#DecodeTheBias fighting algorithmic beauty biases

5 Oct 2021 --- Of the search results given for ‘beautiful skin’, ‘face’ and ‘woman’, 80% do not include women of color. US skin care brand Olay wants to change this, so Olay launched its #DecodetheBias campaign in September to advocate for change. The campaign was largely inspired by the work of Joy Buolamwini, founder of the Algorithmic Justice League (AJL), which was highlighted in the Netflix film Coded Bias.

Click to EnlargeThe campaign is inspired by the work of Joy Buolamwini. Source: #DecodeTheBias.“This campaign tackles two issues - coded bias as it manifests in the exclusionary representation in beauty imagery, and the need to create more equitable opportunities for young girls of color, especially in the fields of STEM (science, technology, engineering and mathematics),” highlights Buolamwini.

Released in 2020, Coded Bias explores the work of MIT Media Lab researcher Buolamwini, who discovered that facial recognition does not see dark-skinned faces accurately. Following the discovery, she began campaigning in the US for legislation against bias in algorithms that impact our online searches and decision-making processes. 

A poet of code who uses art and research to illuminate the social implications of artificial intelligence (AI), Buolamwini’s MIT thesis methodology uncovered large racial and gender bias in AI services, prompting her to found the AJL. “We want the world to remember that who codes matters, how we code matters, and that we can code a better future,” she says.

“AJL has learned how code, data and algorithms can be exclusionary,” says Stephanie Headley, Vice President of Olay, Procter & Gamble (P&G). For example, she notes: “We found that search terms like ‘beautiful face’, ‘beautiful skin’ and ‘beautiful woman’ do not feature enough women of color.” 

Misclassifying color

In Buolamwini’s research, “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification”, the team embarked on a study to understand racial and skin type bias found in AI systems. The researchers demonstrated an approach to evaluate the bias present in automated facial analysis algorithms and datasets that relate to phenotypic subgroups. Using the dermatologist-approved Fitzpatrick Skin Type classification system, the researchers characterized the gender and skin type distribution of two facial analysis benchmarks. 

Following their research, the team’s results showed that these datasets were overwhelmingly composed of lighter-skinned subjects. The researchers then introduced a new facial analysis dataset that is balanced by gender and skin type. "We evaluate three commercial gender classification systems using our dataset and show that darker-skinned females are the most misclassified group (with error rates of up to 34.7%),” the research study states. “The maximum error rate for lighter-skinned males is 0.8%,” the researchers add. 

“The substantial disparities in the accuracy of classifying darker females, lighter females, darker males, and lighter males in gender classification systems require urgent attention if commercial companies are to build genuinely fair, transparent and accountable facial analysis algorithms,” notes Buolamwini’s research.

Olay’s goal is to create a more inclusive representation of beauty online and take tangible actions towards that goal. In devising its ‘DecodetheBias’ campaign, Headley emphasizes: “We wanted to prompt consumers and the beauty industry with the question, ‘Where are the women of color?’.” 

Following AJL’s research and discoveries that backed Olay’s campaign, Headley details how search algorithms and beauty filters on social media apps are biased towards Eurocentric beauty standards:

• “Algorithms act as digital gatekeepers deciding who gets hired or who gets access to healthcare, often giving undue preferential treatment. Another effect is that algorithms reinforce an exclusionary standard of beauty,” states Headley.
• “Search engines are optimized to present what is most popular. When a system is optimized for popularity, representations of dominant groups are generally what makes it to the top of the stack. This is why if you searched for ‘beautiful face’ or ‘beautiful skin’ you would see search results that are lacking women of color.”
• “Social media apps and filters contain algorithms that are claiming to enhance beauty - by lightening skin or slimming a nose - based on a European standard of beauty, leaving many women of color excluded.”
• “Both of these examples are showing girls that what is considered ‘beautiful’ may not be representative of them. But it doesn’t have to be this way. We have the power to approach the design of algorithmic systems differently to ensure many perspectives and many types of beauty are included.” 

Buolamwini adds: “I remember being a little girl and being teased for my dark skin and being told I was not the standard of beauty, so to have the opportunity to be a face that young girls can relate to is incredible.”

#DecodetheBias campaign goals

Through its #DecodetheBias campaign, Olay hopes to get closer to its goal of doubling the number of women in STEM positions, as well as trebling the number of women of color in STEM by 2030. “This campaign, in partnership with Joy Buolamwini of the Algorithmic Justice League, highlights one example of how a lack of diversity in STEM and computer science fields can reinforce the norms,” states Headley. “By diversifying who codes, we are creating a new future that is more digitally representative of all.” 

Click to EnlargeAn aim of the campaign is to send over 1,000 girls of color to a coding camp. Source: Olay.She adds: “We also hope to help take steps towards a more inclusive digital world. To do this, Olay has partnered with Black Girls Code to send girls of color to its 2022 summer camp.” 

Emphasizing how “anyone can help us”, Headley shares that Olay is encouraging users to use the hashtag #DecodetheBias on Instagram and Twitter to help send up to 1,200 girls to the summer program because, Headley highlights: “As more women and women of color become coders, we will see more inclusive code.”

Kimberly Bryant, Founder and CEO of Black Girls Code, says: “Code touches every aspect of our lives, and showcasing its application in the beauty industry is great because so many girls are passionate about beauty. By sending more than 1,000 girls to our intensive camp, we hope to build foundational skills and foster their longer-term career interest.”

Headley sums up: “Olay remains committed to equality and inclusion and champions a more diverse and inclusive standard of beauty in which there is not one, but many, representations of what is ‘beautiful’.”

By Natasha Spencer-Jolliffe, BPC Insights Senior Journalist