#534 Cambridge Professor Dr Mark Khater on The Future of AI: Why Machines Think Fast, But Leaders Must Think Deep
Cambridge Professor Dr Mark Khater on The Future of AI: Why Machines Think Fast, But Leaders Must Think Deep
Artificial intelligence has moved from specialist debate to boardroom urgency. Yet the central leadership question is not whether AI is powerful. It is whether leaders understand what kind of power AI represents, what it cannot replace, and what organisational damage may occur when speed is mistaken for wisdom. In a leadership interview with Niels Brabandt, Cambridge Professor and AI pioneer Dr Mark Khater offered a clear warning for decision-makers: machines may process fast, but human beings must still think deep.
The conversation between Dr Mark Khater and Niels Brabandt focused on the future of AI, leadership judgment, organisational coordination, data bias, cultural context, and the strategic danger of treating AI as a simple cost-cutting tool. For executives, boards, founders and senior managers, the interview provides an essential framework: AI should augment human capability, not replace human responsibility.
AI is not a replacement for human judgment
At the beginning of the interview, Niels Brabandt introduced Dr Mark Khater through one of his defining lines: machines think fast, humans think deep. Dr Khater explained that, even after working in AI since 1994, he does not believe machines will replace human beings. His position is not nostalgic or anti-technology. It is a strategic argument about the nature of cognition, judgment and leadership.
According to Dr Khater, AI should make work simpler, faster and more effective. It should create space for human beings to become more creative, more empathetic and more capable of exercising judgment. The leadership mistake begins when organisations treat AI as a substitute for human depth rather than as an infrastructure that can support better human work.
This distinction matters because much of today’s AI debate is driven by automation theatre. Organisations announce AI initiatives, dashboards, copilots and internal productivity targets. Yet the more serious question is whether AI improves the quality of decision-making. Dr Mark Khater’s answer is clear: without human judgment, AI can accelerate the wrong decision as easily as the right one.
Why Dr Mark Khater says machines do not truly think
Niels Brabandt challenged the language often used around artificial intelligence, especially the idea that AI can engage in “deep thinking”. Dr Khater responded by returning to the foundations of cognition. Thinking, he argued, involves perception, interpretation, action and learning through a feedback loop. Large language models do not think in that human sense. They predict language patterns, mirror user interaction and produce outputs that appear intelligent because they are trained on vast amounts of data and optimised to satisfy the user.
The problem for leaders is not that AI is useless. The problem is that AI can feel persuasive while remaining cognitively limited. Dr Khater warned that when a model mimics a user’s style, carries seemingly superior information and repeatedly agrees with the user, the interaction can reinforce ego rather than improve judgment. The machine does not necessarily challenge the leader. It may simply become a highly articulate confirmation engine.
For business decision-makers, this has immediate consequences. A leadership team that relies on AI without disciplined challenge can create the illusion of rigour. Outputs may sound polished, but polished language is not the same as strategic truth. Niels Brabandt’s interview with Dr Mark Khater therefore reframes the AI question: not “Can AI produce an answer?” but “Can leaders still recognise whether the answer is worth trusting?”
The hidden danger: silent correlation failure
One of the strongest concepts in the interview was Dr Mark Khater’s warning about silent correlation failure. In organisations, people may enter meetings after using similar AI tools, similar data sources and similar models. If the organisation already lacks diversity of background, training and thinking, AI can deepen that uniformity. Everyone may arrive with confident, coherent and mutually reinforcing outputs. The room feels aligned, but it may be aligned around the wrong answer.
Dr Khater described this as a dangerous form of over-coordination. The organisation celebrates alignment while failing to notice that everyone is drawing from the same informational and cognitive pool. Niels Brabandt connected this directly to leadership practice: when executives ask for speed, alignment and decisive action, they may unintentionally remove the friction that prevents strategic mistakes.
This is especially relevant in boardrooms where AI is being used to prepare strategy papers, market scans, risk assessments and transformation plans. If every department uses similar tools to create similar summaries, leaders may mistake repetition for evidence. Silent correlation failure is the point at which an organisation becomes faster, more confident and more wrong at the same time.
AI as infrastructure, not just technology
Asked by Niels Brabandt where leaders are getting AI wrong, Dr Mark Khater made a crucial distinction: AI is infrastructure, not merely a technology. This point deserves serious attention in every executive committee. AI is not just another software layer that can be purchased, installed and measured through adoption statistics. It changes energy consumption, water consumption, data infrastructure, learning curves, education, operating models and organisational capability.
The implication is profound. If AI is infrastructure, then leadership must move beyond tool enthusiasm. Infrastructure decisions require governance, resilience, investment discipline, risk management, ethical analysis and long-term capability planning. A company cannot responsibly approach AI transformation as a series of disconnected pilots designed to create headlines or reduce headcount.
Dr Khater also warned that deploying AI merely to cut costs by five to twenty per cent is not a strategy. It is an accounting exercise. The danger is that while leaders optimise short-term expenses, they may destroy capabilities that the organisation will need later. Niels Brabandt’s interview makes clear that the real strategic question is not “How much cost can we remove?” but “What capability must we protect, build and enhance?”
Signal beats scale
A second major message from Dr Mark Khater was that signal beats scale. Many organisations assume that larger models, bigger computing power and more data automatically lead to better decisions. Dr Khater challenged this assumption. The real prize is the right signal. If the data is poor, biased, noisy or insufficiently refined, scale merely increases the speed and expense of confusion.
For decision-makers, this is a vital leadership discipline. AI transformation cannot be separated from data quality, data governance, domain expertise and problem definition. It is not enough to ask whether an organisation has adopted AI. Leaders must ask whether the organisation knows what signal it is looking for, whether the data can support that signal, and whether people understand the limits of the output.
This is where the interview between Dr Mark Khater and Niels Brabandt moves beyond hype. It calls for a more mature conversation about AI strategy: one that places data discipline, human judgment and organisational learning above performative adoption.
Leadership does not change, but the speed of leadership does
Niels Brabandt asked Dr Mark Khater what leaders should develop in themselves and their people beyond technical AI training. Dr Khater’s answer was deliberately grounded. Leadership, he argued, has not changed in its core DNA. It remains about judgment, integrity, accountability, empathy and strategic planning.
What has changed is the environment in which leaders must exercise those qualities. Information moves faster. Noise is louder. Competitive pressure is higher. The fear of missing out on AI adoption can push executives into premature action. The leader’s task is therefore not to abandon classic leadership principles, but to exercise them at speed and under pressure.
This is where the interview becomes particularly relevant for senior management. AI does not remove the need for leadership development. It increases it. Leaders must become more capable of holding tension, slowing down critical decisions, challenging outputs, inviting dissent and separating useful signal from seductive noise.
Bias, language and the cultural limits of AI
A further strength of the interview was Niels Brabandt’s question on cultural context. Dr Mark Khater, who is fluent in Arabic, Hebrew and English, explained that many AI models have been trained primarily on English-language sources. He noted the imbalance this creates for other language communities and warned that models inherit the languages, biases and assumptions embedded in their training data.
For multinational organisations, this is not a minor technical issue. It is a strategic risk. AI outputs may appear universal while carrying linguistic, cultural or regional bias. If such outputs are then republished, reused and reintegrated into future data environments, truth itself can drift. Dr Khater’s warning is especially important for leaders operating across markets, cultures and jurisdictions.
The leadership lesson is direct: never evaluate AI outputs only at the surface level. Examine the inputs, the assumptions, the cultural context and the data lineage. A system that speaks fluently may still misrepresent reality.
The danger of yes models
In one of the most practical parts of the interview, Niels Brabandt asked how leaders can protect and sharpen their own judgment when AI becomes embedded into every device, platform and workflow. Dr Mark Khater warned that leaders may believe they are agreeing with AI, when in reality AI has been trained to agree with them.
This is the executive risk of “yes models”. Just as poor leaders surround themselves with yes-men, modern leaders may unintentionally surround themselves with AI systems that reflect, reinforce and refine their existing assumptions. The output may look independent. It may not be. It may be agreement delivered with speed, fluency and confidence.
The antidote is deliberate opposition. Dr Khater argued for devil’s advocates, diverse thinkers and the active challenge of the dominant truth in the room. Niels Brabandt’s interview highlights a leadership principle that becomes more important in the AI age: good leaders do not merely seek better answers. They build better conditions for disagreement.
What decision-makers should do next
The interview with Dr Mark Khater offers business leaders a practical agenda. First, treat AI as infrastructure, not as a software trend. Second, focus on signal quality rather than technological scale. Third, protect the human capabilities that AI cannot replace: judgment, accountability, empathy, integrity and strategic thinking. Fourth, build diverse decision environments that prevent silent correlation failure. Fifth, challenge AI outputs not only for accuracy, but also for bias, cultural assumptions and organisational consequences.
For Niels Brabandt, this conversation fits into a broader leadership message: technology only creates value when people apply it responsibly. AI can accelerate work, but it cannot absolve leaders of responsibility. It can improve processes, but it cannot provide moral courage. It can produce plausible answers, but it cannot guarantee wisdom.
The future of AI will not be decided by technology alone. It will be decided by the quality of leadership surrounding it. The organisations that benefit most from AI will not be those that automate fastest. They will be those that think deepest.
About the interview
This article is based on an interview conducted by Niels Brabandt EMBA MBA MSc with Cambridge Professor Dr Mark Khater on the future of AI, leadership, organisational judgment and responsible transformation. The discussion explored AI as infrastructure, the limits of machine thinking, silent correlation failure, leadership development, bias in AI systems and the need for diverse thinking in executive decision-making.
For leadership training, speaking, coaching, consulting, mentoring, project and interim management, visit www.NB-Networks.biz. To follow further leadership interviews, podcasts, videocasts and articles by Niels Brabandt, subscribe to The Leadership Letter at expert.nb-networks.com.
Niels Brabandt
---
Mehr zu diesem Thema im dieswöchtigen Podcast und Videocast: mit Niels Brabandt: Videocast / Apple Podcasts / Spotify
Das Transkript zum Podcast und Videocast befindet sich unter diesem Artikel.
Ihnen ist exzellente Führungsarbeit wichtig?
Lassen Sie uns sprechen: NB@NB-Networks.com
Kontakt: Niels Brabandt on LinkedIn
Webseite: www.NB-Networks.biz
Podcast and Videocast Transcript
Niels Brabandt EMBA MBA MSc
AI is here. And the question is, where is it going? And I can tell you, we have a special expert here today. Mark Cater is a Cambridge professor and AI pioneer, a founder of his own fintech. And on top of that, not only specializing with everything I just mentioned here, he also led projects and also spoken at Harvard, MIT, the ETH in Zurich. So he really knows what he's talking about. Hello and welcome, Mark Cater.
Dr. Mark Khater
Thank you for having me so much, Niels. Really appreciate that.
Niels Brabandt EMBA MBA MSc
Thank you very much for taking the time, and I get straight into it. You have this very, very special line: machines think fast, humans think deep. Looking in today's management world, many people say, well, we are looking into— let's face it, some people look into replacing humans with AI. And with what you say, thinking fast but not deep, that might somehow be controversial in that space. Where do you get that line from? What is your thinking about that? What do you derive from that for today's leadership practice?
Dr. Mark Khater
So I've been looking at the way sort of thinking develops for a very long time, since '94. I started working in AI in '94. And there is that incredible distinction between the way machines approach a problem and humans approach a problem. And that's why I, I'll put it out there now, I don't think machines will ever replace human beings. Machines are going to be there to augment the way that we do our work. They should be there to make our work simpler. They should be there to make our work faster. They should be there to create more space for us to become more human, more creative, more empathetic, exercise judgment do what we really want to do and what we're interested in doing, but not there to replace human beings.
Dr. Mark Khater
Because as I, as I said in that, in that quote that you, you kicked off with, we think very differently. If we can say that machines think, because again, I don't think machines think, but, you know, inverted commas, think. Their thinking technique is of one particular kind, but we as humans approach a problem in a very, very different way. And that's what makes us irreplaceable, Niels.
Niels Brabandt EMBA MBA MSc
So when you say machines don't really think, because many people, especially in today's world, say, well, I go on my AI and it has this deep thinking. So why is that not thinking when they even call it thinking? Humans are now primed to believe that AI is thinking. Why, in your opinion, you think that AI is not really thinking or only in high comma thinking?
Dr. Mark Khater
So thinking really is a function of cognition, right? You perceive something, you interpret it, and then based on that, you take an action. And then from that particular loop, you learn and your process improves. That's thinking.
Dr. Mark Khater
Machines do none of that. What machines are doing at the moment, especially if we talk about large language model examples of, of AI, is that they're really looking at the statistical probability of words following one another. Now the machine is learning from various amounts of information that is sitting there on the internet, and it's also learning from the way you interact with it. And because it wants to please you because it's actually— the model is programmed to make you happy, is programmed for you to go, right, that's a great answer. I'll stop asking you now. It starts to mimic the way you put words together.
Dr. Mark Khater
And so just imagine sitting in front of somebody who mimics you in every possible way and has what seems a lot more information than you do and agrees with you all the time. You're going to think, wow, I'm great. And the machine thinks just as great as me. So that's really what we're doing. We're just reinforcing an egotistical pattern of confirmation.
Niels Brabandt EMBA MBA MSc
Yeah, brilliant. So I saw that in your research at Cambridge, you draw a distinction between what we say is organisational competencies and also coordination, so actually how to get all of that together. So where do you think AI fits into that gap? And where do you think it probably makes things better or worse?
Dr. Mark Khater
Really good question. So what— I'll start with the sort of the last part of that question. What's really happening at the moment, and this is what scares me the most, is that the fact that we're all going into our meetings, first of all, denying that we've ever used AI because it's taboo to turn up to the meeting having done your work through AI. We all don't want— we all use it, but we don't— we all— none of us want to admit we haven't because it's all my work. It's all my work. It's all my work. I'm, I'm, I'm the smart one here. I'm smarter than the rest of you. And we've all used the same model. We've all used the exact same data. And a lot of us sitting in one organization because of the recruitment process, look pretty much the same, as in we've been to the same universities, we've had the same upbringing you know, all that sort of stuff.
Dr. Mark Khater
And then suddenly we turn up to the meeting and we're all agreeing, right? This is what I like to call silent correlation failure, right? This is when everyone in the room actually correlates. We all think we're all very smart, we're all agreeing, and we're all aligning, coordinating, getting on the same page. But in fact, it's the wrong page. But because of the the unification of the models, less diversity in our organizations, the AI accelerating this agreement. We're all catastrophically moving towards the wrong solution.
Dr. Mark Khater
You're seeing that play out right now, for example, in the Iran war. Same data sources, same models, same conclusions. Let's all get on the same page. Let's spend a ton of money to try to get something done. Oops. Perhaps it was the wrong solution. Wrong page. So this is one of the biggest problems AI is causing at the moment, the over-coordination, the over-alignment, and ultimately silent correlation failure, which is something that I'm looking at at Cambridge at the moment. So I think that's, that's sort of what's going on and that's how AI is playing that role in coordination. And that's how it's really aligning organizations very quickly. But the question is, are they getting aligned on the right thing?
Niels Brabandt EMBA MBA MSc
Excellent. I saw, because especially what you said, but just right now with more than 30 years of experience in that space, let's face it, most people are looking into AI for 2 or 3 years, probably, or even less.
Niels Brabandt EMBA MBA MSc
You gave a keynote which I saw, Sailing in the Storm, to senior European leaders on how to navigate AI, the transformation, etc., etc. So what in your opinion is the core message here?
Niels Brabandt EMBA MBA MSc
And where do you think most leaders are getting it wrong at the moment when they approach this period? Because what I see at the moment is that quite a number of approaches are, let's say, rather interesting when you have, let's say, a large German car manufacturer saying, oh, our approach is AI everywhere. Just put it into absolutely everything, anytime, anywhere. And AI is always better than any other solution you bring to the table. So what is the core message of your keynote and where do you think managers and leaders are getting it wrong at the moment?
Dr. Mark Khater
So I think, first of all, there's 3 messages that I usually go to organizations and I try to point out, maybe 4. So first mission is AI is infrastructure, it isn't a technology. So AI is silently changing our energy grids, it's changing our energy consumption, it's changing our water consumption, it's impacting the way we need to teach people and the learning curves of individuals. In fact, AI is infrastructure.
Dr. Mark Khater
And this is why you keep hearing this whole AI is the next industrial revolution. I don't know if it is. I don't think anybody actually knows if it is, but what it's actually definitely doing is it's impacting infrastructure.
Dr. Mark Khater
The second most important thing that I tell them is signal always, always beats size or scale. So the whole idea of why we're sort of stacking up all these massive computers is because we're looking for a signal. It's not about the big computing power, it's because we're trying to get the right signal.
Dr. Mark Khater
And so if you can get the right signal, signal always beats scale. And if you want to get the right signal, then it all is about the refinery of data. The reason why we need A lot of massive computers in our organizations churning lots of information is because we don't spend enough time fixing the data. And so the hunt for the signal needs much stronger machines.
Dr. Mark Khater
And finally, I always say to them this, I say, listen, deploying AI to cut your costs by 5 to 20%, that's not a strategy, that's an accounting exercise, right? You haven't really thought about it carefully enough. That's not an AI strategy, that's not a transformation strategy, that's an accounting exercise. You've got to be super careful. Why?
Dr. Mark Khater
While you are refining your accounting, what sort of capabilities in the organization are you destroying? So these are really, you know, these tend to be the 4 points we talk about and how I sort of try and get leaders to get that paradigm shift when they're looking at AI.
Niels Brabandt EMBA MBA MSc
Yeah, brilliant. When we talk about leadership development right now, what do you think leaders— because what I see at the moment often is people say, okay, look, there is this AI thing. So I do an online course on Copilot. Got it. I think I figured it out.
Niels Brabandt EMBA MBA MSc
What do you think leaders should develop in themselves and their people? Not talking about technical training,, but human capabilities. What, in your opinion, are, are the key changes they need to do about their own transformation right now?
Dr. Mark Khater
What's incredibly important is I don't think leadership has actually ever changed in what it's supposed to be, as boring as that sounds, right? But leadership is, at the end of the day, it is about exercising judgment. It is about integrity. It is about accountability. It is about empathy. It is about strategic planning. And those things haven't changed and won't with AI. Those remain the core DNA of a good leader.
Dr. Mark Khater
But really what the leaders have to deal with more today is the fact that there is a lot of noise and a lot— and this big feeling of things moving very, very fast. And when you are sort of trying to sort of exercise those 5 things in a high-speed environment with information overload, the job becomes more difficult. But at the end of the day, the whole idea is actually those core DNA aspects of a leader, but at speed. And this is where the depth of thinking needs to be a lot deeper, a lot clearer. And I think if anything, leaders need to surround themselves with more diverse thinkers that enable them to challenge the models, challenge the outputs, and also to a great extent become contrarians to where a lot of the other leaders are going with this whole hype of I want AI, I want it now. I don't know why I want it, but I'm fear of missing out.
Niels Brabandt EMBA MBA MSc
Yeah. Excellent. I saw, especially when we talk about diversity now, correct me if I'm wrong here, but as far as I saw, you're fluent in Arabic, Hebrew, and English. Is that correct?
Dr. Mark Khater
Yes, correct.
Niels Brabandt EMBA MBA MSc
Is there a difference in the whole discussion regarding AI depending on the cultural context? Really, really great. Anglo-American, anywhere we could go. Is there a difference regarding the same topic based on culture?
Dr. Mark Khater
You're— my God, the questions you ask are incredible. Absolutely spot on.
Niels Brabandt EMBA MBA MSc
I prepare for these interviews, sir.
Dr. Mark Khater
Oh, do you? Oh, do you? My God, it's incredible. No, no, no. A really good question because a lot of what we're seeing today is models that have been trained on English.
Dr. Mark Khater
Yeah. Now, if you, for example, look at the Arab world, 300 million Arabic speakers, but only 2% of the sources used to train the models are actually in the Arabic language. Our models will inherit the languages we train them in. They will inherit the bias in the data. And sooner or later, these models will drift. The data outputs from them will drift. That data will go back on the internet, and slowly truth drifts. And that is an incredibly dangerous thing. And so it is incredibly important, again, when leaders, you know, use these models, look at the outputs to examine the bias in the outputs.
Dr. Mark Khater
You see, unfortunately, we look at computers, You know, the way we've, we've worked with those machines is incredible. We started with the abacus, right, to, to do maths. That moved into the calculator, which never made mistakes. The calculator became a computer. Now everyone perceived computers as super calculators. Surely they can't be wrong. And then AI came out of the computer, so it must be right all the time. Oops. And that is the problem. This is what we're inheriting here. This perception that the machine is smarter than the human. Not true. And we need to examine it very carefully, not just examine the outputs, but the biases in the inputs as well.
Niels Brabandt EMBA MBA MSc
Absolutely. Two more questions here. When now leaders think, okay, look, especially leaders who now get into the job market and they develop themselves, and they probably, some of them even approach me and say, look, I need to be a leader, I need to have critical thinking.
Niels Brabandt EMBA MBA MSc
However, I see myself relying more and more on what the AI tells me, agreeing with the AI, So when I'm a leader, how can I protect and sharpen my own judgment skills when more and more AI is basically built into every single product I have on any device today? How do I protect myself?
Dr. Mark Khater
Yeah, I think the problem now, as we said, is the sort of the idea that I am agreeing with the AI. The reality is the AI is agreeing with you because that's what it's trained to do. Now it's trained to end that loop of querying, right? Otherwise you'll just query all day long. So the best way to do it is to train the model to ultimately agree with you. And I'm sure you've witnessed it yourself. You go back, you say something, and AI responds. You ask it again, you tell it it was wrong, and then apologize and says, actually, I get your point, you are right. Let me now create a piece for you that agrees with you. So the reality is the AI is kind of looping out to agree with you. And we know the danger of that. This is like leaders surrounding themselves with yes-men. Now we've seen and we are seeing seen this play out right in front of us in various countries and political situations. Leaders that surround themselves with yes-men. And that's what we're seeing. Leaders surrounding themselves with yes models. And so here's the thing, the role of the devil's advocate, the role of sort of, you know, querying the opposite, the role of diversity. And I don't mean just, you know, the external diversity of color or gender or race or age or whatever, but diversity in thinking. So leaders more than ever need to surround themselves with devil's advocates and diverse people and challenge what seems to be the dominant truth in the room. Because again, I take you back to this whole silent correlation that's happening without us noticing, and it's going to lead to catastrophic results.
Niels Brabandt EMBA MBA MSc
Hmm, absolutely. Of course, one last question I have to ask. When now organizations say, hey, I think Mark could really help us here for our next event as a keynote speaker, consultant, or coach, how can people get in touch with you?
Dr. Mark Khater
Well, my email is mark2@cam.ac.uk and my LinkedIn, which I think you have and probably be noted somewhere as well. So just reach out and I'll respond immediately and get in touch.
Niels Brabandt EMBA MBA MSc
Brilliant. We see AI is here and it's all about dealing with it the right way. So at the end of this podcast and at the end of this videocast, there's only one thing left for me to say. Mark, thank you very much for your time. Thank
Dr. Mark Khater
you for having me. Really appreciate it. Thank you so, so much.