EM: Before we start talking about the machines, let’s discuss the method. The UAE built a reputation for governing to measurable targets, and you have watched it from the inside of the talent market that staffs it. You have operated in the UAE since 2007. What specifically changed in how the government recruits, holds accountable, and replaces its own leaders over that period, and when did you first notice it?
I arrived in Dubai in 2007, and the first searches I ran here were won or lost on relationships and pedigree. The question in the room was who someone knew and where they had served. Around 2010, as the first government strategies arrived with hard targets attached, that changed. Clients began asking me for people who had turned something around, and they wanted the evidence. I remember a chairman stopping me mid presentation to say, tell me what this person actually delivered. The deeper shift came later, when entities became willing to move a respected leader who was not delivering. That is rare here, and the first time I saw it done calmly, without drama, I understood the system had matured.
EM: People praise the UAE for governing to measurable targets. From where you sit, placing senior leaders, what does that culture actually demand of a person that a more traditional bureaucracy does not?
A traditional bureaucracy rewards a leader for following the process and surviving the year. The system I work inside here rewards a leader for delivering something a citizen can see. That demands a particular kind of person. They have to be comfortable having their name attached to a number, and comfortable being measured against it in public. They have to move at a pace most public careers never require and deliver across departments that do not report to them, through influence. But, the hardest thing to assess is appetite. Many capable people want the title and the proximity to power. Far fewer want to be held to account for an outcome. When I assess for these roles, that appetite for accountability is the first thing I test for, because everything else can be coached and that cannot.
EM: Where do you think the "deliver, not just administer" model has clear limits, or a cost that its admirers tend to leave out?
Every model has a bill, and this one has two. The first is time horizon. When a leader is measured on what they can show this cycle, the work that only pays back in ten years can quietly starve. Some of the most important things a government does, building institutional memory, developing people slowly, earning trust in a community, do not fit on a dashboard. I have said for years that resilience is built in the quiet years when nothing seems to be happening. A pure delivery culture is impatient with quiet years. The second cost is human. A system that runs at this pace burns people, and the ones it burns are often the best, because they care the most. Admirers of the model rarely mention either. I admire it too, and I still watch for both.
EM: Other governments send delegations to study this model. Give a concrete example of something that you think does not transfer across a border, and why.
The dashboards transfer. Delegations photograph the smart government centres and the performance systems and take the playbook home. What does not transfer is the thing underneath it, which is continuity of command. The reason a decision here can travel from idea to execution in months is that the leadership is aligned and stays in place long enough to see it through. A government built on a four or five year electoral cycle, with a coalition that has to be renegotiated constantly, cannot replicate that even with the same software. I have watched ministers from other regions study the model with real admiration and then return to a system that will not let them act on what they learned. The tools travel. The political continuity that powers them stays at home, and that is usually the part that matters most.
EM: What is the difference, in practice, between a government that performs and a government that merely looks like it performs, and how would an outsider tell them apart?
The simplest test is what happens after something goes wrong. A government that performs will publish a number that embarrasses it and then move a respected leader because the result was not there. A government that performs for the cameras announces, launches, photographs, and then quietly lets the same people carry on regardless of the outcome. An outsider cannot read the strategy documents, but they can see two things. They can see whether a service works when they use it themselves, at the counter or on the app, with no one watching. And they can see whether anyone is ever held to account when it fails. Announcements are cheap and easy to stage. Consequences are expensive and hard to fake. Follow the consequences and you will know very quickly which kind of government you are looking at.
EM: In 2025 the UAE announced an AI system would sit as an advisory member of its Cabinet, and a separate drive to move half of government to autonomous AI. This is the hinge of the conversation. An AI advisor now has a seat, non-voting, at the Cabinet table. As someone who advises boards, what changes about a room the moment a machine is one of the voices in it?
I have spent a career in rooms where the real decision happens in the pauses, in who speaks last, in what is left unsaid. A machine changes the physics of that room. It is always prepared, never tired, never managing its own career, and it speaks with a fluency humans instinctively trust. The first thing it changes is tempo. The analysis arrives faster than the group can digest it, and the temptation is to move at the machine's speed rather than the institution's. The second is deference. There is a quiet gravitational pull toward the most confident voice in the room, and the machine is engineered to be confident. That makes the chair's job harder and more important. The chair has to protect the space for human disagreement and let the machine inform the judgment in the room without quietly replacing it.
EM: You assess humans for a living. How do you assess the judgment of a board or a cabinet that is leaning on a machine's analysis, when the machine is persuasive and fast and the humans are tired?
I assess it the way I assess any leadership team, by watching how they handle a confident answer. A board with judgment receives the machine's analysis and immediately asks what it cannot see. What is missing from the data. Whose experience is not in the model. What happens to the person at the edge of the policy who is not a number in the system. A board without judgment receives the same analysis and feels relief, because the hard thinking appears to have been done for them. The fatigue you mention is the real danger. Tired people hand judgment to whatever is fluent and available, and the machine is always both. So the question I listen for is a simple one. Someone has to ask whether we should, after the machine has told us we can. If no one asks it, the machine has quietly stopped advising the cabinet and started running it.
EM: Where, precisely, should the line sit between what the AI partner decides and what a human must own? Give an example of a decision you would never hand across that line, and one you happily would.
I draw the line at dignity. Anywhere a decision touches a person's dignity, their liberty, their livelihood, or the trust between a citizen and the state, a human has to own it, sign it, and answer for it. I would happily hand a machine the allocation of ambulances across a city at three in the morning, the detection of fraud in a benefits system, the balancing of a power grid. These are problems of optimisation inside values that humans have already set, and the machine will do them better than we can. I would never hand across the decision to take a child into care, to deny a person their freedom, or to choose who is helped when not everyone can be. Those are judgments about what kind of society we are, and they need a human being who can be looked in the eye and held responsible. A machine cannot carry responsibility, and responsibility is the whole point.
EM: The promise is that this brings government closer to its people. The risk is that it puts a layer of machine between them. Which do you think is more likely, and what determines it?
Both are possible, and which one happens is a choice leaders make. Take a benefits system. Used well, the machine clears the thousands of straightforward claims in a morning, and the human officer spends the afternoon with the family whose situation is complicated and painful, the one a form was never going to solve. That brings a government closer to its people. Used badly, that same family is the case the machine cannot categorise, so it loops them endlessly and no human ever appears. What decides the outcome is whether there is always a human door behind the machine, and whether leaders measure the right thing. Measure only cost and speed and you will build a wall. Measure whether a citizen left feeling understood and you will build a bridge. The machine will faithfully deliver whichever one you actually ask it for.
EM: If you were briefing a UAE leader on the single human capability they must strengthen as they delegate execution to AI, what is it, and why that one?
I would tell them to strengthen their judgment, and I would be specific about which kind. As the machine takes over execution, the scarce thing at the top becomes the confidence to sit with a fast, fluent, persuasive answer and still ask whether it is the right thing to do. That requires two things, both built deliberately. Enough depth in your own field that you can sense when a confident answer is subtly wrong. And enough self-trust to overrule the machine in front of a room that can all see the machine agreed with itself. I choose this one because everything else is being automated. The forms, the analysis, the first draft, the optimisation, all of it is moving to the machine. What stays with the human leader is the final judgment, and a leader who has not deliberately built that capacity will hand it over without realising they have done it.
EM: Contrarian prompt: make the strongest case you can that an AI advisor in the Cabinet is a mistake, even if you do not believe it.
I can make that case, and I can make it strongly. Start with responsibility. The moment a machine sits at the table, every hard decision acquires a place to hide. The model recommended it becomes the most comfortable sentence in government, and accountability quietly dissolves into the software. Then values. Every model carries assumptions inside it, often built somewhere else, by people a cabinet will never meet, reflecting a world view no one in the room voted for. You are seating a foreign philosophy at your own table and calling it analysis. Then pace. Wise government depends on a certain slowness, time for objection, for second thought, for the wisdom of delay. A machine compresses that time and calls the compression progress. And finally the cost of disuse. A cabinet that leans on a machine for ten years will lose the ability to reason hard things through on its own, the way a man who stops walking loses his legs. That is the case, and it is serious. The response is to keep a human spine in the room strong enough to carry the weight the machine cannot, and to make sure responsibility always has a name attached to it.
EM: The Human Skills Economy: Let's Pull back from government to the whole labour market. The claim is that as machines take routine work, the human-only skills, empathy, judgment, creativity, trust, rise in value faster than supply can meet. You see the demand side directly, in what clients will pay a premium to hire. What human skill has actually repriced upward in the last few years, and is that visible in the offers, not just the rhetoric?
What I see repriced, in the actual offers and not in the brochures, is the ability to build trust quickly across boundaries. Five years ago a brilliant technical leader who struggled with people could still command a top package on expertise alone. Today that same person clears a lower bar, because the expertise is increasingly available from the machine, while the trust still has to come from a person. The premium has moved to the leader who can walk into a room of suspicious, competing stakeholders and have them working together by the end of the week. I placed a transformation leader last year whose technical depth was good without being exceptional, and the client paid well above the band for one reason. Every previous attempt had failed because the organisation would not follow the leader. This person made people want to be led. That capacity is now the most expensive thing on the market, and the hardest to fake.
EM: "Empathy, judgment, creativity, trust" can be a slogan. When you assess a candidate, how do you tell real judgment from the performance of it?
I tell them apart by asking about failure and then staying silent. Anyone can narrate a success with judgment painted on afterwards. So I ask a candidate to walk me through a decision they got wrong, and I watch what they do. The person performing judgment gives me a tidy story with a lesson neatly attached, usually one that flatters them. The person who has it slows down, shows me the moment they realised they were wrong, and tells me what it cost. Real judgment carries the memory of being wrong, and it changes how a person holds their own certainty. I also test it live. I will push hard on a view they hold and see whether they can update in front of me, or only defend. The ones who can change their mind in the room, without losing their spine, have the real thing. Certainty is easy to perform. Earned doubt is almost impossible to fake.
EM: If demand for these skills is running ahead of supply, who is failing to supply them, the schools, the employers, the individuals, and where would you put the most blame?
The honest answer is all three, but the blame is not shared evenly. I put the most on employers, and I include my own clients in that. For two decades organisations quietly dismantled the machinery that used to grow human beings. They cut the apprenticeships, the long graduate programmes, the patient mentoring, the tolerance for a young person being unproductive while they learned. They decided it was cheaper to buy a finished leader than to grow one. Now everyone is bidding for the same small pool of finished leaders and acting surprised that it is expensive. Schools carry the second share, because they kept optimising for what is easy to grade, and judgment and empathy are very hard to grade. I put the least on individuals. People responded rationally to the signals we gave them. We told a generation the certificate was the prize, and they went and collected certificates. If we want different people, we have to reward different things, and that has to start with employers, because we are the ones writing the cheques.
EM: Is there a human skill that you think is quietly losing value in this shift, not gaining it? Name it.
Yes, and it is uncomfortable to say out loud. The skill quietly losing value is deep individual expertise held as a private possession. For a whole generation, the expert was powerful precisely because the knowledge lived in their head and you had to come to them for it. The specialist who knew the regulation, the engineer who knew the system, the consultant who knew the benchmark. That knowledge is now a query away, and getting cheaper every month. The expert who built their identity and their authority on being the holder of answers is slowly being repriced down, and many of them cannot feel it happening yet. The expertise still matters, and it works best now as a foundation a leader stands on while making a wise call and carrying people with them. Each year, the value moves further toward the person who can judge what to do with the answers.
EM: For a young professional in the region reading this, what would you tell them to spend the next two years getting good at, and what to stop wasting time on?
I have sat through the same scene many times. Two finalists for one senior role, one who knows more, and one who can make a room understand a complex idea in three sentences. The second one gets the job, again and again, because everyone they will ever lead has to understand them first. So spend the next two years getting good at three things. Learn to write and speak so clearly that people understand exactly what you mean and want to follow it. Clear expression is becoming rare, and it is the visible edge of clear thinking. Learn to make decisions before you feel ready, with incomplete information, and to own them. That is the muscle every senior role will demand and no machine will build for you. And learn to work with these AI tools at a serious level, so you can direct them and catch them when they are confidently wrong. What to stop. Stop collecting certificates for skills the machine has already absorbed. Stop polishing work to perfection that a tool will redo in seconds. And stop hiding behind being busy. The market is about to pay far more for judgment and for the ability to move other human beings, and far less for simple activity. Aim everything there.
EM: AI in Higher Education - This is where the spine lands. UAE universities are making AI a condition of graduation, yet the stated aim is sharper human judgment, not technical fluency. Manolopoulos has hosted public conversations on exactly this. You have sat across from people rethinking the university, including Ben Nelson. What is the most uncomfortable idea about higher education that you came away agreeing with?
I have had these conversations, including with Ben Nelson, who rebuilt the university from a blank page around the idea that you teach people how to think and treat the transmission of facts as the smaller part of the job. The uncomfortable idea I came away agreeing with is that a great deal of what a traditional university sells has very little value left in it. The lecture that transmits information, the memorised content, the degree as a four year container for facts, all of it is being hollowed out now that information is free and instant. What keeps its value is the part most institutions treat as a by product. The formation of a young person's character and judgment. The friendships and the network. The encounter with a mind that changes how you see the world. I find this uncomfortable because I work with universities I deeply respect, and the logic says many of them charge a premium for the part that is fading and give away the part that is priceless. The ones who survive will reverse that.
EM: A university now requires AI to graduate but says the goal is human judgment. Is that a coherent aim, or a contradiction the sector has not resolved? Push on it.
It can be coherent, and in most places right now it is not yet. It becomes coherent when the AI requirement exists in order to build judgment. You put the machine in every student's hands precisely so they learn where it is brilliant, where it is confidently wrong, and what it cannot weigh. A student who has argued with the machine for four years, caught its errors, and felt its limits, comes out with exactly the discernment the university is aiming for. That is a beautiful design. The contradiction appears when an institution bolts on an AI course as one more technical box, teaches students to operate the tool, then writes judgment into the brochure and hopes it arrives on its own. Judgment does not arrive on its own. It is built by making young people wrestle with hard, ambiguous, human problems where the machine cannot rescue them. So the aim is sound. The question I would put to any provost is whether their actual curriculum builds the judgment, or just assumes it.
EM: If a machine can supply any answer, what is the one thing a university must still teach that it cannot outsource? Be specific about how you would teach it.
The one thing is judgment, and I mean the specific human act of choosing well when two good things are in conflict and the data cannot settle it. A machine can tell you what is true. It cannot tell you what is right when two truths collide, because that takes values, context, and the willingness to be responsible for the outcome. How would I teach it. You cannot lecture someone into wisdom, so I would build the whole degree around small rooms where students argue real dilemmas with no clean answer, in front of a teacher who pushes them and will not let them hide in theory. I would make them decide, commit, and then live with a real consequence, even a small one, so the decision costs something. And I would put them next to a person of judgment for years, the way every craft has always passed wisdom down, by apprenticeship and proximity. You learn to judge by judging, badly at first, in the presence of someone who already can. There is no shortcut, and the machine does not change that.
EM: You recruit the output of these institutions. What does a graduate who was "formed," not just informed, look like across a table, and can you actually spot it in an interview?
I can spot it within twenty minutes, and it has almost nothing to do with their grades. This kind of graduate is comfortable saying I do not know, and then thinking out loud with you instead of freezing. They ask better questions than the ones I am asking them, which tells me their curiosity is real and self-driven. They can take a challenge to their view and consider it, turning it over rather than defending a wall. They listen in a way that changes what they say next. And they have a centre, some set of things they believe and can defend, so they are not simply mirroring whatever the room wants. The graduate who was only informed gives me fluent, correct, complete answers and leaves me feeling I have learned nothing about the person underneath. The one who was formed makes me curious about who they will become. After thirty years across the table, that feeling is the most reliable instrument I have, and no transcript has ever given it to me.
EM: Where could making AI compulsory in every degree backfire, and how would you design against that risk?
It can backfire in a precise way. If you put a tool that does the thinking into the hands of a young person before they have learned to think, they will skip the struggle, and the struggle is where the capability is built. A student who lets the machine write every first draft never develops their own voice. A student who lets it solve every problem never builds the mental strength that comes from solving problems yourself. You can produce a generation that is fluent with the tool and hollow underneath, and you will not see the damage until they meet something the tool cannot do for them. I would design against it deliberately. I would protect long stretches of the degree where the machine is switched off and the student has to struggle in the old way, because that difficulty is the point. I would assess the thinking out loud, in person, where you cannot hide behind a generated answer. And I would teach the wisdom of when not to reach for it at all. The goal is a graduate who can use the machine fully and does not need it to think.
EM: Contrarian prompt: argue that the university, as we know it, does not survive this, and that something else should form the person instead.
I can argue it, and parts of it I half believe. The university as we know it is a thousand year old technology built to solve a problem that is disappearing, the scarcity of knowledge and of experts to transmit it. Both are now abundant and nearly free. What remains is expensive, slow, and bundled: four years, one location, one price, for a mix of content, formation, signalling, and friendship that no longer has to be bought together. Unbundle it, and the content goes to the machine, the signal goes to better measures of real ability, and the formation, the part that actually matters, could be done better by other means. Intense residencies. Apprenticeships inside great companies. Small academies built around a master and a craft, the way they existed before the modern university and produced extraordinary people. A young person might one day be formed by two years of real work beside people of judgment, and look back on the lecture hall as a strange detour. I do not fully believe it, because the university also protects knowledge from power and gives the young a place to become themselves before the world demands they produce. That is worth defending. But any university that assumes it is safe has not understood the century it is in.
EM: If we run this interview again in five years, and the UAE has gone further than anyone expected with machines in government and the classroom, what will you want to be able to say held firm on the human side, and what are you most afraid will have been lost?
In five years I want to be able to say one thing held firm. That through all of it, the human stayed the author of the decisions that define us. That we used the machines to carry the load that was crushing us, the volume and the routine and the noise, and we spent the freedom they gave us on the things only a person can do. Sitting with someone in their hardest hour. Making the call that cannot be modelled. Forming the next generation with patience. If we can say that, we will have used this technology to become more human, which is the only use of it worth wanting. What I am most afraid of is quieter and harder to see. I am afraid we will lose the muscles we stop using. The patience to sit inside a hard problem instead of reaching for the instant answer. The comfort of not knowing, which is where real thinking begins. The slow, inconvenient, irreplaceable work of one human being forming another. Those capacities fade quietly, with no dramatic announcement, in a generation that was never asked to build them. My life has been built on a single belief, that in the end the human is the answer, whatever the question. Everything returns to that. My work is to make sure that in five years, and in fifty, it is still true.
