Moravec's Paradox ☆

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.