"Actually I'm not sure even THIS approach works since even tech applied for pure aesthetics can do harm (imagine an app that makes cute cartoon selfies, but because of a limited training set breaks on people of color). But if you're gonna use ML at all u have to outline its limits"
You mean like has already happened? Along with a bunch of other bad results for people of color (in prticular but others as well) in the results of algorithms based on biased and/or incomplete data?
(no subject)
You mean like has already happened? Along with a bunch of other bad results for people of color (in prticular but others as well) in the results of algorithms based on biased and/or incomplete data?