Job A costs 8+ human hours a day, and Job B costs 8+ human hours a day for something that doesn't really deserve a human at all; simply some humans are so desparate that they need to do it.
The market signal is intentionally changed from "automate A preferably to B" to "it will make financial sense to automate both A and B, do it".
Yes, I think that is the likely outcome if availability of capital is not the constraining factor. And my read of the U.S. economy is that availability of capital is not the constraining factor currently: lots of companies are sitting on huge piles of cash that they don't know what to do with. So they'd happily automate any A or B if it made financial sense to do so, i.e., if ratio of labor cost to robot cost favored the robots.
Why not automate 'em a few years earlier, and get a jump on both deploying new tech and moving people into something better?
Because it would be more efficient to invest in automating other processes?
Suppose you have two processes - one process (call it A) costs a human $12/hour, another (B) costs a human $6/hour. You only have enough money to invest in automating one. It's obviously more efficient to automate A.
If the minimum wage is raised to $12/hour, then the market signal to automate A preferably to B is lost.