The Fourth Dimension: Building Human Capability into the AI Strategy with Rahim Hirji, Superskills

Terry O’Dwyer: Welcome Rahim Hirji. You’re the founder of Superskills, an advisory practice built around human capabilities that AI can’t replace. You’ve spent three years talking to more than 200 organisations on this. What is it that you’re seeing that most people are missing?

Rahim Hirji: The thing I’d say is that most companies have got a process in place. They say they’ve got all of their AI strategy covered off: process, automation, cost reductions, revenue opportunities and new products. These are all legitimate. But in my research, I think they’re missing this whole other dimension — and it irks me a little bit because it’s what I call the fourth dimension, which is human capability. I think it’s the game changer. It sits across all of the AI prongs or pillars you might have.

Rahim Hirji: The way AI is being perceived by leaders is often black or white — you’re either on the boat with it or you’re off. But I see it very differently. I think that people can become more human in this moment and not less, and they can lean into a different set of skills: curiosity, change readiness, empathy. These are the super skills my book is based on. But the game-changing question that almost nobody’s asking is: what does an organisation need its people to become to succeed in the new world? I don’t think people think about the human bit first — they think about the tech first. And I think that’s a mistake. Because this is the first time that technology has come for our brains. We’ve lived through multiple different eras and every other era came for different things — our hands, our processing speed. This one is coming for the thing that we thought would make us irreplaceable. And we’re still thinking about it as a tool problem. CEOs are still thinking about it as a tool problem. We need to reframe AI strategy as a thinking problem. Then the conversation shifts.

Terry O’Dwyer: What do you mean by human capability in layman’s terms?

Rahim Hirji: Let me give you an example. Back in the day, when I was younger, if you’d go and get insurance, you’d go to your insurance broker. He worked out of his garage. He had his desk, his big leather chair, a biscuit tin with penny coins because insurance cost 19.99. He thought his value was in writing out the name and all the details in the form. But the real thing I saw in him was trust. I trusted him. I’d go back to him every year. The thing is: it’s really about the value that humans bring. People forget about that value. There is this value in this work, and you have no idea what your value is at this point in time because everything’s been thrown up. It’s about taking a step back and saying: this is where my value lives as a human.

Rahim Hirji: I consult across multiple different industries. I spoke to a bank recently that was very pleased it had automated part of its research function. Five years ago, before ChatGPT, nobody was saying: we need to automate research for cost savings. No one was asking that question. And AI showed up and they said: oh, we can do this. But I said: this is a big mistake. Because research is where someone’s actually using their brain to ask the right questions and to decide what the business does next. What they were doing was outsourcing that bit to the machine. Let’s assume they had a team of ten, which I think is roughly what they had, and reduced that to a team of two. They had missed out half the questions that they used to ask before. They were no longer asking them because they’d lost all the wisdom from people who’d been in that industry and seen different waves in the past. Secondly, there was no one really to provide oversight — to make sure the AI is covering the right areas — because they didn’t have all the know-how on what was going on. Subsequently, they’re now bringing some of that team back because they realised they were missing a trick. What happens is AI comes in as a process, a procurement item, and gets rolled out through tech or IT. It gets measured in different metrics — maybe licences activated or how many times people use AI — but it never really gets to the CEO’s desk. And you can’t lead with what technology can do. You have to lead with where you want to go as an organisation and make that decision.

Terry O’Dwyer: For a mid-market business where oversight sits with someone already doing three other jobs, how do you guard against failures caused by incomplete data?

Rahim Hirji: The thing that terrifies me is incomplete data. You walk into this world where AI is the default and not the exception, and the human stuff is the context that AI can’t know or feel. It doesn’t know the relationship. It doesn’t know the conversation you haven’t recorded because it was inappropriate to record it. It doesn’t know the gut feeling you have because you’ve spoken to someone a number of times in person. There’s no field for it on the system. No dashboard for that type of stuff.

Rahim Hirji: The guard for that is not more oversight applied more regularly — you haven’t got time for that. It’s asking one question of every workflow: where in this process does a human need to be present? Is it at the beginning? Is it at the end? Is it when four or five different things come together? Is it before you send things out? Where could things coalesce in a way that the agent won’t catch? I don’t think it needs to be everywhere. I think it’s just at one specific point in time. It’s going to be different in different industries. But I see oversight as a new era or new area that humans are going to move into — it’s currently underrated. Everyone talks about acceleration, deployment, adoption. But no one talks about oversight as a discipline. There’s a level of wisdom to it. And oversight is only useful if you have the power to intervene. Intervention is what makes you human in the loop, not just witnessing the information. If you can only monitor what’s happening, you’re not overseeing it — you’re just a friend to the agent. Real oversight is when you can ring the alarm, stop the machine and say: something’s not quite working. We need to look at this differently. There’s a level of authority that people in this new managerial role of oversight are going to come into.

Terry O’Dwyer: Where should a mid-market leader start?

Rahim Hirji: You need to reposition for this new world. It’s a thought experiment of sorts. Let’s assume AI takes 80 per cent of what you’re doing off your plate tomorrow. Don’t worry about your job — let’s assume all those things are okay. What would you do with your time? Most leaders in mid-market companies say: well, I’d do more of what’s left. I think that’s completely the wrong answer. We’re moving from a world of knowledge management and knowledge thinking — that’s old school thinking. It was a world where productivity was highly regarded. If you had more Outlook pings and more Slack messages, you were doing lots of work. I just call that busy work. And I think we’ve been mistaking that productivity for quite a long time.

Rahim Hirji: What you’d do is look at what you’re doing right now and start to protect your thinking. Once a week, sit down as an individual — and then come together as an organisation — and say: this is what I’m going to be doing for the following week and use that time for thinking time. I remember being in organisations where people would block time off and I thought: that’s a bit arrogant, they’ve got time to think. I had a very busy diary and was doing the stuff. In fact, they were the ones who were right and I was the one who was wrong. Because thinking time is what’s actually going to become really important. And we should stop saying human in the loop and start saying human first. When I’ve got a problem, instead of going brain to fingers to keyboard — which is what most people do, or increasingly are starting to do with prompting — I start writing first. I start scribbling to try and figure out those problems.

Terry O’Dwyer: So it’s protecting that authorship time — AI-free, computer-free space — before diving straight into ChatGPT?

Rahim Hirji: Yes, absolutely. And I think it is moving from a world of execution to authorship. For the last 25 or 30 years, execution was really important. If you could research it, build it, format it, coordinate it, optimise it, make it programmatic — you were in a great position. Now we’re in this world where agents do all of that. And the thing they can’t do is decide what’s worth executing, which is different. They can’t set the direction. They can’t weigh up competing values or author intent. That judgment piece becomes really important. When you get to authorship, the important conversation shifts from collecting the data and doing the busy work to: what are we going to do with that data? What are the questions we’re going to take forward?

Terry O’Dwyer: Tell me briefly about the seven super skills.

Rahim Hirji: The seven super skills are: curiosity, change readiness, big picture thinking, principled innovation, empathy, global adaptability, and the augmented mindset. They all work in a specific order. The two I think are really important for mid-market companies are curiosity and the augmented mindset — curiosity is first on the list and the augmented mindset is last, and they form an arc. I think curiosity is the skill that’s most under threat right now because of AI. People come back to me and say: well, everyone’s curious. No one says they’re not curious. And I say: well, it’s not an obvious one because we all think we’re curious. Curiosity is the skill that’s most under threat right now because of AI. The reason is that we’re tricking ourselves into thinking we’ve come up with the answers when we’ve just gone to the AI. We’re losing the muscle of being curious and asking questions of every single thing that we’re doing. It’s more important to be asking the questions before you get to the AI than after. Having a hypothesis before you get to the AI is really powerful. Take a step back and note those things down. Why was I thinking about that in the first place? Was I thinking about that because I was influenced by an algorithm or something I read? Or is this something that’s forming in my head and I need to form it myself first?

Terry O’Dwyer: And the augmented mindset?

Rahim Hirji: In the interest of time: human plus AI is more powerful than humans alone, or more powerful than AI alone. The example I bring is Tony Stark in Iron Man. Tony Stark isn’t great because of the suit. He’s great because he decides when to wear it and when not to. The suit gives him extra capabilities, these superpowers, and he brings intent, he brings the judgement we’ve been talking about, he essentially brings the whole game. And crucially, he makes the decision when to wear it and when not to. That’s really what the new world is going to be like. When do you think and use your brain and use your capability and use your judgement? And when do you say: hey, I’m going to press the button?

Terry O’Dwyer: How do we distil this into something practical for a C-suite team at a mid-market business?

Rahim Hirji: A couple of things. Status updates are going to happen and they should be happening. The question is: what questions do you need to ask that you could be doing? I was working with a company that, instead of having a status update meeting, were voting on the questions that were most important to the business going forward. The reason was that they didn’t know what the world was going to turn into. The North Star that a company’s had — five or ten years out, these are the things we’re going to be doing — has changed because the world is moving so quickly. You need to start pivoting more quickly. Sit down and say: in our industry, with competitors who are going to have access to the same technology, what is it in our people that’s going to differentiate us amongst the competition?

Rahim Hirji: The other thing: back in the late 1990s and early 2000s, you’d have X-teams — the people who are most au fait with the technology and the strategy of the business — getting stuff done and energising the organisation. Especially mid-market ones who are right at the point where they could grow in a massive way, but could also die off quite quickly — get the people who are going to spark this change, sit down with them and say: this is your new remit for the next one or two years, make that change happen. It’s not just one person. It’s multiple people with different mindsets coming together. They have a shared feel for what’s going on in the market, in the industry, with AI and the specific competition. But they also have the human capability to think outside of the box and see what’s actually possible.

What is the fourth dimension of AI strategy that most organisations are missing?

Rahim Hirji argues that most AI strategies address three dimensions: process automation, cost reduction, and revenue and product opportunities. The fourth dimension — human capability — sits across all three and is the one that determines whether any of them actually stick. His core question is one most organisations never ask: what does our organisation need its people to become to succeed in the new world? Without answering that question, AI is treated as a tool problem rather than what Rahim calls a thinking problem. The tools matter. But the human capability to direct them, question them and lead with judgement is what separates organisations that scale AI from those that stall.

What happens when organisations automate research functions with AI?

Rahim Hirji cites a financial services firm that reduced its research team from approximately ten people to two, using AI to handle the research function. The problem was not the technology — it was what the organisation lost along with the people. Research is where humans ask the right questions and decide what the business does next. When the team shrank, half of the questions the business used to ask simply stopped being asked. The institutional wisdom of people who had been in the industry through different cycles disappeared. There was also no longer sufficient human oversight to catch what the AI was missing or getting wrong. The firm subsequently began rebuilding the team. Rahim uses this as a case study for the difference between automating tasks and stripping out the thinking that gives those tasks meaning.

What is oversight as a discipline and why is it underrated?

Rahim Hirji argues that oversight — the active management of what AI is doing — is becoming a critical human function that most organisations have not yet built properly. His distinction is important: monitoring is not the same as oversight. If you can only observe what the AI is doing but cannot intervene when something is wrong, you are not overseeing it. Real oversight means having the authority, the knowledge and the mandate to ring the alarm, stop the process and say: something is not working here. He advocates for asking one question of every workflow: where in this process does a human need to be present? The answer will be different in different industries, but it should always be answered before the workflow is deployed.

What does it mean to move from execution to authorship?

For the last 25 to 30 years, Rahim Hirji argues, execution has been the primary form of professional value: research, build, format, coordinate, optimise, make programmatic. AI agents can now do all of that. What they cannot do is decide what is worth executing, set direction, or weigh up competing values and author intent. Authorship is the shift from doing the work to deciding what work is worth doing and what questions are worth asking. In practice, this shows up in meeting design: rather than status update meetings where someone has pulled the numbers, reformatted charts and chased updates through Slack, authorship means the data is already there and the conversation becomes: what are we going to do with this data? What are the questions we need to take forward?

What is the augmented mindset and how is it different from just using AI more?

The augmented mindset is Rahim Hirji’s seventh super skill and the culmination of his framework. His definition is simple: human plus AI is more powerful than either alone. The Tony Stark metaphor captures what makes it distinct from simply using AI more: Tony Stark is not great because of the suit. He is great because he decides when to wear it and when not to. The augmented mindset is about judgment over when to deploy AI and when to rely on your own thinking instead. In an age when AI is becoming the default, protecting the moments when you think first — before reaching for a tool — is what keeps that judgment muscle functioning.

Why is curiosity the most under-threat skill in the age of AI?

Rahim Hirji’s argument is that most people believe they are curious, so they do not think of it as a skill at risk. The problem is more subtle: people are increasingly going to AI tools before forming their own hypothesis or question. They get an answer back, it looks complete, and they feel as though they have done the thinking. But the intellectual work that should have preceded that prompt — the questioning, the framing, the hypothesis — did not happen. Over time, this erodes the curiosity muscle. His practical prescription is to write before you prompt: get something rough on paper, form a hypothesis, identify what you actually want to know and why, before opening any tool. This protects the thinking that makes human input worth having.

What practical steps should mid-market C-suite teams take now?

Rahim Hirji offers two concrete recommendations. The first is to shift from status update meetings to question-prioritisation sessions — rather than reporting what has happened, teams vote on the questions most important to the business going forward. This is particularly relevant at a time when the North Star of a five-year plan can become obsolete within months. The second is to form an internal X-team: a cross-functional group of people who are au fait with AI and the strategic direction of the business, given a clear remit to drive change over one to two years. This group should include people with different mindsets and perspectives, not just technical specialists, and should be empowered to think outside the current operational constraints of the organisation.