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.
With the enormous amount of big data, companies are eaguer to throw at it any algorithms for data nalysis, prediction and recommendations to derive insight in the name of profit. To accomplish this, companies have to be bias toward algorithms to reflect customer segmentation or clustering. Thereby sometimes targetting a race, culture or anything to do with human factor.
Recommender Systems can reinforce human bias because the largest subgroup of users will dominate overall statistics; if other subgroups have different needs, their satisfaction will carry less weight in the final analysis. This can result in an incomplete picture of the performance of the system and and obscure the need to identify how to better serve specific demographic groups.
Here are examples of few recommender systems that are bias:
There are many ways that recommender selection could reinforce human bias. Consider few examples:
Recommender systems are a tool, and therefore they can and will be used in both ethical and non-ethical fashions, for good and bad purposes, fairly and unfairly.
What is even more important is the long-term effect of recommender system bias. The long term effect will lead to polarized groups of users.