Artificial intelligence can perform some mind-boggling tasks faster and better than any human. Time and time again, robots have proved their superiority over humans in Jeopardy, chess, and even stock trading.
But strangely enough, AI systems still struggle at tasks that come naturally to human, even toddlers.
Pieter Abeel, a roboticist at the University of California, Berkeley and co-founder of Gradescope, told Tech Insider that what today's AI struggles most with with are simple sensory tasks — including touch, vision, and locomotion.
Called the "Hans Moravec's Paradox," it's a problem that plagues today's most sophisticated AI. The paradox is named after the AI researcher who wrote about the problem in his 1990 book "Mind Children: The Future of Robot and Human Intelligence."
25 years ago Moravec wrote that "it's comparatively easy to make computers exhibit adult-level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception or mobility."
That paradox holds true today.
"This [problem] is well appreciated by researchers in robotics and AI, but can be rather counter-intuitive to people not actively engaged in the field," Abeel told Tech Insider. "Replicating the learning capabilities of a toddler could very well be the most challenging problem for AI, even though we might not typically think of a one-year-old as the epitome of intelligence."
Working on unraveling this paradox has given Toby Walsh, a professor of computer science at the National Information and Communications Technology Australia, a profound respect for the human brain.
"To see my daughter (my wife is German, so my daughter is bilingual) simultaneously learning two languages, just blew my mind about the capabilities of the human brain," he said.
This video, taken at a competition called RoboCup where robots play soccer, exemplifies the Hans Moravec Paradox. The robots struggle to stand, walk, and kick while trying to play soccer.
There is one possible explanation for why vision, bipedal locomotion, and sensory tasks are so difficult to get right. According to Moravec, AI hasn't learned from "a billion years of experience about the nature of the world and how to survive in it."
These years of evolution gave humans remarkable sensory and motor abilities so we can understand threats and take action.
Today's AI also lacks the ability to "look at objects we've never seen before and apply our common sense to understand how they work and what we need to do," in the situation, Walsh said.
"Some of the early pioneers in AI, John McCarthy and his colleagues, identified common sense reasoning as being a fundamental challenge for AI," Walsh said. "There's a huge amount of implicit common sense that we pick up as children and apply to our daily lives that we're perhaps not actually that aware of."
Join the conversation about this story »
NOW WATCH: Only 6% of Americans surveyed can answer these basic science questions — how would you do?