That was a solved problem in 1998.
1. Is the rate of creating new possibilities equal to or greater than the rate of killing jobs?
If yes:
2. Is the difficulty of retraining our workforce today equal to or less than the difficulty of retraining our workforce in the past?
If yes:
3. Is the rate of retraining our workforce equal to or greater than the rate of creating new possibilities?
If yes:
Then yes, we don't have a problem.
We are now entering a period in which middle and high paying jobs being replaced by machines. High-frequency trading replaced a whole profession in the financial sectors. There are now machines that are capable to analyze an x-ray image and determine the cause of the problem, leaving the doctor with "just" the job of delivering the news.
So, yes, we do have a problem. I'm not a Neo Luddite. The purpose of this is not to destroy the Internet and all the machines, but I do think that we need to have an ongoing debate about the social structures that we are dissolving as much as about those that we want to build up.
But it also created one. Someone has to both conceive and write each algo. This looks like the old profession, except with different people because Excel is being replaced by C++.
http://www.vanityfair.com/business/2013/09/michael-lewis-gol...
The algos can be created by fewer people. Traders are being replaced by the algos, and fewer people are required to create the algos than do the original trading. So your point does not have a 1-to-1 job creation-destruction ratio.
Yes, but those efficiencies open new possibilities. Transistors killed abaci and slide rules, and made skills in their use suddenly redundant.
Fast forward, and I'm reading about people employed to make realtime peer to peer video between any two people in the world.
Yes, truly self improving AI would change everything, and probably spell the end of the traditional economy, but I think we're a long while off yet.