case I look further.

SPEAKER_00 12:38

Yeah. I mean, I think um so first thing I'd say is um despite the somewhat misleading title of my book, I don't think all technology has to default to AI, if that makes sense. So I think um when I think about the world, I think about what we should automate. And when we automate something, AI is one of the vehicles available to us. But you don't have to use AI for everything, something very deterministic, and therefore you don't need a probability-based AI tool to do it. You just need a bloody good decision tree, for example. Um, so I think I think that's point number one. I think people have to default, uh, stop defaulting to every answer being AI now what's the question. And the second thing I think people should do is really spend the time to rationalize where technology helps them to make decisions and where it doesn't. And do that in a really considered way, right? So rather than uh react to whatever you see in front of you, you know, if you take a functional area or you you you spend a bit of time workshopping, the decisions you make on a day-to-day basis, and of those decisions, you know, whether they hit the criteria for being automated and whether they hit the criteria for human talent, even if you just did those two things, you'd start to have the basis of both a talent strategy and a technology strategy. And then if you said, what decisions sit with humans today, but actually if we had the right data, could we start to automate? Then you start to get a data strategy as well, right? But what you're most importantly doing is you are systematically thinking about the decisions you make as a company and where it's suited, best suited to either humans or technology. And then you go and find the technology or you build the technology as the case may be. But what you don't start is with the technology back and then work out what problem it's trying to solve, if that makes sense.

Procurement Automation Gets Real

Kieran Gilmurray 14:27

It so does. I I do wonder and worry about that though. Let me explain that. If you look at it, I think it was Bain BCG, I can't remember which they said 72% of us make bad decisions every day, or businesses do, which basically means us. And they this was around the topic of decision intelligence. Basically, it says there are four types of AI descriptive, diagnostic, the insight, what happened, why it happened, predictive and descriptive, why did it happen before? And that's part advocating for that particular science. The bit that I think businesses struggle with today is not necessarily the data, not that they can't do some science with it, and let's put the word science as opposed to some math because it sounds so much better, but it's actually forming the right question in the first place. And before the technology, I would love to see a culture of curiosity being built inside of an organization. So, for example, if I say, you know, how many uh customers churn this month? Well, I'm going to get some level of an answer. And let's say the answer is 400, but I've actually learned to ask a really great question, you know, how many customers churned in region X and what were the top three reasons for them, and what can we do about them? Then long before I get to, you know, attempting to put AI or deterministic or whatever other simple analytics around the particular problem, if I understood where my business was going and learned to ask better questions, and if everyone learned to ask better questions every day, we'd come up with a better business. Because that's what a business is. A business is the sum of the decisions that everybody makes every single day. And therefore, the better the question originally, the better the strategy, the better the question, and then put decision intelligence around it, you're more likely to come up with a better answer than you know, the computer said A or the computer said B, or even forecasting forward is that it might be C or D up to a percentage. So, how does all this, how do we get all this to work then, David? So, for example, you are a procurement specialist, you know, procurement is an area where often you know operational decisions meet financial reality, for want of a better phrase. So, how's AI changing procurement decision making, and then doing a little bit of predictive and prescriptive analytics yourself or verbal gymnastics? What will normal look like in about two years from now that feels new today?

SPEAKER_00 16:47

Yes, I mean I I think to your point, procurement's uh a really good example of both extremes of uh of what we talked about. So, on the one hand, you've got some highly repeatable type activities that are frankly very tactical, very repeatable. You want to be automating the hell out of. Um, and that's things like you know, 10 years ago, we were still manually converting purchase requisitions to purchase orders, for example. Um I mean, which which staggers me. Um, and then we were manually looking at invoices to determine whether those invoices should be paid. You know, fortunately for the large part, we've now automated a lot of that, so we're no longer doing the conversion. You know, machines do that for us. If we've got the right upstream processes, you can three-way match. So when an invoice lands, it takes the requisite information off the invoice, compares it to the purchase order, a good receipt note, and it just gets paid, right? So take something that used to cost you, I don't know,$25 an invoice and turns it into something that costs 50 cents, right? And you should be driving more and more of those things that used to have a decision a human in the loop should be automating. So I think there's really good examples of stuff that we're doing today that no one would argue should be done. Um there's there's no value in having a human sit above that, aside from the laziness in automating the process in the first place. Um in terms of where the future's going, I think that the future is pretty cool in this space. Um, so yeah, people are starting to look at autonomous negotiation, for example. So yeah, if you've got a uh a five year manual coming up with Microsoft, you're probably not going to use a machine to do that. But if you've got highly repeatable commodity-like tail spend, you don't need a human in the loop. Um you can you can actually use intelligent agents to start doing that for you. And there are examples of organizations that are doing that at scale with the existing technology. So that's not something coming down the line, that's something that already exists. Just takes a bit of bravery and a bit of imagination to uh to use it. Um but I think one of the things that uh I think could look really interesting in the future to your point about something you can imagine that's not here today quite yet. I'd love to see the majority of contracts automated, um, which doesn't mean you no longer negotiate them, but it means you use machines to automate them. So you know, take an NDA. I don't know how many millions of person hours a year are spent negotiating NDAs, but fundamentally it's a pretty straightforward principle that says I'm gonna give you information, you're gonna protect it, vice versa, you'll apply appropriate standards. And yet we have teams of lawyers that can engross themselves in negotiating that for days and weeks on end, it slows down business, and I would argue add almost no value. So why couldn't you automate that given how formulaic it is? Um, and then once you've done that, why can't you extend into other types of contracts? So, again, you know, how much time do people spend arguing the toss over limits of liability, warranty provisions, indemnities? Yeah, there's tons of data that exists out there to train machines on it. You can set your risk parameters so that it always gives you a certain level of review when you receive something. But I think that the days of having big teams of highly paid legal specialists sat across relatively repetitive contracts, I would like to think, uh, are starting to come to an end. Um, and then when you start to connect the dots between this, yeah, you should end up with smaller teams of higher impact people that do the cool stuff we were talking about, but where the more transactional high-volume stuff has just been automated?

Kieran Gilmurray 20:33

So we're moving up, for one of a better phrase, see if lawyers, by the way, in tiers, uh, listening to your removing their substantial income out of that particular activity, and many more, it must be said, too. Uh but we do do our lawyers at the best of time. So this sounds like you know, we're moving up the intellectual or work curve. So, what leadership skills will define high-performing organizations in the next three to five years?

Soft Skills And Responsible AI

SPEAKER_00 20:59

So I think uh the the word you used earlier, I think, is key. So curiosity. Um, I think um, I mean it may sound like a soft skill, but I think we are moving into the uh the era of soft skills differentiating leaders. So curiosity is first and foremost amongst that, because if you're not curious, then you can't explore the brave new world in the way that we're we're describing. And then I think it's going to be organizations that use technology efficiently to drive that kind of transactional work, that highly repeatable work, um, but then really invest in their talent strategy alongside that to bring in a different type of leader, right? A leader that has really high EQ, that has really outstanding soft skills, you know, influencing skills, communication skills, somebody that can empathize, for example, but without effectively um without being able to forego giving hard messages, because again, that's something that humans are uniquely good. That you know, relaying feedback in a way that's both meaningful and impactful is a hard thing to do, but it's a necessary skill if you're going to drive improvement. So I think once you get to the point where technology is more uh ubiquitous, where everyone's effectively got that, it's no longer a differentiator. What's left? What's left is the human element that sits on top. Um, and that's really hard to replicate. Yeah, a really effective culture, really strong leaders, um, a really powerful vision that the organization are unified behind, they're the things that will differentiate us in five years versus some of the things that maybe five years previously have been the uh the co-differentiators.

Kieran Gilmurray 22:41

Yeah, I like that. Do you know what I mean? Maybe it's just the human in me likes it more than anything, but I do sometimes worry and wonder, you know, um, if we spent as much time getting excited about people as we did technology, if we invested as much in the right people as we did co-piloting or another technology, you know, wouldn't organizations be much better places? Now, that's not a call to arms for all the philosophy students who are now rubbing their hands together with the words, I told you so. I will be a leader someday, now that my curiosity is coming up to the top. Although sadly, I was talking to a friend of mine recently who is a clinical psychologist who did say that only 36% of us are emotionally intelligent, which was rather shocking. Uh so she said, When you're in your room with three people, you know, look left and look right, feel sorry for two, but recognize only one of you is actually correct and you haven't quite realized which one it is, you know. So fantastic. So if you're given advice. Oh, sorry, Dave, go.

SPEAKER_00 23:42

It's interesting, Kieran, because I think um it could be a great era for philosophers. Um and and I say this with a slight bias given that my first degree was in philosophy. Um, but AI, you could argue, is born out of philosophy, right? And if you if you think about the world we're marching towards where things like um responsible AI, ethical AI, you know, how you how you deploy this stuff within an organization so that it doesn't just deliver an outcome, but it does so in a way that is ethically and responsibly acceptable, as it were. Yeah, do you want to trust an engineer to do that? Um, or are there a broader set of skills that you need to be thinking about and taking into account when you strike that balance? So yeah, I think uh I think it could be a really interesting skill set in a few years.

Practical Board Advice And Closing

Kieran Gilmurray 24:31

Yeah, I think I'm gonna have to get a lawyer and a software engineer on to counteract that uh philosophy bias that's entering into this interview. There's a slight best of word interest in that, but uh slightly best of interest, yes. I think we may have to word with the uh with the education department as well to refund sociology, anthropology, and philosophy, and then and then uh actually we might survive over the next couple of years. So if you were to give David, just to close out, if you give uh you know boards and executive leaders, you know, a bit of advice, what would you be telling them to do over the next couple of years to improve their decision making, to improve their people strategy from what you're describing, to improve curiosity, or or what would you be telling them uh to get themselves fit for the future?

SPEAKER_00 25:19

Yeah, so I think um the key thing is um is to be considered, right? So rather than react to what's immediately in front of you, break break the problem statement down into its component parts. What what are the what are the decisions that really make a difference for our business, right? And it will be different for every business, and then establish to what extent do they hit the criteria we've talked about. So high volume, high repetitiveness, and higher data um data equivalence, as it were, that you want to be doubling the album tech. On the other side, and that's where you build your tech strategy. On the other hand, be equally considered about how you think about human talent. Think about the skills you want in the organization, start planning for it now, because those skills are gonna be in higher and high in demand. Um, and actually, how you upskill and reskill your teams is gonna be a really important part of how they feel about the future. You know, if you don't upskill and reskill, people are gonna feel intimidated and threatened by it. If you do upskill and reskill, they're gonna be excited about it and it's gonna propel your business forward. But the key thing is be curious, be considered. Um, start with the business problem back or the business opportunity back, don't start with the tech.

Kieran Gilmurray 26:33

Fantastic. Folks, we'll end on that. Look, if you find today valuable, follow David's work. Uh, read his excellent book, Organizational Decision Making in the Age of AI, and subscribe to some of the leadership insights that we bring out in podcasts and uh webinars like this. Look together, hopefully to goodness, we're building better leaders for a new era. And that has started today. David, thank you so much indeed. Awesome. Thanks for your time, Kenneth.