Forget the Singularity
Ray Kurzweil has popularized the idea of the "singularity," the point when computers will become more intelligent than humans, changing everything.
This idea assumes that there is one form of intelligence, general intelligence, and that computers are getting better at it. In reality, there are many forms of intelligence.
Machines became more intelligent than humans at doing arithmetic in the nineteenth century, when the first mechanical adding machines were invented.
Computers became more intelligent than humans at playing chess about a decade ago, when computer began to regularly defeat the world's greatest chess players. But the computers play differently from humans: human chess masters think about whether a position gives them a strategic advantage, while computers calculate every possible combination of moves many moves in advance.
Rather than a single moment when computers become generally more intelligent than humans, computers will gradually become more intelligent than humans at one skill after another as programs are developed that reduce one skill after another to calculations - as chess was reduced to calculations.
And it seems that many forms of human intelligence can never be reduced to calculations, so computers will never become good at them.
This idea assumes that there is one form of intelligence, general intelligence, and that computers are getting better at it. In reality, there are many forms of intelligence.
Machines became more intelligent than humans at doing arithmetic in the nineteenth century, when the first mechanical adding machines were invented.
Computers became more intelligent than humans at playing chess about a decade ago, when computer began to regularly defeat the world's greatest chess players. But the computers play differently from humans: human chess masters think about whether a position gives them a strategic advantage, while computers calculate every possible combination of moves many moves in advance.
Rather than a single moment when computers become generally more intelligent than humans, computers will gradually become more intelligent than humans at one skill after another as programs are developed that reduce one skill after another to calculations - as chess was reduced to calculations.
And it seems that many forms of human intelligence can never be reduced to calculations, so computers will never become good at them.
0 Comments:
Post a Comment
<< Home