The Japanese also believed that useful generalized robots were coming int he 1990's and spent billions researching it. They eventually had to admit to themselves that the hardware/software just wasn't there and wouldn't be for quite awhile.
If you look over the last 15 years there has been an explosion in AI. One small example is 15 years ago airlines did the maximization of seat use on their flights with humans, and it wasn't sure how long it would take computers to be able to do it. Today computers do it all, booking the seats, setting pricing, scheduling their flights to maximize revenues. Its almost taken for granted that a computer would do those things now.
Also realize the commercial incentive is going to be insane. Right now there is no point trying to develop an AI robot. Because your average computer still does about 100,000 times less FLOPS than a human brain. That is a big step up from 20 million times less FLOPS a decade ago, but still something that a human brain would recognize in 1 second, would take the robot over 24 hours to figure out!!
However in the 2030's when a computer might figure the same question out in 1 second, its a different picture. At that point the googles, microsofts, facebooks and friends of 2030's are going to be in a race to bring out useful AI, and updating at a furious pace. No doubt there will be many hard problems.. but they will be able to put thousands of super genius programmers and philosophers and working on them.
The problem with the people who are betting against technology and saying its impossible.. is they are betting that of thousands of teams of geniuses around the world, some with near unlimited budgets, none of them will make any breakthroughs.
Its the same problem with the peak oil crowd. They are always betting that mankind will never be able to come up with new ingenius ways of extracting the oil. All it takes is one engineer somewhere to make a breakthrough and their theory is shattered.
No comments:
Post a Comment