Why MBA education must evolve in the age of AI
As artificial intelligence commoditises data synthesis and analysis, modern leaders require the strategic conviction to make critical decisions
Reflecting on his time at IIM Ahmedabad and his tenure leading BCG India, an experienced executive explores how modern management education, crisis leadership, and decision-making under uncertainty have evolved in the age of artificial intelligence.
Reflecting on his time at IIM Ahmedabad and his tenure leading BCG India, an experienced executive explores how modern management education, crisis leadership, and decision-making under uncertainty have evolved in the age of artificial intelligence.
Reflecting on his time at IIM Ahmedabad and his tenure leading BCG India, an experienced executive explores how modern management education, crisis leadership, and decision-making under uncertainty have evolved in the age of artificial intelligence.
I STILL REMEMBER the embarrassment of a quant class at IIM Ahmedabad: cold-called on a case, stumbling through it in front of 50-plus classmates, the professor unconcerned about my discomfort. We were never given enough information. That was the point, though none of us saw it then. We complained about incomplete exhibits the way people complain about the weather. Three decades later, I understood that the missing data and learning to operate outside one’s comfort zone was the curriculum.
The MBA at IIMA did not give me just a body of knowledge. Most frameworks I learned in 1994-96 are now obsolete or free on the internet. It gave me the ability to decide with 60 per cent of the facts, and a structured way of asking what the other 40 per cent would have to look like for me to be wrong. In a crisis, the cost of waiting for clarity is almost always higher than the cost of being partially wrong and correcting fast. B-school taught me to break a large problem into manageable pieces, find the data that answers each, and assemble the jigsaw. It also taught me, through a hundred cases about companies that no longer exist, that most failures are not failures of analysis, they are failures of decision, or of timing.
I had the privilege of leading BCG in India for six years, and the pandemic fell bang in the middle. March 2020 is the only time in my career I have felt the ground move under an entire population at once. We had 4,500 people across five offices, every one of them sent home within a week, a client base going into simultaneous shock, and no precedent to lean on. I have never been more grateful for a training that distinguishes simple, complicated, complex and chaotic situations: that situation demanded an “Act, Sense, Respond” model, and not the “Probe, Sense, Analyse, Respond” model of ordinary times.
The first question was not commercial: are our people safe, and do they believe we will look after them? Everything that followed rested on the answer. We ran a small health services provider—a near-hospital in rented rooms, with doctors and nurses, an ambulance on standby, oxygen concentrators on tap. We made an early call not to take headcount out, and said so out loud before we knew we could afford it. On a spreadsheet these calls were indefensible. On any reading of what an employee-oriented professional services firm is, a balance sheet of trust and nothing else, it was the only option. My management education did not hand me that answer. It gave me the discipline to be explicit about which objective I was optimising for, and over what horizon, so a call that looked irrational annually was coherent over five years.
The second question was about our clients, and here the case method earned its keep. Every Indian company we worked with was asking the same thing: how much of this is a shock to demand, and how much to the model? Those require opposite responses. We spent that year helping boards separate the two. The ones who got it right cut cash burn hard while protecting the two or three capabilities that would matter on the other side, and came out of 2021 stronger. The move is unglamorous but learnable: decompose the shock, put a time constant on each component, and refuse to treat a cyclical problem as structural, or the reverse.
Which brings me to the most interesting question: whether long-range strategy has quietly become a euphemism for crisis adaptability. The premise is half right, and it matters which half. What has died is the five-year plan as forecast, the hockey stick propped on assumptions nobody revisits. It deserved to die, and most Indian boards have stopped pretending otherwise. What has not died, and matters more than ever, is strategy as a set of choices about where to play and what to excel at.
Adaptability is no substitute. A company with no point of view adapts in circles. The firms I admire most have shortened the planning cycle to rolling three-year plans with granular annual ones, while lengthening their conviction to a decade. That combination is harder than either extreme, and it is what strategic management, taught properly, should prepare people for.
AI sharpens all of this rather than replacing it. The analytical phase of a strategy question, once six weeks and a team of four, now takes hours. Synthesis is cheap now. What has not become cheap is judgment about which question is worth asking, and the ability to get an organisation of 30,000 people to do something differently. If anything, AI raises the premium on what an MBA is meant to develop and often does not: framing the question, sitting with ambiguity, the nose for an opportunity, the willingness to own a decision you cannot fully defend with data. One caution: AI makes it very easy to produce something that looks like a strategy. Leaders will need to become better, not worse, at telling the difference.
So where does management education go? Less technical content, more judgment. The lecture on the value chain is free; what is not is two years among sharp, argumentative peers forcing you to commit to a position in public and defend it. I would want tomorrow’s programmes shorter and recurrent, people returning at 35 and 45 when the questions have changed. I would want far more time on the human dimension: how to hold an organisation together when plan A has failed, which is the actual job. And I would want AI treated as a collaborator to be interrogated, not a calculator or a threat.
The case method understood something in 1994 we are only now catching up to. Incomplete information is not a flaw in the exercise. It is the basis of the work.
—As told to Abhinav Singh
The author, an alumnus of IIMA, is managing director and senior partner, BCG.