There's an important difference between the 19th century and today. We are nearing the point when computers will be mentally equal, if not superior, to humans. They are already physically superior in many regards. Even before that point, there are tons of jobs that can be done by machines with minimal intelligence. If a machine can do everything that many humans can, but for less money, those humans will be unemployable. If machines are stronger than you and smarter than you, then there's nothing for you to shift to that they aren't already doing.
I really don't see the "We are nearing the point when computers will be mentally equal, if not superior, to humans." AI hasn't produced any results that have even got within range of human intellect. I am not denying that a lot of factory jobs can be made robotic, but I am saying a lot of skilled labor still cannot be done by robots (e.g. plumber).
In the near term, only unskilled labor is really at risk. Even so, plumbing has no potential for growth in developed countries above the rate of population growth. It's something you only care about when it's broken. The key to avoiding huge disruptions in the face of advancing technology is having new jobs for displaced workers. I don't see it.
There are plenty of jobs that will be automated in the next decade. What will those workers do?
" It's something you only care about when it's broken." - you do realize that plumbers are needed for new construction and renovation? Actually, given the needed improvements in infrastructure and new building projects, a lot of skilled labor is needed. I am really not seeing robot electricians, plumbers, etc. for many decades to come.
I am really unclear what you think unskilled labor is.
I really don't think the labor stats match your belief in regard to the skilled labor, and I the unskilled positions your are mentioning are being limited more by raises in minimum wage / economic conditions versus automation / demand.
We are nearing the point when computers will be mentally equal, if not superior, to humans.
Not even remotely close. In fact, if we ever want to reach a point where computers are as intelligent as humans are, we are likely going to have to start from the ground up with the way we build processors.
Our computers rely on precision, our brains rely on lack thereof. Our brains were "built" from the beginning to match fuzzy patterns. I look at a door and I know immediately that it is a door, even if I can only see part of it and have never seen one that looks anything like it before. Computers are terrible at this. We're getting sortof-kindof-maybe close by applying bayesian filters to images, but you're talking about rooms full of specialized hardware that cannot even compete with the brain of a mouse.
Additionally, that which makes us intelligent (what a delightfully indescribable word that is) is partially our creativity; which is something that computers are absolutely incapable of in their current form.
Sorry, but it is probably going to be a while before engineers, or architects, or programmers, or babysitters are replaced by any sort of "machine intelligence".
You're wrong about this -- neuroscientists know quite a bit about the mechanisms that make the brain work. You don't have he same knowledge because it hasn't been popularized and disseminated yet.
> We are nearing the point when computers will be mentally equal, if not superior, to humans.
No we aren't. We are no closer now than we were 50 years ago. The hardware is faster, but the software has not changed. (And hardware is not getting faster anymore either.)
Seriously? We have cars that drive themselves. The realm of tasks that only humans can do shrinks every year. We're not close to computers equaling humans, but we are nearing it.
I agree. The interesting question is not when robots will be equal or even superior to humans. The interesting question is when they will be advanced enough to take over a significant portion of labor for cheaper than even a chinese slave worker could.
The physical, mechanics part seems to be mostly sorted out. Robots can walk, run, climb stairs, avoid obstacles, get up when they fall, lift heavy things or operate at incredible precision. It won't be long until they exceed us humans in most physical regards (with a few notable exceptions like smelling or touch-sensitive skin, but imho even that is only a matter of time).
In my opinion the really interesting research is now going into the stuff you see in the video above. Making robots discover and interact intelligently with their environment. This is still a far cry from real intelligence but try to imagine the impact when robots become able to execute simple instructions like "Make me a coffee", "Drag the couch from A to B" or even "Repair the plumbing" without further handholding.
That will be the tipping point for a huge shift in society, akin to the invention of bookprinting or the internet, likely even bigger.
Looking at what we have today I would say that specific vision is just a matter of time. Whatever follows then is in the pudding.
Last week, I heard a talk from one of the engineers doing this. He claimed that the main thing holding back robotics is liability concerns. If someone builds a machine that causes widespread death and destruction, the usual cop-out is to blame the operator for losing control of it. But if a fully autonomous robot goes postal, the manufacturer can't pass the buck to anyone.
Perhaps this will be a real issue, or perhaps robot engineers are a bit too fond of Terminator and 2001. Fear of it is apparently causing real effects. It would be sad if the ultimate reason to employ humans is that it feels good to kick their arses when things go wrong.
Depends what you mean by mentally superior. Computers can add billions of numbers a second, work at a very high precision with reflexes measured in nanoseconds, do not have weaknesses like boredom and the need to socialize. If you're talking about creativity and navigating the real world, however, computers aren't there yet.
