We already know Artificial Intelligence can be biased, a number of recent incidents have shown as much. But, exactly how biased can technology be?
As per a new study: Very. A study used AI to analyse social media to see whether the technology can pick up signs of depression from posts made online. The results speak volumes.
Biased AI
The analyses found that while artificial intelligence (AI) can detect signs of depression in white Americans through social media use, it may not be as effective for Black individuals. In fact, it was three times less predictive when applied to Facebook posts by Black people, as compared to those by White people.
“Race seems to have been especially neglected in work on language-based assessment of mental illness,” wrote the authors of the study published in the Proceedings of the National Academy of Sciences (PNAS).
Research methodology: ‘I’ talk
For their research, they used an “off the shelf” AI toll to analyse the language of 868 volunteers, including equal numbers of Black and white adults with similar demographics like age and gender.
Participants were also asked to complete a depression screening questionnaire commonly used by healthcare providers.
As per previous research on social media posts, those individuals who frequently use first-person pronouns like I, me and mine, and certain words or terms that can be considered self-deprecating are more at risk of depression.
Interestingly, the study revealed that the use of first-person pronouns and self-deprecating language was only linked to depression only in white individuals.
Study co-author Sharath Chandra Guntuku said that they “were surprised that these language associations found in numerous prior studies didn’t apply across the board.”
While he acknowledged that social media data alone cannot diagnose depression, he said it could aid in risk assessment for individuals or communities.
(With inputs from agencies)
