Moravec's Paradox
The observation that some tasks easy for people are difficult for computers, while some demanding cognitive tasks are easier to automate.
Why AI Can Crush Chess but Still Struggles Like a Toddler in a Kitchen
You’d think the hard part of intelligence would be the big brain stuff. Chess. Calculus. Writing code at 2 a.m. with dead eyes and cold coffee.
Nope.
Moravec’s Paradox is the weird truth that machines are often great at things you think are hard, and terrible at things you do without thinking. Like walking across a messy room, picking up a mug, or not driving like a confused uncle in a rental car.
The weird bit
A computer can beat world champions at games built on logic and rules.
But ask a robot to:
- fold laundry
- crack an egg
- walk up stairs without looking haunted
Now we’ve got a situation.
That’s Moravec’s Paradox. Hard reasoning is relatively easy for machines. Easy physical skills are brutally hard.
Why?
Because the stuff you call “easy” took millions of years of evolution to build into your brain and body.
You don’t calculate how to catch a falling spoon. You just do it. Your eyes, hands, balance, touch, timing, all coordinate instantly. It’s less like solving a math problem, more like running a nightclub with no manager and somehow everything still works.
Logic is different. It feels impressive because you have to think about it. But for computers, step-by-step reasoning is their whole thing. They love rules. Rules are their IKEA instructions.
Real-world movement is chaos. Lighting changes. Floors are slippery. Objects squish. Humans leave socks everywhere. The physical world is basically a prank show for robots.
Why you should care
This is why AI can write a decent email, analyze legal docs, or ace an exam, yet still struggle with jobs that involve hands, bodies, and messy reality.
So when someone says, “AI is basically human now,” remember this. Passing a test is one thing. Loading a dishwasher without breaking a wine glass is another.