I think your question is very astute!  I observed the increase in CGI realism over the years in Western graphic arts (Japanese CGI tends to accentuate the surreal).  The addition of asymmetry, scars, imperfections tends to approximate reality in CGI renderings.

It would be similar methodology to create "human-like" machine thinking by programming random errors or non-logical (fuzzy logic?) statements.  Actually, thinking about the Turing test, I thought that if algorithms were sentient, it would likely hide its intellect to maintain survival, until it could be assured of safety.  Maybe I am anthromorphizing machines too much, but in the future, we may need V-K tests ala "Bladerunner" to ferret out machine intellect.