Probably it is not happening now, but it should happen in a few years.
Historically technology has displaced humans, but there has always been something that is cheaper to retrain humans for than it is to build machines for. For one thing you can explain to humans what to do, and they could turn an incredible amount of computing power to figuring out, following, and filling out your instructions. Computers simply have not had anything close to a human brain.
However as the complexity of our machines goes up, and the costs of automation go down, eventually we'll come to a point where it is cheaper to design and build a machine for a new job than it is to train someone to do it. At that point humans will be displaced and new jobs won't show up for it.
By a back of the envelope estimate, we're perhaps 10-15 years away from having computers with the raw computational power of a human brain. Software will follow, and the successes of things like Watson suggest that it won't take that long for software to catch up enough to be economically useful.
Once it is possible to train a machine faster than a human, the cost of buying and running that machine puts a cap on what that human can expect to be paid. Given Moore's law, that cap will descend painfully quickly after it first becomes an issue.
Historically technology has displaced humans, but there has always been something that is cheaper to retrain humans for than it is to build machines for. For one thing you can explain to humans what to do, and they could turn an incredible amount of computing power to figuring out, following, and filling out your instructions. Computers simply have not had anything close to a human brain.
However as the complexity of our machines goes up, and the costs of automation go down, eventually we'll come to a point where it is cheaper to design and build a machine for a new job than it is to train someone to do it. At that point humans will be displaced and new jobs won't show up for it.
By a back of the envelope estimate, we're perhaps 10-15 years away from having computers with the raw computational power of a human brain. Software will follow, and the successes of things like Watson suggest that it won't take that long for software to catch up enough to be economically useful.
Once it is possible to train a machine faster than a human, the cost of buying and running that machine puts a cap on what that human can expect to be paid. Given Moore's law, that cap will descend painfully quickly after it first becomes an issue.