Short answers, pulled from the story.
Moravec's paradox is the observation, articulated by Hans Moravec in 1988, that computers find it comparatively easy to match adult human performance on intelligence tests or games like checkers, but find it difficult or impossible to replicate the perception and mobility skills of a one-year-old child. The paradox was also developed in the 1980s by Rodney Brooks, Marvin Minsky, and others.
The most widely cited explanation is evolutionary: ancient skills like recognizing faces, walking, and catching a ball have had roughly a billion years of natural selection to be refined in animals, while abstract reasoning like mathematics is only a few tens of thousands of years old. Because older skills are buried in unconscious, highly optimized brain processes, they are far harder to reverse-engineer than recently developed abstract abilities.
The observation is associated primarily with Hans Moravec, who wrote it in 1988, though Rodney Brooks, Marvin Minsky, and others articulated the same idea during the 1980s. Steven Pinker called it the main lesson of thirty-five years of AI research in his 1994 book The Language Instinct.
In the 1980s, Brooks founded a research direction he called Nouvelle AI, built around the principle of "no cognition, just sensing and action," deliberately setting aside the logic-and-symbol approach that had dominated early AI. His goal was to build machines grounded in perception and movement rather than abstract reasoning.
Minsky emphasized that the hardest human skills to reverse-engineer are those below the level of conscious awareness. He wrote: "In general, we're least aware of what our minds do best," and added that we are "more aware of simple processes that don't work well than of complex ones that work flawlessly."
By the 2020s, computers were hundreds of millions of times faster than those of the 1970s, and machine-learning systems had begun to handle perception tasks as Moravec predicted in 1976. However, there is currently no consensus as to which tasks AI tends to excel at, so the paradox remains a live question in the field.