The Skill That Can't Be Automated

Daniel Pink recently laid out six human skills that become more valuable as AI advances. The first one on his list: asking better questions. His argument is straightforward. AI is extraordinarily good at generating answers. It can produce a hundred plausible responses to any prompt in seconds. The cost of answers is collapsing. Which means the scarce resource was never the answer. It was knowing which question to ask.

The idea has stayed with me because it names something that happens in every leadership meeting where I’ve facilitated a decision.

#The Wrong Problem, Solved Beautifully

Leadership teams fail more often from misdirected effort than from poor execution. A room full of smart, experienced people solving the wrong problem with impressive efficiency.

It looks like productivity. The conversation is energetic. People are engaged, contributing, building on each other’s ideas. Solutions emerge. Decisions get made. And then weeks later, the real issue surfaces anyway, because nobody paused long enough to ask whether the question they were answering was the right one.

The urgency to solve overtakes the discipline to define. This is human. We’re wired for it. An open question creates discomfort, and solutions relieve that discomfort. So we rush toward resolution, and the quality of the question goes unexamined. AI accelerates this. When you can generate and execute answers faster than ever, the cost of the wrong question compounds at machine speed.

#The Research on Question Formulation

Hal Gregersen at MIT has spent years studying how the world’s most innovative leaders think. He interviewed over 200 of them, and found that they exhibit a disproportionately high question-to-answer ratio. In transcribed interviews, they ask significantly more questions relative to the answers they give. They spend more time on formulation than most people spend on solutions.

This resonates with the question-behavior effect I’ve written about before. Asking someone a question activates a fundamentally different cognitive process than giving them an answer. The question creates internal reasoning. The answer creates evaluation. These produce different qualities of thought and different levels of commitment to whatever follows.

Gregersen calls the best questions “catalytic.” They dissolve barriers to thinking and redirect the search into pathways nobody was considering. His question burst methodology works on this principle: instead of brainstorming answers, a group spends a concentrated period generating only questions about a challenge. No answers allowed. The result is consistently a reframing of the problem that opens up solutions the group hadn’t considered.

The question shapes the search space. Get the question wrong and every answer, no matter how sophisticated, is irrelevant. Get the question right and the answers often become obvious. AI can generate those answers instantly. It cannot generate the question. Framing requires contextual judgment that no model has.

#Why This Was Always the Bottleneck

There’s a quote attributed to Einstein: if he had an hour to solve a problem, he’d spend 55 minutes defining the problem and 5 minutes solving it. Whether he actually said it is debatable. That it rings true to anyone who’s worked on hard problems is not.

The ability to frame a problem well has always been the highest-leverage skill in any room. The person who can look at a messy situation and articulate the actual question underneath the noise creates more value than the person who generates the fastest answer to whatever question happens to be on the table.

This was true before AI. But AI made it visible by removing the other constraint. When generating competent answers required significant time and expertise, both skills mattered. Now that answers are nearly free, the asymmetry is exposed. The question is the bottleneck. It always was. We just couldn’t see it clearly because answers were expensive enough to seem like the scarce resource.

#What This Means for Leadership

The leaders who consistently produce the best outcomes are the ones who slow the room down long enough to get the question right.

This is harder than it sounds. The pressure in any meeting is toward resolution. People want to solve, decide, move forward. The person who says “wait, are we asking the right question?” risks looking like they’re slowing things down. And they are slowing things down. That’s the point. The ten minutes spent interrogating the question saves the three weeks spent executing the wrong answer. In an AI-accelerated environment, those three weeks compress into three days, and the wrong answer gets implemented with more confidence and less friction. The cost of a bad question is higher than it’s ever been.

The discipline is specific: resist the pull toward solutions long enough to examine the problem. Ask what’s actually going on. Ask what would have to be true for the proposed solution to fail. Ask what question you’d be asking if you weren’t anchored to the one already on the table.

#Something to Try

Before your next meeting where a decision needs to be made, spend the first ten minutes only asking questions. No solutions, no proposals, no evaluations. Just questions about the problem. Anyone can contribute a question. Nobody is allowed to answer one.

It will feel unnatural. The room will want to solve. That pull toward resolution is exactly the thing worth examining. Notice what happens to the quality of the conversation once the room has been forced to think about the problem before jumping to solutions.

The skill that can’t be automated is knowing what to ask.