There's an economic principle I ran across a while ago which I find absolutely fascinating -- in fact it's a decent part of why I wanted to start writing about this type of thing in the first place. Unfortunately I can't remember its name (the Internet has failed me), but I can describe it.
Part of it is based on the idea that I've discussed in a lot more depth elsewhere of jobs shifting their use of automation over time, relative to each other. With that as a given, it's obvious that taken in isolation, certain jobs would tend to command higher salaries over time, while those left without the benefit of automation over a long period would tend to drop. But human workers don't act in isolation. Instead, there's another driving principle in the economy, which is that jobs, even in different fields, which take a roughly equal amount of skill to perform must provide roughly equal compensation to people across all those fields.
The unnamed principle deals with the result of the obvious conflict between those two forces. On the one hand, fields that don't benefit from automation will tend to drop in purchasing power. On the other hand those jobs aren't getting any easier to perform, so you have to keep compensation in those fields roughly in step with that in automation-friendly fields or face the consequences of deteriorating workforce quality. Not a pretty choice, and the further implications are even worse.
When I first read this, the example was about teaching. Everyone has heard the news stories pointing out the fact that we're facing rising education costs and decreasing quality. Most of them act as if this is unexpected, which is completely wrong. Because teaching hasn't noticeably benefited from automation in centuries. What little gain they have found is probably from their heavy dependence on information transfer processes. Cheap printing and Internet access lower supply costs, and once we eventually make the leap to digital books where the price actually reflects the reduced cost to the publisher, they'll see a little more. But the core requirements haven't changed in a long, long time. You still need a teacher to lead a class and a room to put them in. Growing class sizes give a little efficiency increase, but at a demonstrable cost in terms of quality.
So we cut costs where we have to -- salaries, and facilities. Only then you get a section of the workforce, especially those who are young and can retrain easily, who decide they'd rather do something else than deal with forty screaming children for a near minimum wage salary. Perfectly logical for them, but a drain on the top talent in the field. Seeing your work as a calling can only make up for so much salary differential before you start losing most of the good people, and quality suffers even more.
If our goal were to keep the quality of education exactly the same we would see even more inflated costs than we do today. Almost paradoxically, every time we increase automation in other fields we end up having to spend a larger percentage of our income on the things that didn't get improved. In the short term this might be sustainable. After all, with the increases in efficiency elsewhere we're increasing our overall standard of living, and maybe we end up making enough extra that it more than covers the increases in these fields.
But it isn't sustainable indefinitely. As more and more fields of human effort become "high performing" thanks to automation, the salary for skilled workers across society will keep going up. At some point it will become high enough that we simply can't afford to pay the salary necessary to keep those people in fields like education. Arguably we've already reached that point, given the current near-crisis we're facing in education. Even worse, it seems reasonable that this may cause a feedback loop which magnifies the effect, since lowering education quality will result in fewer skilled workers, raising demand and salaries and further reducing the number who go into education.
There are several ways to fix this, but all of them require reinventing the job. The key is to find a way to make a single skilled worker more effective in the field -- magnify their output, so to speak. You could do this entirely with automated systems, taking a couple of experts and having them put together computer tutorials that children could follow without human intervention -- but that's very impersonal, and arguably the human interaction component is just as important for children as the lesson. You could come up with some kind of scheme where unskilled workers provide the low-level supervision and are overseen by a smaller number of skilled workers who can direct them. This gets back the human element, although removing children from close proximity to the subject expert doesn't sound all that great.
Some kind of hybrid model is most likely to be successful, taking elements of an automated system built by experts that can handle the low-level information delivery and tweaks to the speed and direction of the learning process, unskilled workers in relatively high concentrations for the interaction element, and more specialized experts in lower quantity but close enough that they can direct and handle situations that the less skilled workers can't. Whatever the specifics, it would have to be a pretty major shift from the model we're comfortable with, simply because you have to reduce the number of skilled workers in the system.
That's just education, but this general idea applies to any worker in a position that's significantly harder to automate than others which attract workers of the same skill level. Certain types of manufacturing, such as Boeing, come to mind -- they benefit from machinery and computer design systems, but the actual assembly of each plane is a custom job requiring high levels of craftsmanship. Worse, medicine seems like it might be a candidate for the same effect. Everyone right now is talking about waste, and insurance inflation, but I'd be very surprised if there weren't at least some effect based on the fact that doctors and surgeons can't attend all that many more people today than they could fifty years ago. On the bright side, though, there are some tremendous opportunities out there for people who are good at creating automation to jump in and catch these fields up with the rest of the industrial world.