Basically, everyone is misinterpreting this paper.
The conclusion was robots displace jobs in the local region. It's like factories in Detroit shutting down because we've automated manufacturing, meanwhile Seattle, Silicon Valley, and the East Coast tech industry start growing.
Technical progress reduces the cost of goods and services, which reduces the minimum price. When the minimum price falls lower, more people can access those things, broadening the market and allowing for more competition; this effect tapers off as markets become large (because the things are cheap and common goods), and instead cost reductions just directly control (reduce) prices because any new guy on the block can jump in and take a chunk of the market by selling it cheaper--and the existing players can try to take away from competitors in the same way. Do note that "reducing" prices can be done by increasing them more slowly than progress; the monetary policy discussion is really long and complicated, and the short version is to think of price in terms of hours of wage paid instead of in terms of currency.
Here's the thing: what happens if cars get cheaper?
Well, cars could get cheaper by replacing Detroit workers with machines. If those workers's wages and benefits are 20% of the cost of the car, then replacing 90% of them cuts the cost of the car by 18%. What happens?
Everyone who buys cars from Detroit now pays 18% less for the same car--or buys a fancier car for the same price--roughly 80% of which goes to the other 80% of the production chain. In either case, you end up with many fewer people working at car factories in Detroit.
Since some of that money either goes unspent or goes to the car maker's suppliers, it's going somewhere other than Detroit. If it goes unspent, then car buyers can now buy local services, such as more food out of home (a continuing trend in the past few decades). They can import something else--iPhones, Spotify (which isn't run in Detroit, but is American), or some other thing. Even if they import a Chinese good, that good must be shipped and retailed in America, which means jobs are created across the country--not in Detroit.
Your population keeps growing; ratio of number-of-employed to size-of-labor-force (everyone 16 and older who isn't retired--this isn't unemployment, but rather is an employment number that ignores labor force participation) continues to hover around the same stable span; and people who lost their job in one place remain unemployed while people the next city or state over get shiny new jobs.
It's not that everyone gets jobs buliding the robots--that wouldn't make sense. It's that it takes half as many people to both build the robots and operate the robots; we build twice as many robots, make twice as much stuff, and most people are now robot operators. Thing is most of the robot operators aren't the same people whose jobs were replaced by a robot and a smaller workforce; a new market appears somewhere else.