Are search engines gender biased?
Google image search uses PageRank (despite Google’s plan to phase out of public page rankings over the next few years) and web mines to determine content that you would want to see most. Keywords, timing, the importance of a page and origin of the search are all important in determining what will appear on screen. But do these algorithms do more than just gather and display data? What if they are telling us how to think?
In a case with Google Image Search, when “nurse” is searched, only mainly female nurses are represented. However, when “doctor” is searched, the opposite occurs – mostly men populate the page. Google image search does not use any human method of detection to respond to a query, and the need to present images to a specific audience is not a priority. And yet, the “nurse” and “doctor” searches show blatant stereotyping.
In a similar case, computer Scientists are Carnegie Mellon University found that men were more likely to be shown higher paying job advertisements than women. To understand these result, researchers created a tool called Ad Fisher which tracked the relationship between a user’s behavior and the personalized advertisements that appear to them. They used a series of fake, newly-created accounts to closely track the data with limited variables. All of the profiles and search histories were created and formulated the same, within the one distinction of gender: some were male and some were female. Though advertisers are allowed to target the audience that they want, the results of the study would imply that advertisers are requesting and Google is approving that only men are displayed high-paying jobs.
In a final case, eyebrows were raised at a problem with LinkedIn’s search engine. When typing in “Stephanie Williams,” the site prompted whether the user meant to say “Stephen Williams.” And Stephanie is not the only case of this gender “correction.” LinkedIn claims that its search engine works by analyzing tendencies of past searches and generating suggestions from those searches.
The question remains if it is indeed the search algorithm’s fault or just mere coincidence. No algorithm is perfect, and the potential to code gender bias into the internet is quite likely, and is seemingly a rising issue. Search Engine companies like Google should work on developing bias detection methods that get rid of any human bias in searches or advertisements.
Sources:
http://www.huffingtonpost.com/2015/04/10/google-image-gender-bias_n_7036414.html
http://www.seattletimes.com/business/microsoft/how-linkedins-search-engine-may-reflect-a-bias/
https://hbr.org/2016/10/new-evidence-shows-search-engines-reinforce-social-stereotypes