I understand where you're coming from, but I think you already covered the counterargument to this; in the aggregate, humans may not be more effective reviewers than a purely computational system. There should always be some sort of recourse as a failsafe, but I don't think that spending more money and time on human customer service is a good idea for Google. I think it's something we'll get used to; I imagine there must have been a similar sort of discomfort when ATMs were introduced, but they're ubiquitous now.

This sort of automation is also very much in tune with Google's philosophy (and really, the history of information retrieval itself--it's gradually steered away towards using much human expert input in problem solving from what I remember) and may be the only viable option for some of Google's services given the kind of scalability they need.