Are Machine Learning Search Algorithms To Blame For Stereotypes?

by Laurie Sullivan @lauriesullivan, August 10, 2016

Do machine-learning algorithms processing search engine queries bring on prejudice, discrimination and stereotyping in query results? Search results have been known to highlight these negative attributes in the past. Now researchers at Brazil’s Universidade Federal de Minas Gerais suggest it could be true when it comes to female physical attractiveness in images available across the Web.

The paper submitted to the International Conference on Social Informatics scheduled for publication analyzes how Google and Bing represent female beauty in their image search results, particularly when it comes to different age and racial groups.

Virgilio Almeida, Wagner Meira Jr., and Camila Souza Araujo scraped the top 50 images for “beautiful woman” and “ugly woman” across dozens of international versions of Google and Bing. They then passed the more than 2,000 images through a program, which estimates subject age, race and gender with an estimated 90% accuracy.

For nearly every country analyzed, white women appear more in the “beautiful” results, and black and Asian women appear in the “ugly” ones, per The Washington Post, which initially pointed to the study.

Blackness is considered less attractive in 86% of the countries surveyed on Google, including Nigeria, Angola and Brazil, where most people have black or brown skin.

In some countries such as Japan and Malaysia, beauty is associated almost exclusively with extreme youth. Queries for “beautiful woman” return ladies not much older than 23.

In the United States, searches for “beautiful” women return pictures that are 80% white, mostly between the ages of 19 and 28. Searches for “ugly” women return images of those about 60% white and 20% black between the ages of 30 to 50.

Researchers admit they are not sure of the reason for the bias, but conclude that they may stem from a combination of available stock photos and characteristics of the indexing and ranking algorithms of the search engines.

The search engines may learn from the stock photos and reflect prejudices and bias of the real world that transferred from the physical world to the online world. Given the importance of search engines as source of information, the researchers suggest that they analyze the problems caused by the presence of negative stereotypes and find algorithmic ways to minimize the problem, according to the study. 

MediaPost.com: Search Marketing Daily

(26)