Research Discussion Assignment 3

Research Question

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.

Discussion

Probably the most prominent domain where Recommender systems reinforce bias is in the dating applications like Match.com where users fill out a profile and an algorithm identifies the “perfect” match based on the user’s preferences.

Users approach dating applications to find their ideal mate or dream date which is based solely on their biases, race, age, weight, height, wealth, etc. In order to combat reinforcing these biases and be a bias-free, recommendation system, the apps should eliminate certain choices for their members.

An article by THOMAS MCMULLAN on Wired.com looked into this: “A big motivation in the field of algorithmic fairness is to address biases that arise in particular societies,” says Matt Kusner, an associate professor of computer science at the University of Oxford. “One way to frame this question is: when is an automated system going to be biased because of the biases present in society?”

Summary

Algorithm designers need to be cognizant of their own biases and prejudices when building out Recommender systems. Completely un-biased Recommendation systems need to develop a better understanding of what their users would truly value, not what their biases are.

References

Are the algorithms that power dating apps racially biased?