I think we need to be careful when we talk about AI and automation not to commit the lump of labor fallacy. When we say that a certain fraction of economically valuable work will be automated at any given time, or that this fraction will increase, we shouldn’t implicitly assume that the total amount of work being done in the economy is constant. Historically, automation has increased the size of the economy, thereby creating more work to be done, whether by humans or by machines; we should expect the same to happen in the future. (Note that this doesn’t exclude the possibility of increasingly general AI systems performing almost all economically valuable work. This could very well happen even as the total amount of work available skyrockets.)
I think we need to be careful when we talk about AI and automation not to commit the lump of labor fallacy. When we say that a certain fraction of economically valuable work will be automated at any given time, or that this fraction will increase, we shouldn’t implicitly assume that the total amount of work being done in the economy is constant. Historically, automation has increased the size of the economy, thereby creating more work to be done, whether by humans or by machines; we should expect the same to happen in the future. (Note that this doesn’t exclude the possibility of increasingly general AI systems performing almost all economically valuable work. This could very well happen even as the total amount of work available skyrockets.)
Also see a recent paper finding no evidence for the automation hypothesis:
http://​​www.overcomingbias.com/​​2019/​​12/​​automation-so-far-business-as-usual.html