I wonder if they will purposely train/instruct chatbot help agents to be a bit less helpful, to mimic the friction and frustration that phone trees and such cre

I wonder if they will purposely train/instruct chatbot help agents to be a bit less helpful, to mimic the friction and frustration that phone trees and such create. Or perhaps they'll train them to be cheerfully helpful, but to just dump all the feedback in a virtual circular file. If so, would the chatbot admit this is what happens if the user asked them?