One of the most dangerous things you can do in an aircraft is to deprive your colleague of their in-the-moment competence or confidence. Using modes whose functioning they do not, or may not, understand, or whose precise vagaries in conditional states they do not, or may not, understand, is a great way to achieve that. I don’t really care about the complexities of those modes in the uses proposed, because a pilot who can fly doesn’t need them. I probably wouldn’t be interested in discussing them other than as an academic exercise in considering their design and fitness for purpose, it would be like making an extended study of the plumber’s wrench as a tool for screwing wood screws in - you can do it, but it’s not sensible. There used to be teaching, originating I believe in the NBAA, that in marginal situations, reversion to the next lower level of automation was well-advised, right back to raw data. That seems to have been forgotten.