I think the reason automation isn't taking over as quickly as expected isn't because the efficiency gains are small, but because the capital investment required to automate is often large, and because many of the jobs that haven't yet been automated are complicated to solve in the general case. (This machine, for example, only handles boxing; it can't actually pick products off the shelves because that's a very general problem that's more difficult to solve.)

There are also numerous other economic effects to consider, such as the increased activity from capital spending on automation equipment and software, and (as you pointed out) the economic losses from higher short-term unemployment.

My initial question was only made in reply to a reductionist and intellectually lazy statement implying that the company owner is the only one who benefits from automation. I asked that question in an attempt to get the other commenter to think more carefully about their position (the answer isn't "nowhere" by the way, as the very figures you yourself provided indicate that 70% of it goes directly back into the economy). Unfortunately it doesn't look like they ever replied, but nonetheless I think I've made my point.