Ads and Discrimination
Recently in one of my classes I came across this very interesting article about Discrimination in Online Ad Delivery by Latanya Sweeney. Her article described how when searching for more Black-sounding names, arrest records appear more often across multiple ad companies in Google. Considering how we have been learning about sponsored search, and how we analyze the different algorithms with advertisements (VCG, second-price auction, etc.), it’s very interesting to see how this is incorporated into real-life issues such as discrimination. Sweeney herself breaks down the algorithm: “An advertiser provides Google with search criteria, copies of possible ads to deliver, and a bid to pay if a reader clicks the delivered ad.” So in this case, we have multiple ad companies bidding for space given a formula (this formula being “firstname lastname” searches), and splitting the profit with Google when a reader clicks on the site.
Sweeney finds that across multiple companies, Black-sounding names are more likely to be associated with public records than white sounding names. This means that we need to evaluate how the algorithm works, and the real-life impacts in can have on people. Computers and algorithms merely follow the instructions they’re given. In this case, computers and algorithms are merely running a quick auction as it would any other content. However, we are the ones who build these algorithms and maintain it, we are the ones responsible for making changes so that we address and fix social issues like discrimination and racism in algorithms such as this one. This article presents how there should be an intersection with social issues and algorithms, that one cannot live without the other.