What's being omitted in most cases are (1) the increase in productivity allowed by things like robots at warehouses and (2) the new markets enabled by new technologies.
The first can be estimated in various ways (revenue per worker, costs per worker, packages shipped per worker, etc) but the second is not estimable. The example I like to use is the advent of the automobile. I'm sure there were a lot of horse-and-buggy drivers or train operators that were decrying the disastrous unemployment it would cause, but the automobile industry now employs more people than that industry ever did, not to mention the new industries it has enabled such as long-haul trucking, taxis, private postal couriers, ride-sharing, logistics, etc. On top of that the automobile industry enables new productivity gains in hundreds of not-directly-related industries such as sales, food service, plumbing, medicine, and so forth.
The average American owns more than one vehicle. They're an affordable, quick way to do things like work more than a couple miles from your home, haul goods, or take a vacation. If the horse-and-buggy operators had got their way we'd have deprived ourselves of many of the luxuries we enjoy today.
Agreed. Though there is something to be said for skilled and unskilled labor.
Where the advent of the automobile brought about a labor market for machine operators (taxi drivers, combine driver, etc.), what happens when the machines automate themselves.
Or the bar to acquire the technical skills of an operator (ML engineer) is so high and move too fast coupled with institutions reacting far too slow to adopt policy aiding the unemployed in high skill acquisition.
The bar to acquire the technical skills of an ML engineer could, to continue the example, be likened to the bar to acquire the skills of an automotive engineer. It's a technically skilled profession. But the industries it empowered to arise (drivers) have much lower bars to entry. The point is that ML will create jobs not just in the production of ML but primarily in the application of ML. We cannot anticipate what that looks like, but it may be as simple as a clerk who speaks to a Star Trek-like computer and says "Computer, extrapolate what this person would look like in ten years and display."
The more we pursue new technologies boldly without fear of the consequences the more power we give to the common man to get things done that would've previously been impossible to him.
The first can be estimated in various ways (revenue per worker, costs per worker, packages shipped per worker, etc) but the second is not estimable. The example I like to use is the advent of the automobile. I'm sure there were a lot of horse-and-buggy drivers or train operators that were decrying the disastrous unemployment it would cause, but the automobile industry now employs more people than that industry ever did, not to mention the new industries it has enabled such as long-haul trucking, taxis, private postal couriers, ride-sharing, logistics, etc. On top of that the automobile industry enables new productivity gains in hundreds of not-directly-related industries such as sales, food service, plumbing, medicine, and so forth.
The average American owns more than one vehicle. They're an affordable, quick way to do things like work more than a couple miles from your home, haul goods, or take a vacation. If the horse-and-buggy operators had got their way we'd have deprived ourselves of many of the luxuries we enjoy today.