Why Is Algorithmic Bias So Bad
By Manroop Bains
How can we stop algorithmic bias?
Recently, Twitter users have discovered a bias in the Twitter image-detection algorithm that “crops out black faces in the favor of white ones.” In a nutshell, social media algorithms decide what social media users see on their feeds based on relevancy.
“Technological racism,” or in this case, algorithmic bias is much more difficult to point out because it is “‘largely unavoidable.”’ The chances of a black defendant being unfairly sentenced have increased significantly, but not just by a judge’s verdict, but by sentencing algorithms that work hand in hand with human biases. Predictive models used in hospitals “methodically deny” treatment for some Black and Hispanic patients, some of which are regularly given to less-sick white patients. If this is the norm, just imagine how many Black and Hispanic patients are left untreated.
Frequent exposure to algorithmic bias tends to shape how one can see themself, or others, and can result in long-term psychological impacts in adolescents and mainly for those who spend most of their time online. According to a study by the Pew Research Study Center in 2018 “,95 percent of teens have access to a smartphone, and 45 percent describe themselves as being online “‘‘almost constantly.”’’ Hispanic teens, in particular, spend more time online than their white peers. Given America’s reliance on remote learning during the pandemic, adolescents are likely spending even more time on the internet than they did before.” It was also reported that those who are on the receiving end of discrimination tend to have poor mental health which can alter gene expression across the lifespan. This could result in higher rates of poor mental health in people of color, as opposed to white people.
People of color aren’t the only ones that are negatively affected by algorithmic bias. The content-filtering algorithm in TikTok can place people in “echo chambers where everyone looks the same.” This can affect one’s ability to empathize, negatively affecting the development of skills needed to “thrive” in a country that’s diverse. This also opens a window of opportunity for hate groups to spread propaganda to recruit others and “inspire them to commit real-world violence.”
On the bright side, big tech companies have come to a realization that their platforms rely on algorithmic bias, and as a result, have set up projects to combat it. A team of researchers led by Henriette Cramer at Spotify are working together to address algorithmic bias. Organizations like AI Now Institute at NYU and the Algorithmic Justice Team at MIT are currently creating guidelines for “ethical artificial intelligence. It is difficult to correct bias in humans, so it needs to be corrected at what seems to be the root of the problem, in this case, it’s social media algorithms. Once a bias has been detected, certain checks and balances should be placed. Social media algorithms may seem small and harmless, yet its impacts can be powerful.