AI can now solve problems at the level of an International Mathematical Olympiad (IMO) gold medalist. And yet, as the robotics lab Physical Intelligence likes to point out, it still can't spread peanut butter on a slice of bread. When AlphaGo beat the world Go champion, the AI reportedly couldn't pick up a single stone by itself. Even now, in 2026, AI can solve a math problem but can't write the answer down with a pencil. This curious phenomenon has a name: Moravec's Paradox.

The Hard Things Are Easy, the Easy Things Are Hard

In 1988, Hans Moravec, a roboticist at Carnegie Mellon University, offered the following observation.

"It is comparatively easy to make computers exhibit adult-level performance on intelligence tests or at playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility."

Things that are hard for humans — multiplying ten-digit numbers, analyzing tens of thousands of legal precedents, running complex statistical calculations — are easy for AI. The reverse is also true: things that are easy for humans, like picking a toy up off a cluttered floor, reading the expression on the face of the person you're talking to, or climbing a flight of stairs, are extraordinarily hard for AI.

Why is that? Moravec traced the cause back to evolution. Abilities like walking, seeing, and touching, he argued, are the product of hundreds of millions of years of optimization by natural selection. Abstract reasoning, on the other hand — math, logic, language — is a very recent arrival in human history and hasn't been optimized nearly as much. And the less optimized an ability is, the easier it is for machines to catch up.

Put simply, the more unconsciously we humans do something, the harder it is for AI to imitate.

What This Paradox Means for Your Career

So what should you actually be building over the course of your career?

Plenty of professionals, hoping to get ready for the AI era, learn to code, earn a data analysis certificate, or pick up prompt engineering. None of that is wrong — but through the lens of Moravec's Paradox, the aim is slightly off. Coding and data analysis are precisely the areas AI is best at.

Apply Moravec's Paradox to your career, though, and a different yardstick emerges.

Any job you can spell out in a manual is at risk.

If you can write "just follow these steps," that means AI can learn the task. Filling in report templates, sorting documents by fixed criteria, arranging data into tables — work like this is already being handed over.


The skills that are hard to put into words are the safe ones.

These are the areas where, asked "Why did you decide that?", all you can say is "I don't know — it just felt right." Picking up on the unease in a client's tone of voice, reading the room and knowing exactly when to speak, sensing a shift in a teammate's mood. None of this can be turned into data, so AI has no way to learn it.

An Era That Puts a Premium on the Unconscious

In January 2026, Arvind Narayanan, a professor at Harvard's Berkman Klein Center, raised an intriguing objection to Moravec's Paradox: it won't hold forever. Computer vision, too, was once something "AI could never do" — and after the deep learning revolution of 2012, it fell almost overnight.

It's a fair point. A robot that peels the plastic off a slice of cheese and lays it on bread will surely arrive one day. But one domain still holds out: the things that happen between people. Building trust, defusing conflict, making a call in an ambiguous situation — won't these remain? They operate on a different plane from sensory or motor skills, because no matter how much data you feed it, AI can't replicate the relational context a person has accumulated over ten years in the same industry.

What Moravec's Paradox should teach us is that competitive advantage in the AI era lies not in "what I'm consciously good at" but in "what I'm doing without even noticing." The abilities you can't put on a résumé, the instincts no certificate can vouch for — these are about to become the most valuable skills of all. Decades of experience are baked into the judgments and the conversations we carry out without a second thought every day. AI doesn't know that yet — and that, in turn, is what gives us our edge.