The Human Side of AI Strategy

Great Expectations Podcast – Episode 5: David De Cremer

Paul Sephton:

Hi, I'm Paul, and this is Great ExpectAItions. The Jabra podcast that cuts through the AI noise to bring you sharp, actionable insights from the world's leading minds. If you're steering your business through the age of AI, this is your map. We skip the hype and get straight to what matters. Smart thinking, strategic moves, and the human side of AI. Today's guest is someone who's asking the questions. More of us should be, not just what AI can do, but how it should be done. I'm joined by David De Cremer, Dunton family Dean and Professor of Management and Technology at Northeastern's D’Amore-McKim School of Business. He's also the founder of the Center on AI Technology for humankind at the National University of Singapore. David is a global thought leader in AI ethics, leadership and trust, and the author of the AI Savvy Leader, a must-read for anyone guiding teams through digital transformation. In this episode, we'll unpack what it takes to implement AI in a way that actually serves your business and your people. From aligning leadership and culture to building trust and spotting real success signals, we'll explore how to move beyond the hype and make AI work in practice and for humans. Firstly, Thank you very much for joining me today. I'm really excited to speak to you.

David De Cremer:

Thank you for having me.

Paul Sephton:

I think right now there's this huge narrative around humans versus AI and the human-AI interface. And yet if we zoom out from the scale of it all and zoom in on just what do we do now in 2025? There's this really important role that we're seeing in how AI can start to augment human capabilities rather than replacing anything. So, I'd love to get your take on what's the here and now that we can be doing with AI to enhance human capabilities rather than having a doomsday approach.

David De Cremer:

It's an important question, still today. If you listen to most businesspeople, CEOs, especially in the C-suite or in boards, they will still use the narrative of, is AI a threat to your job? And there's several reasons for this because especially leaders at the top, they still don't understand AI very well. They see it still as something magical. Whereas in reality, AI is a tool and it's a tool that we have to use. So in that sense, for companies, for businesses to make the best use of AI is to use it as a tool and see where it can be used to augment rather than replace. Second thing is jobs are a collection of tasks. So, AI is very good at doing tasks, but a job requires more. You have to bring the tasks together, you have to assign meaning to it, you have to negotiate, you have to influence people, you have to work with teams. It is not really about AI simply taking over because it doesn't understand the social dynamics. It can't work with these, so you still need a human workforce to deliver. People are starting to become more aware of it. And I think the biggest input to why people are becoming aware of this is really that, as you've noticed in the last two years, the return on investment when it comes down to applying or adopting AI in business has been relatively low compared to the investments we make in AI. We throw so much money at the development of AI and tech companies, are leading in this, someone like Mark Zuckerberg of Meta, he believes just throw more money at it and AI will be better. So the focus is completely on input. Now for a company, what matters is, are we more effective? Are we more productive? Are we increasing efficiency? Are we producing more? And in addition, the human workforce will still have to deliver this. Then the question really becomes, oh, it's not simply about replacing, it's about how the human workforce can use it. And along with this comes a really important realization. AI is not a strategy. AI is there to help you achieve your strategic goals. And this is where the shift is happening right now, in my view, that people are becoming aware of this now, companies will realize, just adopting AI is not making yourself more competitive. How do you make yourself competitive? By knowing, first of all, what is your organization about? Who are your stakeholders? What's your strategy? What are your goals? What's your purpose? So, for leaders, to move on and focus on augmentation. There needs to be a realization where they can align a certain AI comprehension, with their knowledge about their organizational purpose and their stakeholders. If you can align those, you'll see that AI adoption actually starts with a business question. What is it that you need to achieve and how can AI help in achieving that?

Paul Sephton:

How do you think you create some type of structure to that because there's so many layers there's the C-suite, the board, you've got people on all levels across different types of departments, whether it's finance or legal, marketing, R&D, engineering, and I'm talking about in a medium to large size company now. And then you get every strata and layer of management coming through with that. In such a rapidly changing space where everyone's half trying to figure out what it can be used for, half trying to figure out what's the best application for it, then you combine all of those layers of management day-to-day work tasks and functionality and talk about, okay, now we're gonna write the strategy as though it's like something that you put down and then there's a line in the sand and from here on forth, we are an AI company, but in reality it's a lot more complex than that. So how do you advise people to break it down into something more bite-sized to be able to it at the pace of change that it's operating at right now?

David De Cremer:

I always outline three ideas explaining why and how business leaders need to be involved. So first of all, it's bringing the idea of AI as a reality to the workflow. Why is AI needed? That's why I always say business leaders today in the AI era, they become narrators. They need to be AI savvy enough. To bring AI in the context of their business question, and at the same time bring AI experts and business experts at different levels, R&D, marketing, sales, and finance together, that they communicate and can collaborate. That's the first step. Once that idea is there, the second dimension is bringing the idea of AI to life within the job context. So that means you need to create a cultural change where you allow sandboxes for experimentation. So that's why most AI projects start as small pilot projects, this should not be in isolated ways. Each department can have their own sandbox and business leaders have to keep aligning it with the main business question. This means you need to install a climate of failure, which is fine, but also flat communication so that there's quick feedback and that people can see if this works or doesn't work. Now, the interesting part here is it's participative, so it increases commitment to the new vision where AI is a part. At the same time, they're experimenting within their job. So you're actually building the jobs of the future. I see so many companies always waiting for a whole list of new jobs to arrive. Well, basically, no. Each company is basically building those jobs themselves. But you need to work on this principle that I just said. You need to have that learning mechanism and quick feedback. The feedback aligns very well with how AI basically works. If you think about your GPT prompts as reinforcement feedbacks. So the same thing in your pilot studies where, and that's why teams have to be involved. It's a collective learning process because you get feedback much quicker. So you can adjust once and it cannot be isolated again because this is one of the big problems that we see is, if AI adoption is only seen as an engineering project, it's only done by people in IT departments, by tech people. they don't know the business side. it's so isolated that it never creates value across the entire company. So that's why these pilot projects are so important and our leaders so important to be the narrators to bring both business and tech together. And once you've got this culture going, the final step, now can the job with AI create value? So to facilitate that value creation process, it comes back to your original business question, who are your stakeholders? 'cause don't forget, everything goes so quickly, most businesses don't use the latest of the latest AI advances. Most of the AI is developed within a lab where there's hardly any stakeholders. So risks are low. But once you bring AI into business, you have different stakeholders, it becomes risk management. You can't run the risk to harm the interest of different stakeholders. from that point of view, the real value creation is how you bring it to the different stakeholders. Just as an example of what I see all the time in organizations is companies are struggling internally to have employees accept the AI and integrated across the board, but at the same time, they've already told clients, we're undergoing a digital transformation and we're gonna bring AI to you in terms of customer service, whereas the company's not even equipped yet to deal with the AI. the very simple principle is here, you will never use AI more advanced than the AI you're using within the company. it will always be less advanced because you need to be sure it's a high risk management and you need to be sure you can deal with it internally and that it's integrated in how your organization operates in pursuit of the goals and the purpose of that organization. So those are three levels, and as you can see, a very much parallels a change management process that leads to meaning, acceptance, and use of it.

Paul Sephton:

Do you think that we should be accepting right now? There's gonna be, let's say like a 20% dip in productivity as we learn and figure that out. Because if I'm trying to manage this in an organization, there's a degree of risk, as you say, that's involved. A lot of trust that's gotta be paved in that change management. But there's also potentially a lot of fear from employees around this idea or concept of replacement. You as a manager empower your teams to go. It's fine if you spend half a day trying to figure something out. Now we know that doesn't work. Or how do you go about accepting a short-term loss to be able to actually lean into a long-term gain and empower your teams to not feel that this is like implementation of a new tool.

David De Cremer:

That's a very good question because if we say it's change management, that means you're changing something. That means there's gonna be a dip in terms of how efficient your organization will work. So this already makes one thing clear. Your best practices, have to stay in place. It also makes clear, like I said, AI's not a strategy. AI is not like a hammer. Many companies mistakenly view AI as a hammer, where everything looks like a nail that you simply hit. not everything within your organization will need AI. You're experimenting with this. You're asking the right questions, you have to expect that it's a J curve. The change, it's gonna happen. You can't ignore it your best practices where you don't need AI, those need to stay in place and the investments need to stay there. So you can't go too fast in changing everything without knowing whether it works and it's integrated. That's why I mentioned while this is happening within your organization, you can never over promise to customers, oh, this is the AI we're gonna use or use exactly the same level of AI adoption for employees. As for, for customers, because you will have a problem. 'cause customers are not forgiving, they're not gonna say it's fine. You're trying to integrate AI, you're trying to adjust to it. If I have a little bit less service, that's fine. I'll give you that benefit of a doubt. No, of course not. So it's really gonna be trying to balance out the best practices and the experimentation with AI. That's gonna be extremely important. going back then, okay, who makes those decisions? Who leads this? Yeah. Those are business leaders. Again, those are not tech experts. So it becomes, again, important to have the right communication and to have the right expectations. Business leadership needs to be seen as supportive of trying this. So I see many companies where they say, okay, we want to be AI driven. And like you said, it may come across as it's fear based because it's a demand. People are okay, I have to, but there's no buy-in. If that's the case, you'll run the risk that people will not use it in the most efficient way possible because they don't understand the meaning very well. That's one issue. The second issue is I still see today that up to 70% of most employees are still doubtful to tell their supervisor they've used AI in the execution of their task. That's because norms and expectations are not clear. There's still the fear, despite the fact that most companies say, we want AI. I tell my boss, then he thinks I'm probably lazy or not that competent because I need AI. Which means leaders need to lead by example. Business leaders need to be part of the transformation as well. So when I set the collective learning process as a leader, as a team leader at every level, so leadership is not only the CEO or the board, it's really at every level leaders talk, the walk, and are part of it. That doesn't mean you need to be the AI expert, but a certain level of AI savviness, is needed and some openness because you have to model this.

Paul Sephton:

In terms of that with the AI savvy leader, do you have an idea of you see people who are talking the talk and walking the walk, I'm sure you can pick up quickly when someone's just throwing AI in as every third or fourth word to a sentence versus actually like modeling the behaviors that will truly drive successful implementations. So I guess the question is, what would an AI leader who others should be scared of look like? you've outlined some of those behaviors, but are there any , things that you can call out that you're seeing as consistent themes across

David De Cremer:

I've started paying more attention to language as well. And I found one really interesting effect is when business leaders say we're an AI driven company, most of the time their AI projects do fail to be integrated well because they see AI as a strategy. Because AI driven means AI drives everything. So it's a strategy in itself, which is completely wrong. But you have people who also talk about, we're an AI enabled company. I like that much more because those show a certain awareness of it's enabling us to do the business that we've always been doing, but in more efficient, optimized ways. Those, you can see, those are people who were asking the right questions, who are much more business savvy and tech savvy at the same time. One consistent trait I have observed is humility and open-mindedness. It's all about learning. AI savvy leaders engage in lifelong learning. This goes back to your earlier point when you said AI advances so quickly. This is where it's useful to stay updated. on what are the big trends that are changing. So you need to invest in yourself there. The second thing is people who are able to build climates. You've already mentioned the word of trust of psychological safety because. Failure needs to be seen as feedback But that comes along only if trust exists. Of course, because those are the sandboxes, the innovation itself is taking place there. That touches upon the soft skills that people talk so much about Because I have still many business leaders who ask you, but why do I focus on interpersonal skills as a leader? Shouldn't I learn to code? I said, no, that's already done. It's all bottom up. So it's learning from data and AI already codes now. So it's really, how do you use it again, you see it's still that being a, the admiration, the magical thinking of AI itself is gonna do everything. It's very difficult to break through that. And as long as you don't break through that, AI is seen as the strategy, the hammer. So and seeing that value in it, because if you have to work together, get feedback, collaboration is essential. So basically emotional intelligence, understanding where people suffer, where their difficulties are, and establishing that collaboration those those leaders who can bring that together, that people do communicate. That's important. So openness. Open-mindedness, humbleness because they allow you to learn building that trust and psychological climate, because that learning orientation is translated into sharing information and then collaborating with the right narrative that you have, which is the AI savviness, bringing tech and business experts together. And then at that moment when you can do this, when you can do this, you really delegate them to tech experts.

Paul Sephton:

It's really interesting the amount of data as well that's out there around as well that's out there decisions made in silos around AI or of when you look at a C-suite level of management, how how the AI decision maker or AI officer, whatever position is being implemented whatever position is being implemented, is usually the furthest degree of separation away from the rest of that C-suite. And it does just seem like a silo, but at the same time, I think I think people are probably thrown a lot by. Trying to separate the signals from the noise when it comes to how do I understand and orient or qualify success versus failure what's an irrelevant signal and what's the right signal to understand whether we're making progress on this? how do you start to look at those quantifications in a meaningful way

David De Cremer:

Remember that I said earlier, oh, oh, we're so obsessed and so focused on inputs. How much money throw do we throw at it? How accurate is the model? It seems like that's the KPI, of course, for tech, for the tech companies. Oh, Oh, this is a very accurate model, very predictive. Well, that's great, but then we buy it. It has to reveal something in terms of something in terms of results for a business, which is, do we sell more? Are we productive? Do we get more clients or do we keep our clients in a more competitive market? So those KPIs, in a way don't change, but people tend to forget that because like I said here, my mantra again, AI seen as a strategy as soon as we have AI. That's fine already, but that's of course not true because everyone uses AI and it's usually the same AI. So, um, So, the question of measuring success is really one that doesn't get enough attention. And that leads to a lot of noise, but noise also in saying we are successful. But what does that mean? Because most surveys are really based on perceptions. We ask leaders in companies, are you successful? Yes, I think so. Oh, well, 90% of companies say they're successful. basically. It really means nothing. So that's why in my book, I also say what I like today, when we talk about AI as augmenting AI, as that capitalization, AI is enabling your business, you need to look at it holistically. And holistically means it's much more than simply the return on investment in terms of finances, for example. It's really about. First of all are you augmenting the skills that are most useful in making your company succeed? So you need to know already as a good leader, what are the skills and the competencies required for your industry, for your business, and can you augment those? So that's, one measure already already for yourself to see whether the alignment is there, because because when the alignment is gonna be there, you're gonna produce results. and it's more holistically as well, because like I said, it may well well be if it's if it's demanded because we're an AI driven company and most of these companies will spend 70, 80, up to 90% of their tech budgets to simply buying the AI and then realize, oh, do we have the infrastructure? Are we equipped for it? Do we have the right talents on the right positions? Are our people Trained? And suddenly they think, oh, we have no budget left for that. It's holistic; you need to consider humans, AI, your organizational culture, and stakeholders, establishing measures for all these dimensions. Are your customers happy with this? How is this being seen in terms of your corporate social responsibility towards society? These things you also need to measure, especially if it's a long-term strategy, AI, needs to be a long-term strategy because remember, it's a J curve. It takes some time to adjust to it, to learn and then create value. So all those levels need to be measured. They need to be holistically to pre predict that long-term value. But of course, you wanna see your sales go up, you wanna see more customers. And like I said, this plays a role as well. Are you teams accepting it? They're adjusting and they're being integrated in the current jobs. Those are very important measures to see. It's so, it's not simply ticking the box. Are the teams using AI? It's also how are they using it? Is there an alignment with the skills and the focus and what the customers demand? And also, for example, one thing that they never think about, the communication. I’m going back to what I said earlier. Are you bringing it in the organization? How are you dealing with your customers at the same time? I'll give an example. There's a lot of consultancy companies today. They're helping you with your your AI adoption projects. Let's say the big five. Now, if you look at two and a half years, three years ago, how much do you think in terms of percentage of their revenue came from AI for those big five for those big five companies, what do you think the percentage of their revenue was?

Paul Sephton:

I would imagine it would be quite low sub 5%.

David De Cremer:

Well, yeah, very good. very good. Guess even less between zero and 2%. Today it's 20%. So this means that in just two and a half years, 20% of the revenue now comes from AI projects, either that means they've integrated AI in the service they already provided, and what I see happening is and what I see happening is they help you with your AI adoption projects in terms of making the teams ready, but they don't work outside of your company in terms of seeing whether customers accept it. So other measures that you need to see is how far has your AI adoption worked within the organization. Is there a balance with maintaining customer loyalty? And once it starts working for you, are you able then to translate it and uplift the customer service? None of these consultancy companies are thinking about that. None of them are asking about it. So, but those are things that you have to think yourself. So before you throw away good money.

Paul Sephton:

It's a tricky one, but it's certainly not something that a Band-Aid is going to be able to solve without the bigger picture thinking happening internally at

David De Cremer:

No. Exactly. And that's the thing. You hire companies who basically are going through the same phase as you are, trust me, their boards are also saying, can't miss out. It has to become part of our portfolio. And how do we do buildable hours with AI? That's their biggest concept.

Paul Sephton:

It seems like it falls into the bigger fear of missing out of am I an early adopter or am I a laggard? And what type of compounding intelligence advances are going to happen because if it's moving this quickly now, it will only compound further. And, who knows if we'll end up with a two-tier workforce.

David De Cremer:

Yes. Yeah. But and a consultancy companies are going through exactly what you're saying themselves. So they're serving clients who have the same problem as they have. So that's already the first problem, I see. But the second one is what you exactly said is, well before we lose out, and I wanna emphasize again, you need to learn to walk. Before you run, you need to learn to walk. So it's the same thing here. You need to understand what AI means to your business. That doesn't mean you need to change your business. You need to see how AI can be brought into your business. So there's, they're having that same problem. So I see these consultancy companies say, so I ask them, do you have your own AI innovation labs where you test how this may work for different industries? No, we don't invest in that because we're, we don't know yet how to use it ourselves, but they're still, but they're already selling their service to an organization. So it's the same thing here. Yes, of course AI will keep advancing, but for the next decade, the biggest question is really about the AI and business transformation, how to make it part of your business, because these consultancy companies are facing this challenge themselves. So obviously this is for you also the thing. So what is gonna be the newest AI is actually not the priority, but that's the second thought to say. So once because AI has a certain number of principles, and that's what I call AI seven, is that you understand what AI's capable of and how to use it in your business. Once you have to decide the type of AI, whether you're gonna focus on robotics, whether you're gonna focus on large language models, predictive models, just data analytics that comes once you understand the jobs that come along with pursuing your business.

Paul Sephton:

Agentic AI is something that you've lent into in the new book, and it seems to be the area on a heat map that has the most focus right now. What are your early signals of and wrong ways? Because with that with that everybody's talking about adaptive agentic models specific to a certain company or vertical or acting as a major accelerant within a business. Is that something that you think is us trying to run before we've learned to walk or necessarily crawl?

David De Cremer:

It's a yes and no in my view. As a company, what you wanna see is what AI can do for us. And from that point of view, it's augmentation mainly. So that doesn't mean augmentation, doesn't mean that there's no automation. And this is where we enter AI, ag, AI agents or agentic AI. So if you look at what AI agents mean today is really that they they can do what we've been talking about, the boring, repetitive, mundane tasks. They can do this autonomously. So So an AI agent is an autonomous agent. it's actually part of an augmentation structure. what you do on a daily basis, which is repetitive, if you can automate this in a way, the automation is augmenting you because you're saving time now. now. Whether it's really augmenting you is how you're gonna use that time. Are you gonna use that time to uplift yourself, upskill yourself, and focus on the tasks and elements in your job that are directly responsible for productivity? That's when we're gaining. Yeah. Do Then we're gonna have gains. Well, companies have a certain way of looking at, well, you save time. Let's do more of the same. And this is opposite of what AI gurus have always been saying. AI is gonna liberate us from the mundane, repetitive tasks, so everyone can be creative and the creativity is gonna provide the innovation, and that's gonna bring business forward. Well, Well, most companies actually don't. They don't They don't manage their company like that. They just don't use that time to uplift people. But here's a problem. How many creative jobs can you actually have? If you look today at how many jobs are really creative, 15% max. this addresses that issue of what are the jobs of the future? And this goes back to what I said earlier, don't wait for the list to arrive. Don’t wait for the list to arrive. You are working on this yourself, so you have to learn to integrate again. And experiment and do the collective learning. The second thing is AI agents, and this relates to, yes, you're right. It may be a little bit we're running before we're walking. It may be a little bit of a myth as well. Autonomous are artificial intelligence. Think about is Think about is this can only function when you can reduce risks, huh? In companies, it's risk management, so that means AI works perfectly and that principle has never changed over the years when all information is available. So that means it's a closed system. So when you see, when AI agents work best is in tasks within a system that's closed, meaning hardly anything can go wrong 'cause all the parameters are known, it is so repetitive that there's hardly any risk for things going wrong. Those are the things you put an AI agent on right now. When you have in what we call in what we call open systems, meaning it's very volatile, information changes all the time. Like within departments, between departments or between companies, there's a lot of noise on the communication there, and it becomes more risky an AI agent is not necessarily right. Across the different layers of hierarchy, but more effective within. The same layer of hierarchy. so for example, this was a Chinese hospital. Amazing. You have your doctors, you have your nurses, you have administrators. You're there, you need to get your medication. the AI agents are basically robots picking up the medication within the logistical space that has been identified. Everyone knows where, which medication is, and people nurses doctors only have to push the button and say, oh, it's this medication and everything's being delivered the whole time. the whole time. You don't have to run anymore. It's very easy. It's serving you, it's automation serving you to provide better service. To your customer, in this case, the patients. And it saves time for a doctor to pay more attention now now to the patient. so it is helping, but under certain conditions these conditions need to be very clear and you need to use that time saved to do something that helps and uplifts your company.

Paul Sephton:

Do we have a productivity crisis on our hands? The definition of productivity that is most adhered to is unsurprisingly from McKinsey. The idea, in a nutshell, is what's going to drive your bottom line. But do we have to if you took the European stance, it would be AI's gonna give everyone, instead of a two month holiday over the summer, a three month holiday over the summer, because we can get our work done more quickly the states. It might be something that's more aligned with, we'll get even more work done every single day. And we've touched on some of these subjects now, but do you think it's a case of by company figuring out what's our primary objective of succeeding with AI or that there's I think the doctor one is a great example because there, I think more human interaction as a result. Automated AI work Is the best outcome, or is it just that it's interaction instead of dealing with your inbox instead of dealing with

David De Cremer:

Now I like what you just said about US and Europe. For example, Europe may say, okay, we get three months of holiday instead of two months. Us may say no, no, no. People need to stay on the job and they just need to do more of it the same. So our productivity overall goes up. Now, I believe that both of them will ultimately fail in really creating the right value with AI for humanity in the long term. And the reason why I'm saying this is with Europe, that's a status quo thing because you're happy with what you're producing. And it's the old idea that was there in the sixties and seventies already, robots are gonna make sure we only work four hours a day. Which never materialized because if you look at the discussions about retirement and pensions, we have to work longer than ever. so what is happening here? With the US it's yeah, we'll just burn out people even more. To produce, but ultimately that will fall flat as well because we're underutilizing the technology and the potential of AI. Why am I saying that? That boat will fail and we under utilize the potential of AI. AI should be there to actually uplift humanity, which means we need to create these new jobs. And like you said, that's the responsibility of organizations. Obviously, companies would rather like to have people flow out of their companies and society pays for them rather than that they have to invest in creating new jobs. That's always a little bit of attention there as well, but now. You have to think about, let's say the current jobs that we all know AI will always get better and it may well be that we end up very soon at 95% of jobs. Technically could be done by AI. Now you see the problem. Are we helping humanity? No. If we don't change in terms of looking at what jobs are and what the new jobs should be, we're basically on a road to 0% human participation

Paul Sephton:

I mean I saw interestingly, I think it was last week, he posted on LinkedIn that D’amour McKim has just been recognized as having one of the leading MBAs with an AI track in it in it. Do you go about advising people who. Are maybe only gonna be in the workforce in what seems like a relatively short amount of time, three, four years from now. But where we can comfortably, estimate that AI will look radically different to what it does today skills, the competencies, the frameworks for thinking or traditional hard skills that you've gained during your time we think is gonna most set you up for year career

David De Cremer:

Yeah. And thank you for referring to our business school, but so the reason why we've been betting a lot on it is, the leaders of tomorrow, they need to be socially responsible leaders who can take different perspectives because judgment is something an AI doesn't have. So we want our leaders to have better moral compasses and it's, and at least have a reflection there. So seeing bigger pictures, asking why questions. So Second thing is really the resilience. You have to learn to adapt, which is the lifelong learning as well. . And this fits very much what, what we say, the economy of tomorrow is a feeling economy. the impact of hard skills will go down and the impact of soft skills will go up. This is the direction in which business schools are moving. So we will have a STEM designated MBA from next year on, thinking like an engineer is great, have designed thinking. Yeah. And be quant oriented. But I call this empathy. It needs to go together with your soft and that's the leadership. So the future of education, especially when for business leaders is empathy, stem, and empathy together, In these studies, the effects vanish immediately. As soon as you reveal the AI has written it, the effects are gone. So, because that means, oh, once they realize this, it's in our brain. It's not authentic. So an AI will, an AI can, I mean I can, for example, I can ask you a number of questions. If I were an AI, I can ask you a number of questions and see how you respond to it with your facial expressions, with your nonverbal behavior and everything. And then, okay, we'll do a game. And based on that information, I can already see what you're gonna do because I know the relationships between your facial expressions and what you'll do. And you'll say, wow, that's amazing. Yeah, that's true. But it's still based on imitation, based on observation, and it can imitate probably emotions. But is it really intent with the intent? No, there is no intent in AI yet. I don't know if this will ever happen because then we're talking about super intelligence. So we're not sure that's gonna happen because that's a completely different level, which AI is nowhere near. Just wanna make that clear as well. So, so there's two things, and AI, as for the next 10 years as I see it, there will not, there will nothing be authentic that what we perceive authentic AI will not be authentic because it will not have that in those intentions. That's why AI can make it decision, but it cannot make a choice. A choice is requires judgment, and that's what we still prefer the most. So that's in our brain. And I think it's also partly related to evolutionary tendencies because authenticity. How we are hardwired is related to morality, and morality is related to humanity. So in a way, implicitly, it doesn't connect with us. So we don't, we, we don't see an AI as being a part of that evolution. But of course, the more we talk about it also explicitly we don't want it. And this is a discussion that a lot of people forget when we say, oh, AI is all this potential and can do all these things. Look, at the end of the day, we still have a choice. We can still decide that we don't use AI for certain things, even if it's capable to do so. That's a normative approach. And sometimes I feel in the whole AI discussion, the fact that we as humans also have a choice and that we can make a moral choice when to use it or not, seems to be completely eliminated from the discussion because it's all about efficiency and productivity. And that's why I call also for a holistic approach. So a human-centered approach to AI, which is helping humans to become better versions of being a human and grow as a human and explore more possibilities. But it's holistically, if we narrow it down by simply focusing only on efficiency and productivity from a philosophical point, you can say, we are reducing humanity already now, and who's reducing this AI? No. The idea of AI and how we talk about it puts us in a certain mindset where we are actually reducing our own humanity.

Paul Sephton:

We're in one too many layers of, of a matrix at the moment with that in terms of if you got into the philosophical You think, we always talk about AI ethics from the perspective of the models rather than the people implementing them. But is there like a crash course you would give leaders in terms of this is how you can ethically implement AI within your teams or organizations?

David De Cremer:

That's needed. It's just simply needed. Look, of course tech companies are not gonna help you with this. At one point Google had ethics as a service, so I literally met them. People who said, ethics, that's now an AI issue. No AI can solve this themselves. We just need better data. I'm but better data, no data are a reflection of reality. Humans act like this. my argument there is always, always, we wrote a paper about this is, that's, that's great. AI amplifies our biases. But you know what? Use AI as an education tool for humans to learn about their biases. great. I don't see a problem there, to be honest. Because you can never de-bias data completely. It's impossible because it's based on the principle of what is the observation of reality there. It's scary how good they are in imitating humans, of course, because they know the reality as it is. But the thing is, humans can be Blind to their own biases, but that's because it's unconsciously. But when you make people aware of it, they can learn. So AI can be that learning tool in a sense as well. So again, this goes if we use it as a learning tool, it goes back to your earlier point. Oh. So then it's important that business leaders today basically learn to have moral awareness, recognize moral dilemmas in reality. Yes.

Paul Sephton:

It's a prompt that that I saw earlier this week which was, I think, I'm not gonna it, but it was to the effect of every single is created from a first person subject matter expert and then critiqued by a third person perspective. And I thought it was quite quite an interesting feedback loop of getting AI to basically come up with the most. Logically sound sound or factually knowledge based knowledge based sound insight, and then critique any shortcomings that it actually actually had. So So what you were getting back as an ultimate response was one that already had a measure of balance built into it.

David De Cremer:

That's where, what you're just describing is what tech companies since last year have started to say AI can reason. Now, yes and no because this is basically a reasoning process. But what hasn't changed is, yes they can work, for example, based on counterfactuals we worked a lot on introducing these counterfactuals that they say, oh, it's one thing. And then they provide reasons for the other and because they're rational, they can do so without emotions. So it's an amazing tool, get a balanced perception. However, two critical elements remain: we must still identify the problem as real-world participants and interpret the results, which does not change regardless of AI's advancement. If we become replaced and we don't upskill ourselves in what we are really good at, at the end of the day, that's a reductionistic because then AI will determine what is useful and interpret. This is useful for us, I don't think AI is there yet. They can't do it. And second of all, we shouldn't allow it either. If we say we care about humanity, look, if people say, if we say you know what humans are useless, that's fine. Let's eliminate ourselves. And as some people are saying, it's the next step in e evolution, human intelligence was not meant to survive. That's a different question, but I, and that's a different discussion, but I still like to believe that. So I call this humanization, human humanization, but then AI in the middle because AI is central. So the com the compromise is really towards the future. AI is core to how we as humanity can grow, but it serves our growth.

Paul Sephton:

And it needs to be Remaining that same vein, the more that we develop it moving forward.

David De Cremer:

Yeah. And that's that, and that is needed. Why? We need people to be able to recognize this, discuss it, to see the logic, and to make judgment calls.

Paul Sephton:

So there's, I'm conscious of time and not wanting to take up too much of yours, but there's three things that I think I'd love to end with What are your rapid-fire three tips for managers in 2025 for taking the next steps and getting things right? The three tips for knowledge workers, anyone on a team, anybody who's coming into an office every day and sitting behind a desk to make sure that they're leaning into this change. then ultimately how you would see 12 months from now the best case scenario, looking for where we are as a result of having pushed things forward.

David De Cremer:

This is for both knowledge workers and managers. I would say, first of all, AI is not an option anymore. You have to accept that go out, experiment, learn, That's the first step. The second one then, for managers, it's really, um, make AI adoption a collective participative learning process as much as possible. Because that will facilitate your communication, that will facilitate feedback and that will facilitate where you basically, your people management, where you need the talents and if you don't have them, Train them or hire them. So that's gonna be extremely important. Also, the third point I'm gonna emphasize is, again, again, in itself, AI is not rewriting leadership many of these behaviors of leadership Remain the same. It just needs to be put in the context of AI. So those are three things So those are three things for knowledge workers. Like I said, AI is not an option anymore. Train, be open to it. Do are aware, yes, it's a cognitive revolution. Yeah. But that means there's no point in being a human calculator anymore. what I mean by this is don't adopt your way of working to the way a computer works. Retrain yourself. Find ways of what it is that an AI can do and make yourself better in those things that that's where I set the feeling economy. So your soft skills need to be there, but it's really, like I said, the critical thinking, the perspective taking the judgment itself and what you can do with that knowledge. I think that's gonna be extremely important. And leverage AI in making your work more efficiently. See where it fits and where it doesn't fit with what you do. And where are we gonna be in 12 months from now? Well, one thing that's still gonna stay there there is, I'm pretty sure next year, and you will see this, we're still gonna be talking about, oh, AI as a business transformation. How are we gonna do it? How do we create most value? Are we gonna see better models? Yes. But we are also gonna see that that AI is gonna be driven into the territory of other discussions like sustainability, the environment, energy, which is already happening. What I hope is that we're gonna have a little bit more of of a serious discussion about ethics and regulation, because I don't need to explain why, but in the US an era of deregulation has arrived and that's, it's gonna be interesting at the global level because you have different malls. Now US is deregulation. Europe has a lot of regulation and very specifically China is very principle base, has a certain rules. And this basically state based. So what I think the big discussion definitely next year is gonna be regulation and human centered approaches of AI. So again, the ethics is gonna be more important than ever.

Paul Sephton:

David, I think we could go into that. I'd love to do a separate episode on it if we had the time. But yeah, before I ask another question, I will stop myself short and just say thank you so much for the time today.

David De Cremer:

No worries. I very much enjoyed it, Paul.