This is exactly the taboo I'm talking about. Police, health care, mining, teaching, and surprisingly many others aren't going overseas right now. A lot of jobs are disappearing, and a lot aren't. Why assume that e.g. the American economy won't have enough jobs in the near future? There's no data on HN in all these discussions, and no real trends to extrapolate from except in very specific sectors.
Mining is becoming heavily automated. The work of a dozen men even twenty years ago can now often be done by two with a good computer model.
We are pushing very hard for computer vision, which is the real barrier to practical robotics that could replace your electrician, plumber, nurse, miner, driver, etc. There is no sacred cow of labor besides what is mandated by the molasses slow state (because I would absolutely argue that learning systems like Khan's Academy, the availability of resources online, and the technological organizational potential of software solve all the criteria to substitute teachers with a security guard watching the kids while they listen to robo-instructor, solve the problems themselves, and ask the neural-net for help when they run into problems - and the kids are of course not all in the same room in an expensive upkeep building called a school). That kind of change though must also be culturally accepted since a lot of effort goes into separating public education from market forces.
Point is, the jobs that we could not send overseas have tremendous pressure on them as a result to be absolved entirely by AI, robotics, and software. It is why the pencil pusher desk job of 1980 - rows of typists at typewriters transcribing documents - went the way of the Dodo and did generate a ton of unemployment that we still have not seamlessly solved.
Pressure, yes, but I want numbers - what percentage of jobs have disappeared over, say, the past 20 years, and could increased demand realistically replace those jobs in other sectors? You can't simply say "manual labor is over, because... singularity." There's no singularity yet. Until then, you need data before you can really extrapolate from current trends. The easiest things to automate are already automated, and harder things are coming along slowly. It might be true that jobs disappear faster than prices fall from the automation, or prices might fall fast enough that most people can have jobs that don't pay much, and still get by.
Like I said, there are demonstrable mechanisms, and in-lab and even in-market examples of how these technologies are going to replace almost all these jobs. This is not science fiction. This is market adoption at this point. And we are not at the start of the process, we are in the middle of it.
None of this really answers the question. Of course technology eliminates jobs. This has been happening since before Ford was founded and before all those automotive jobs were originally created. See https://en.wikipedia.org/wiki/Luddite for instance. The question isn't whether specific jobs are being destroyed, but rather whether all unskilled jobs are being destroyed, and none of those links even begin to give any data on the subject.
By definition of unskilled labor you imply only work done by physical action.
If you really don't think you cannot replace the human arm with a machine, you haven't been paying attention for two centuries. How is that even a question? We could replace all unskilled work today, if all you are doing is removing mechanical components of labor.
Remember, unskilled labor is "work to be done without training or certification". Driving a truck requires a Class C license, which is a certification. Working at Mcdonalds takes no certification. You need to have state certifications to work as a carpenter, plumber, or electrician. Your janitor doesn't need anything. So when you ask "is there any evidence unskilled labor can be completely replaced" then I would ask what unskilled job is not being replaced right now, from self checkout to vending machines to roombas to combine harvesters.
Certification is a red herring; assembly line workers have skills that I lack. I'm really only interested in jobs of that could be filled by the people who used to work in the Rust Belt, whatever you actually want to call that type of employment.
Thought experiment: let's say a trend starts for everyone earning 6 figures in Silicon Valley to hire a butler and a cook. How many jobs would be created? How many have recently been destroyed? How long before robots are as good a substitute for human butlers as the robot in The Jetsons?
The fact you do not have significant pressure to have a butler or cook today is in large part due to automation making them effectively obsolete. Anyone hiring one is making the informed decision that they are cost ineffective at jobs your phone, roomba, and local tv dinner manufactuary or new-age restaurant where the foot is prepared by machine are more efficient at.
