Without intervention, technological advances could do more harm than good. 

Artificial intelligence (AI) is already present in many facets of our lives – the voice assistants that we ask to set timers or dim lights, the predictive text in our messaging apps, and the algorithms that recommend what YouTube videos we watch.  Nearly everywhere we look, AI plays a role in what we see, hear, and do, and not without controversy.  For instance, YouTube’s AI-based recommendation engine has come under fire for radicalizing users with extremist content[1].  But perhaps what’s most troubling is how AI is impacting the lives of people of color.

The biggest issue is AI’s built-in bias.  The workforce is overwhelmingly White; at Google, a global leader in AI, only 2.5% of its staff is Black[2].  And it’s just slightly better at Facebook and Microsoft – 4%.  Though existing AI professionals may not purposely encode their biases, they have racial blind spots, and those shortcomings surface in the products they create.

A 2019 Wired report showed that even the best algorithms struggled to accurately identify Black faces[3].   Case in point, in 2015, Google was widely criticized after its Google Photos AI tagged two Black users as gorillas[4].

It’s easy to think – so what if a smartphone app misidentifies someone?  But it matters greatly because the same underlying technology is being used in more significant ways.  In just the past year, multiple reports have surfaced about facial recognition software misidentifying Black men as criminals.  One man, Nijeer Parks, was falsely accused of shoplifting and trying to run over a police officer, spending 10 days in jail and paying $5,000 in fees – all because of AI[5].

When Black AI executives, like Google’s Timnit Gebru, speak up, they face scrutiny, pushback, and even termination[6].  There are activist groups and organizations mobilizing to fight for equity in AI products and applications, such as Black in AI.  But we can’t only lean on a handful of professionals who are severely outnumbered in their field.

We have to sound the alarm about how AI reinforces social inequities and injustices, under the guise of making our lives easier.

Imagine AI-assisted hiring software, programmed by a team that doesn’t include a single person of color.  This AI is fed a list of criteria and then sorts through job applications in search of qualified candidates.  What happens if that criterion is accidentally discriminatory?  Maybe it searches for certain zip codes or name spellings, or maybe it looks for degrees from a certain caliber of university.  Applicants from certain backgrounds – who have the work experience and credentials – could be excluded, simply because of the programmers’ oversights.

The more we incorporate AI in all aspects of society, the more we risk fortifying the existing structures that uphold racism.  AI does represent a new frontier in modern living, but without proper checks and balances, and more inclusivity behind the scenes, it will only improve life for a chosen few.

[1] Roose, Kevin. (June 8, 2019). The Making of a YouTube Radical. Retrieved from: https://www.nytimes.com/interactive/2019/06/08/technology/youtube-radical.html

[2] Howard, Ayanna and Isbell, Charles. (September 21, 2020). Diversity in AI: The Invisible Men and Women. Retrieved from: https://sloanreview.mit.edu/article/diversity-in-ai-the-invisible-men-and-women/

[3] Simonite, Tom. (July 22, 2019). The Best Algorithms Struggle to Recognize Black Faces Equally. Retrieved from: https://www.wired.com/story/best-algorithms-struggle-recognize-black-faces-equally/

[4] Zhang, Maggie. (July 1, 2015). Google Photos Tags Two African-Americans As Gorillas Through Facial Recognition Software. Retrieved from: https://www.forbes.com/sites/mzhang/2015/07/01/google-photos-tags-two-african-americans-as-gorillas-through-facial-recognition-software/?sh=16fca499713d

[5] Hill, Kashmir. (December 29, 2020). Another Arrest, and Jail Time, Due to a Bad Facial Recognition Match. Retrieved from: https://www.nytimes.com/2020/12/29/technology/facial-recognition-misidentify-jail.html

[6] Hao, Karen. (December 4, 2020). We read the paper that forced Timnit Gebru out of Google. Here’s what it says. Retrieved from: https://www.technologyreview.com/2020/12/04/1013294/google-ai-ethics-research-paper-forced-out-timnit-gebru/