Racial Bias in artificial intelligence

by Molly Dye
February 3rd, 2022

As the world grows more and more dependent on technology and artificial intelligence (AI), there are many implications and flaws to such advances that impact people’s daily lives. One of the biggest issues that perpetuates the use of artificial intelligence pertains to racial and economic inequities and biases that are inherent to AI.

The issue can be as simple as algorithms like Google’s search engine producing ads related to criminal activity when users searched traditionally associated Black names, or Twitter’s image-cropping system “preferring” white people over Black people or simply not recognizing their faces at all. Most recently, in November 2021, reports revealed that Facebook’s algorithms were directing content with hate speech towards users of color. Though these issues may not directly impact people’s lives at first glance, the unregulated and persistent growth of racism and bias in AI deepen and worsen discrimination and exacerbate injustice and inequality.

In Twitter’s imagine-cropping case, the algorithm was not only not recognizing Black faces, but also those with white hair and aging skin, those wearing Hijabs and people in wheelchairs. The primary faces it would recognize were white, slim and young. Representatives at Twitter noted the issue as “unintentional bias,” but there was still a human being behind the screen who wrote that algorithm to contain biases affecting people’s lives. As the ACLU reports, the tech industry faces a lack of representation of employees who can understand and address the inequalities and harm that stem from flawed AI.

The problem goes beyond racist and biased programming on social media sites. In 2019, an algorithm used in U.S. healthcare facilities was found to be less likely to refer black people over white people with similar health issues to programs that would improve their care. Over 200 million people could have been impacted by algorithms like this one in hospital care. Other algorithms and AI tools have contributed to housing discrimination, including inequality in tenant selection and decisions on loan pricing that have discriminated against lenders of color. Employers use algorithms to screen applicants which impact multiple minority groups – particularly, people with disabilities – which contributes to a high percentage of poverty in these groups.

When considering the level at which artificial intelligence is used at every level of daily life, it is essential that companies and government entities begin to address the bias that impacts minority groups in the United States. Many tech giants function with little to no regulation in terms of their algorithms, and virtually any company, institution, or organization uses algorithms to speed up processes that used to be done by hand. These biases in AI come at the cost of equality in healthcare access, housing, and employment access among other issues.

Organizations including Public Citizen and the ACLU are calling on federal agencies and the Biden Administration to take action to hold AI systems accountable for their potential to cause harm. The goal maintains that the technology industry prioritizes creating anti-racist AI system in an effort to avoid deepening inequalities at every societal level in the United States.