Beauty is in the eye of the beholder. However, it seems Google Image’s search engines have their own ideas of what is “beautiful”. “This is something we should care about” urges Safiya Umoja Noble, a professor of Information Studies and African American Studies, regarding the racial biases evident in some of our most trusted search engines. In a TED talk at the University of Illinois, she explains; “Search engines are an amazing tool for us, but what we don’t see is the ethical biases that are inherently built into them”. Umoja Noble, who is African American, recounts an anecdote based on her friend’s experience with Google Image search: “When she did a search for ‘beauty’: this is what came up”. The screen fills with images of young, white women. Looking concerned, she explains that this is reinforcing the social biases of society: “they get replicated in our search engines”. How are we, as a society, to combat racial inequality if the tools that we depend on the most, inherently reinforce these barriers? Is this an algorithmic issue, or simply a reflection of the views of Google as a company?
Unsurprisingly, this is not the first time Google search algorithms have come under fire. In a piece by Mind Matters, an ex-Senior Google Search Engineer, Gregory Coppola, contacted a watchdog group (project veritas) to warn users about political biases in Google’s search engine results. He states: “No private company should have either the right or power to manipulate large populations without their knowledge”. I reached out to Coppola and asked him what his approach would be to identifying biases in a search engine such as Google. In an exchange on the issue, his response was to the point – all of it is biased.
In his blog, Coppola proposes “Coppola’s Law” which states “the social bias in a software product is the social bias of the organisation that produced it”. And Coppola isn’t the only one with concerns. Gabriel Weinberg, CEO of rival Duck Duck Go and critic of Google search engines: “This filtering and censoring of search engines and news results in putting users in a bubble of information that mirrors and exacerbates ideological divides” quoted in The Observer.
Google search engines have a lot of critics, but then if they are so bad, why are they so popular? According to internetlivestatistics.com Google processes 1.2 trillion searches per year worldwide – that’s 1.2 trillion opportunities for reinforcing (or addressing) social divides worldwide. Surely something we are so dependent on cannot be as big a social evil as it seems? So says Google’s CEO Sundar Pichai. According to CNBC In 2018, he appeared in a hearing before US Congress to explain how Google’s algorithm worked, including the results that it favoured. He was vigorous in his defence of the search tool. “Getting access to information is an important human right” explained Pichai. But core principles of human rights are equality, non-discrimination and respect for the worth of every human, irrespective of race and culture. Search results that suggest beauty is confined to Caucasians simply to do not bear that out.
We decided to see the racial biases in Google Image search engine for ourselves and carried out a small practical experiment. We typed the search term “beauty ideal” in 6 different languages (English, Irish, Arabic, Indonesian, Japanese and Korean) in the Google Image search tool that most digital natives use for just about anything. The aim was to identify the different representations of racial groups that each translation of the term produced through the search engines ranking system using the Google Image Scraper tool.
These are some of the top search results from each language:
At first glance, the top results seem to be heavily representative of young, thin, white women which is a backwards step of beauty standards in many ways. The graph below shows the representations of racial groups in the first 20 results of each query:
According to Google, being white seems to be a “beauty ideal” across language and culture. What’s more disturbing is that some races are not even represented in some of the top search results! This is unrepresentative of the diversity in society’s actual views, as “beauty ideals” are unique, not just to individual opinion, but cultures and ethnicities. How challenging will it be to secure equality in society, when one of our most used tools reinforces some of the very barriers that we are trying to break down?
In Copolla’s approach, can we assume that Google, a worldwide multi-billion dollar tech giant, is racially biased against certain groups? Or would these results change if I wasn’t a young, white woman myself and these results are simply what Google thinks that I want to see based on the data it has on me? The mystery of how Google’s search algorithms work makes identifying what the causes of this misrepresentation of certain racial groups difficult to uncover. But, as far as I can see; one thing is for sure – there is a lack of diversity in the image representations of “beauty ideal”.
Does this make Google racist, or is it just a user-pleaser? What about representations of gender, sexual identity or age? There are so many areas for problems to occur in search results, especially in something as subjective (and arguably vague) such as a “beauty ideal”. Search engines like Google cannot please everyone. Something they can work on however, is keeping up society’s standards of inclusion and equality.
Although we don’t need Google to tell us if we meet the standards of beauty, there is something to be said about the lack of representation in digital media’s depiction of beauty. Google is used by everyone, so everyone should feel represented. This may seem like a challenge, but Google is one of the greatest innovators – this one should be a breeze. Come on Google, aren’t we all beautiful?