As more systems and sectors are driven by predictive analytics, there is increasing awareness of the possibility and pitfalls of algorithmic discrimination. In what ways do you think Recommender Systems reinforce human bias? Reflecting on the techniques we have covered, do you think recommender systems reinforce or help to prevent unethical targeting or customer segmentation? Please provide one or more examples to support your arguments.
Recommender System reinforces human bias because its prediction is based on the data. If the data is biased towards or against an item/person or the output of the algorithm will also be biased. Evan Estola pointed out in his presentation that the bias could have been introduced by the recommender system’s designer or the company that commissioned it. I would say that Recommender Systems reinforces what the data suggests or what it is told to force.
A Recommender System built using user-based collaborative filtering may reinforce customer segmentation. Customer segmentation may not necessarily be a bad thing e.g. a video streaming service may restrict or recommend items appropriate for a viewer that’s seven years old. Item-based collaborative filtering can also reinforce customer segmentation. e.g. an item-based recommendation system would not recommend a multi-million dollar home to a family that earns a low income.