Moravec's Paradox
Hard things are easy, easy things are hard — still true, still shaping robotics.
You're looking at a concept explainer: what this term means in plain English, and the records it ties together.
Moravec observed (1988) that abstract reasoning — evolution's newest trick — is computationally easy, while perception and dexterity — refined over hundreds of millions of years — are brutally hard. Fifty years of AI confirm it: models pass bar exams before robots reliably fold laundry. It explains why the robotics entries on this site trail the language ones, and why world models and fleet-scale robot data are the current bets to close the gap.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Related
Sources
Moravec, Mind Children (1988) ↗
Record reviewed as of 2026-01-15 — see Methodology.