Interestingly, each time we have lost jobs, we have diversified into many more types of jobs that cannot be swept away with a single automation breakthrough. ~90% of workforce were once farmers and almost every single one of them lost their job when mechanization came. Now, with self-driving cars, we are worried about the ~2% of the workforce who drive for a living. That is already a big change from the past, and each iteration how shown that a smaller and smaller portion of the workforce will be affected by automation at any given point in time. That is not to say that everyone will go unscathed, but mass unemployment becomes less and less likely.
It was very easy to go from one job to the other. weather it needs to be required or not, many jobs today effectively require 20 years of education (12+ school, 4 undergrad, 2 masters, 2 years experience).
we talk about coal mining jobs going away as if they should have seen it coming, but tech can also be vulnerable. people forget what it was like to search for a tech job even years after the 2001 bubble burst.
also, the reason why people are worried about automation isn't the number of people effected by each iteration. it was widely accepted automation would destroy jobs and more than what was destroyed would come.
the 2001 and 2008 recession seems to indicate that is becoming less true. i would categorize those recessions as mass unemployment, thus mass unemployment is not becoming less and less likely.
When I started I thought it'd be fairly short. It wasn't, and much of the difference wasn't theoretical CS stuff—it was "stupid computer shit". Tools, the command line, daemons, implementation details for a dozen key technologies, debuggers, and on and on and on.
I think I'd underestimated how much all that time I'd spent futzing with DOS to make my games work, or playing with operating systems, or running a message board for my friends, and so on, had provided the foundation for my success as a developer. The hours weren't spent optimally, but they were spent in great quantity, and most people didn't grow up with that kind of running head-start into the industry. It's a giant hurdle to overcome.
http://thehill.com/blogs/congress-blog/healthcare/266610-sho...
(and of course it isn't necessary to immediately double the number of slots in order to soak up the extra MDs)
A great way to spend more money on education would be to target lowering high school graduation age to 17, with at least a year of training oriented education for students that weren't interested in an academic track and a faster academic track for students that did want it. The idea that everybody gets value out of Algebra II is pretty confused and at my Big 10 university, I often knew more history on days I hadn't done the reading than most of the people in my discussion class (that is, our "well prepared" high school students aren't coming out of high school well rounded anyway, lets drop the pretense).
That in no way suggests that there isn't a role for the hyper-educated / trained doctor, or for even more highly trained specialists. The point is just that not every situation requires them.
I'm tired of hearing this argument again and again. It amounts to saying this...
> I did a thought experiment and in that experiment is was theoretically possible to replace humans with AI in any job. Therefore we're all fired!
This is not far from saying in a Universe containing 100 billion galaxies there are probably aliens out there willing to work for less than the minimum wage, just dying to make us all redundant! It's also theoretically possible.
But just because something is theoretically possible, doesn't mean it will actually happen. Most jobs won't get replaced with AI because humans will remain the better economic option.
The automotive industry is an exception for economic not technology reasons. There are a small number of car makers making enormous revenue selling cars that are vastly under utilised. Grabbing a share of the revenue by disrupting that industry with self-driving cars makes economic sense; the massive investment required to develop (and maintain!) self driving cars.
Even more, there's no definite proof (yet) that self-driving cars will even "catch on" and replace traditional human driven cars at all. OK, for shipping / transport and what-not, the economic drivers are pretty clear. But considering that at least some personal driving is actually recreational and that there's no super-strong economic incentive to favor self-driving cars, it's hardly out of the realm of possibility that the self-driving cars will just be a passing fad, or will never amount to anything but a niche market.
Note that I'm not saying that things will necessarily play out this way. Just pointing out that it is a realistic alternative to the prevailing "self driving cars are the future" narrative.
Right now researchers are demonstrating AIs that can learn to play video games vastly better than humans. There's not a huge leap between that and learning to control a robot. Hell I'm watching a video series right now where a random programmer builds a self driving car in GTA.
A robot I can call about a plumbing problem, that can realize it needs to dig up a pipe in the front lawn, successfully dig up the pipe in the front lawn and fix a leak. Not to mention the rest of the things a plumber does. Now that is hard task and I can't see it being solved any time soon.
Any job that has difficult "IO" is a long way from automation in my mind.
The fact is on the human lifetime timescale, jobs go away (farming from 1900-1940; various factory jobs in the late 1980s, coal mining 1970s-2000s etc). People with decades in these jobs may not be able to move into equivalent (in renumeration & status) replacement jobs as their skills are likely not transferable.
Thus to the tractor manufacturer in her 50s it is irrelevant whether there is high demand for experienced web designers or not. Perhaps her kid can go into the job, but she very likely can't. This is part of the reason for the demand for low-skill (and low paying) jobs.
So either way we have to plan for rolling mass unemployment. The only question is the shape of the derivative.