Skip to content
Surprising Science

Why Computers Are Actually Rather Stupid

During its famous Jeopardy! match, IBM’s natural language computer named Watson was given the clue, “What grasshoppers eat.” When this computing marvel answered, “Kosher,” it looked dumb. 
Sign up for the Smarter Faster newsletter
A weekly newsletter featuring the biggest ideas from the smartest people

What’s the Latest Development?


When the world’s most advanced form of artificial intelligence, a computer capable of understanding the natural language of humans, debuted on Jeopardy!, it took home the cash but it gave some spectacularly stupid responses. To the clue, “What grasshoppers eat,” Watson replied: “Kosher.” Just ten years ago, the most innovative Internet search engines were incapable of understanding that President Clinton was the same person as Bill Clinton. That search engine, Ask Jeeves, barely overcame 1966 natural language technology. 

What’s the Big Idea?

Artificial intelligence, though the dream of many computer enthusiasts, seems a long ways off. The computer visionary Alan Turing, who broke the German ENIGMA code during World War II, thought that computers would achieve 128 megabytes of memory and successfully imitate humans by 2000. He has proven right about computer memory but terribly wrong about computer intelligence. In a 1989 experiment, a computer program passed the ‘Turing Test’, i.e. a human was made to believe the computer was also a human, by asking series of profane and aggressive questions.

Photo credit: shutterstock.com

Sign up for the Smarter Faster newsletter
A weekly newsletter featuring the biggest ideas from the smartest people

Related

Up Next
Russia’s intellectuals, creative class and Golden Youth – Russia’s modern-day Decembrists – failed to bring about revolutionary political change this weekend in Russia despite an enthusiastic past three months that […]