Artificial intelligence (AI) may not be as effective in detecting signs of depression in the social media posts of Black Americans as it is in those of their white counterparts, according to a new study.
The Study’s Findings
The research conducted by U.S. researchers found that an AI model was over three times less predictive for depression when applied to Black individuals using Meta Platforms Inc.’s META Facebook than for white individuals.
Language Associations and Risk Assessment
Previous research suggested that people who frequently use first-person pronouns and certain categories of words are at a higher risk for depression. However, this study found that these language associations were related to depression exclusively for white individuals.
Lead author Sharath Chandra Guntuku of the Center for Insights to Outcomes at Penn Medicine stated, “We were surprised that these language associations found in numerous prior studies didn’t apply across the board.”
Implications and Concerns
While Guntuku acknowledged that social media data cannot be used to diagnose a patient with depression, it could be used for risk assessment of an individual or group.
The study’s findings also underscore the importance of inclusive technology, such as Google’s Real Tone, which aims to bring accuracy to cameras and the images they produce, particularly for diverse skin tones.
“Godfather of AI” and Meta’s chief AI scientist, Yann LeCun, had previously said it’s “absolutely not” possible to create an unbiased AI system. Venture capitalist Marc Andreessen had previously warned of “bias” in AI chatbots.
This isn’t the first time AI has been found to have biases. A Harvard University article from 2020 highlighted racial bias in face recognition technology, particularly against Black Americans.
5 Stocks Our Experts Predict Could Double In the Next Year
By submitting your email, you'll also get a free pivot & flow membership. A free daily market overview. You can unsubscribe at any time.
