The Answer Is Cheap. The Decision Is Not.
Why employers increasingly value people who can make the call and explain the tradeoffs.
A founder I spoke with recently was not worried about whether graduates could produce an analysis. AI had already solved much of that problem.
What he wanted to know was simpler and harder: “why did you make that call?”
He was not asking which framework appeared in the deck, how many sources had been consulted, or whether the presentation looked professional. He wanted to understand the reasoning, the tradeoffs, and whether the person in front of him understood the consequences of the decision.
That is the shift business schools need to take seriously.
As intelligence becomes cheaper, judgment becomes more valuable.
The founders and companies I speak with are not short of information. They’re literally drowning in it. What they need are people who can decide what matters, reject what does not, choose a direction, and explain why.
That is increasingly a form of triage.
AI can research a market, challenge assumptions, generate options, identify risks, and polish the final presentation. The problem begins when a plausible answer is mistaken for an owned decision.
A plausible answer can sound intelligent without anyone truly believing it. An owned decision requires someone to say: this is the option we chose, these are the alternatives we rejected, this is the evidence we trusted, and these are the consequences we are prepared to accept.
Yet most business school assessments are not built around that.
The system rewards completion, coherence, confidence, and a numerical grade. Students quickly learn that a polished answer is safer than an uncertain decision. AI has not created that weakness; it has simply made it much easier to see.
So I give students more constraints, not fewer.
They complete several team and individual submissions, each one building on the last and carrying forward the consequences of earlier decisions. They must choose left or right, reject alternatives, define thresholds, and explain what would make them stop.
They cannot hide behind “it depends.”
As I have written before, I use a continuous incentive progress system. Students receive precise, actionable feedback in small instalments throughout the course, with bonus or penalty points as their work develops. One weak performance does not destroy the course, and one polished submission does not rescue it.
But I also punish polish when it outruns proof.
And from my experience, this substantially changes how students use AI. It stops being a machine for producing the answer and becomes part of the reasoning process. They use it to test logic, expose assumptions, compare alternatives, and sharpen a decision they still have to own. As well as strengthening the decision making logic, and gives the ending more weight without becoming theatrical.
I acknowledge that this is easier in strategy, innovation, and entrepreneurship, where there is rarely one correct answer. In mathematics or engineering, the final answer may be fixed. Even there, the path matters. Employers still want to know how someone approached the problem, what they tested, what they rejected, which tradeoffs they accepted, and why they trusted the result.
But in my view, the larger obstacle is time.
Judgment does not develop in a 15 hour elective. It comes from repetition, mistakes, feedback, memory, and the slow process of making ideas your own. Yet business schools keep compressing courses while expanding grade requirements and ECTS accounting.
That is not a cheap criticism. The system was built for a different age, and changing it will take time.
But the direction is clear.
AI will keep giving students more information, more options, and more convincing answers. But producing intelligence will no longer distinguish them.
The scarce graduate will be the one who knows when enough intelligence has been produced, understands the tradeoffs, chooses what not to do, makes the call, and accepts responsibility for what happens next.