But automation is continuing at a rapid pace. Currently the designs are such that complex machines are often replaced rather than being repaired. (When is the last time you heard of a disk drive or a keyboard being repaired? Refridgerator?) And that's NOW, not when those currently in grade schoold have graduated from college/vocational school.
FWIW, I would have no idea what to recommend people learn...beyond certain basics: logic, critical thinking, how to evaluate reliability, etc.
A few decades ago I recommended that one should "be a garbage man", on the grounds that this would probably not be quickly automated. Well, automation hasn't advanced quite as quickly as I expected, but garbageman has remained a viable career path, if not a particularly attractive one. Part of the reason for this recommendation was that there aren't many openings in top management, and they are the ones who decide what jobs won't be automated, so that's the last job that will go. Today, though, I'm not so sure. Automated trucks are on the horizon, and that will eliminate huge swaths of jobs. And garbage collection has already been redesigned to increase the automation. (I'm not real inspired by the efficiency of the automation, but it has reduced the number of workers/truck and, perhaps, increased the speed of collection.) Further redesign is clearly needed, however...and by the time that happens, it's quite likely that the truct will drive itself.
Supermarket checkout clerk? RFID tags are already changing that. Currently there's no requirement that the cashier be more than very minimally literate. Self-checkout is spreading. How far will it spread?
Have you heard about the automated paralegal? It does searches through legal cases for useful references. This is work that used to be delegated to the entry lawyers of a firm. I don't know how widespread it is, but it's clearly something that is amenable to improvement.
Etc. I can't predict where automation will strike next, or how rapid will be its proliferation. But predicting what will be needed 20 years from now seems more than a bit risky. I might venture 5 years. At 10 years I'd be likely to overestimate the changes. At 20, however, I'd be likely to grossly underestimate the changes. That's enough time for something to be expected to come out of left field that will totally change things in unexpected ways. And education is supposed to prepare one for the long term.