Yes, certain things COULD be automated... at considerable expense to student achievement. One big thing they have found - remove the personal feedback and connection to students --> lose the motivation of students. If a teacher (the same teacher) isn't interacting with a student consistently at nearly every step of the learning process, the feedback doesn't stick and the student loses motivation.
It would be interesting to looking to the basic research behind the feedback-achievement connection and stereotype threat to start.
Hope that helps you address some of the problems with the automate/delegate solutions so often thrown at teachers.
I seriously doubt that letting teachers automatically grade arithmetic tests will hurt student achievement. The fact is that many teachers do that sort of thing at home in what should be considered overtime hours. I would like to hear how automatic grading causes student achievement to suffer.
Likewise, I'm skeptical that it should be solely educators' responsibility to make sure chairs are not being thrown out windows. Letting teachers focus on educating and not babysitting seems like a good thing.
1. Routine things that could be orders of magnitude more efficient (or even fully automated) given enough resources. In most cases, the resources needed would be fairly modest compared to the aggregate amount of effort teachers everywhere spend on them. Writing and grading elementary-level math tests, for example, shouldn't take any time at all given the right software.
2. Routine things that couldn't be automated well but could easily be done by some sort of entry-level assistant. Babysitting and discipline tasks don't require college degrees.
It strikes me that the economics of education are structured in a way that there is marginal impetus to improve efficiencies in the day-to-day work of teachers.