metaphors of machine learning

in my growing research into machine learning, I am interested in the metaphors (of more higher level concepts) applied by researchers to describe the constructed systems, complexity and ‘behaviour’ of machine learning models. These no doubt come from lived experience/‘real (human) world’ insight and I wonder if placed side by side with pedagogical practice broadly and historically if there is some is something interesting here to glean

particularly thinking of the relationship with power within institutional structure of learning