Nobody Defined Critical Thinking. They Just Started Demanding It.

Korn Ferry’s 2026 Talent Acquisition Trends report landed with a stat that should stop you mid-scroll: 73% of talent leaders now rank critical thinking as their number one hiring priority, ahead of every AI-specific skill on the list. Scott Erker, one of the firm’s skills experts, put it bluntly: he can’t imagine someone being great at AI without exceptional critical thinking skills.

I’ve spent two posts on this site circling this gap. In 2024 I watched the Apple Intelligence and Grammarly demos and wondered out loud whether we’d quietly handed our thinking over to our tools. More recently I wrote about the paradox underneath the corporate enthusiasm for critical thinking: everyone claims to want it, almost nobody tolerates it when it shows up, and the freshman-to-senior data out of higher ed shows the skill barely moves across four years of college. Both posts were diagnosis. This one is an attempt at something more useful: a definition.

Here’s the question the Korn Ferry stat raises that nobody in the report answers. If 73% of talent leaders are suddenly desperate for critical thinking, what exactly did they think they were hiring for over the last several years? The skill didn’t just become valuable in 2026. It became *named* in 2026 — which means for a long stretch before that, organizations were either not asking for it or not recognizing it when it showed up.

Part of the problem is that “critical thinking” means whatever the person using the phrase needs it to mean. In a job posting it usually means “doesn’t need much hand-holding.” In a leadership meeting it sometimes means “agrees with me, but with better-sounding reasoning.” And it’s worth noting that we’ve also lived through a stretch of public life where questioning expert claims got treated less like a skill and more like a liability. That instinct doesn’t just disappear the moment a talent report decides the skill is suddenly back in demand.

So let’s name the skill instead of just demanding it.

Here’s what critical thinking actually is. You take in information, form a hypothesis about what it implies, and identify a test — a concrete, verifiable consequence that would have to be true if the hypothesis holds. If a particular food is causing a skin reaction, the hypothesis generates a prediction: cut the food, the reaction should fade. You’re not guessing and you’re not doubting for sport. You’re building an if-this-then-that chain and checking whether the world actually behaves the way the chain predicts. Not contrarianism. Not devil’s advocate for the sake of friction. Not “doing your own research” in the way that phrase has come to mean instinctively distrusting anything you didn’t personally arrive at.

I grew up doing a version of this before it had a name. Bikes first — chains, derailleurs, figuring out why something wouldn’t shift right. Then cars, starters, alternators, batteries, oil! Then, in my college years, troubleshooting computers, where one wrong jumper setting or line of code meant the whole thing wouldn’t work and you had to backtrack through every possible cause until you found the real one. No YouTube. That instinct never left. I’m still doing it today — for vendor pitches, for AI tools that hand me a confident-sounding answer, and most days, as the unofficial IT department for my family.

That’s pattern recognition, sequencing, and hypothesis testing — skepticism as method, not skepticism as identity.

If organizations keep using “critical thinking” as a vague, feel-good label next to “strong communicator” on a job requisition, they’ll keep hiring for it without being able to test for it or recognize it when an employee actually does it. Worse, they’ll keep building AI workflows that quietly reward the opposite: accept the model’s first answer, move fast, don’t slow down to interrogate the output. You can’t want hypothesis-testing minds while optimizing every process for hypothesis-skipping speed.

So start with the mechanism, not the buzzword. Pick one recurring decision in your work — a recommendation an AI tool hands you, a trend your team assumes is true, a claim in a vendor pitch — and run it through the sequence: what’s the pattern, what does it predict, what’s the cheapest way to check that prediction before you act on it. Do that consistently and you’ll have actual evidence of whether you, or the people you’re hiring, can do the thing everyone suddenly claims to want.