tl;dr:Bias exist in both algorithms and people. Bias caused by an algorithm is symptomatic of a poor data set. If we create a framework to improve data sets - to account for everyone - an algorithm can be trained properly to negate bias. Retraining people is harder.