FOR YEARS the conventional wisdom on the relative cognitive strengths of humans and machines has held that humans excel at recognizing faces and other kinds of pattern matching, while computers rule on anything that smacks of number crunching. That may no longer be the case. The success of Foldit—an online puzzle created by biologists and computer scientists at the University of Washington—proves that human intuition can outperform computer algorithms on complex scientific problems.
Foldit presents players (all nonscientists) with a partially folded protein on-screen and challenges them to twist it into an ideal shape based on simple rules. Not only did players predict correct protein structures much more quickly than any algorithm could (a brute-force search of all the possibilities would take millions of years), they were also able to intuit solutions that a computer might never have found at all. “To fold a protein into the right shape, you might first have to bend it in a couple of directions that seem totally wrong,” says Seth Cooper, a Washington computer scientist and one of Foldit's inventors. “A human being playing with a virtual object can see the big picture and recognize those tricky solutions.”