When it comes to search engines such as Google, it turns out that online tools, like racist robots, reproduce the biases that persist in the social world. They are, after all, programmed using algorithms that are constantly updated on the basis of human behavior and are learning and replicating the technology of race, expressed in the many different associations that the users make. This issue came to light in 2016, when some users searched the phrase “three Black teenagers” and were presented with criminal mug shots. Then when they changed the phrase to “three White teenagers,” users were presented with photos of smiling, go-lucky youths; and a search for “three Asian teenagers” presented images of scantily clad girls and women. Taken together, these images reflect and reinforce popular stereotypes of Black criminality, White innocence, or Asian women’s sexualization that underpin much more lethal and systemic forms of punishment, privilege, and fetishism respectively.34 The original viral video that sparked the controversy raised the question “Is Google being racist?,” followed by a number of analysts who sought to explain how these results were produced:
The idea here is that computers, unlike people, can’t be racist but we’re increasingly learning that they do in fact take after their makers … Some experts believe that this problem might stem from the hidden biases in the massive piles of data that algorithms process as they learn to recognize patterns … reproducing our worst values.35
According to the company, Google itself uses “over 200 unique signals or ‘clues’ that make it possible to guess what you might be looking for.”36 Or, as one observer put it, “[t]he short answer to why Google’s algorithm returns racist results is that society is racist.”37 However, this does not mean that we have to wait for a social utopia to float down from the clouds before expecting companies to take action. They are already able to optimize online content in ways that mitigate bias. Today, if you look up the keywords in Noble’s iconic example, the phrase “Black girls” yields images of Black Girls Code founder Kimberly Bryant and #MeToo founder Tarana Burke, along with images of organizations like Black Girls Rock! (an awards show) and Black Girls Run (a wellness movement). The technical capacity was always there, but social awareness and incentives to ensure fair representation online were lacking. As Noble reports, the pornography industry has billions of dollars to throw at companies in order to optimize content, so advertising cannnot continue to be the primary driver of online content. Perhaps Donald Knuth’s proverbial warning is true: “premature omptimization is the root of all evil.”38 And so the struggle to democratize information gateways continues.