A recent report has uncovered alarming racial biases within popular AI tools like ChatGPT and Gemini, suggesting covert racism that could significantly impact various aspects of society.
The study, conducted by a team of technology and linguistics researchers, sheds light on how these AI models are racist against African-American dialects.
AI Becoming More Racist
The report by Cornell University suggests that AI tools harbor stereotypes against speakers of African American Vernacular English (AAVE), a dialect predominantly used by Black Americans.
According to Valentin Hoffman, a researcher at the Allen Institute for Artificial Intelligence and co-author of the paper, these biases pose serious concerns, particularly in scenarios like job screening.
Targets Depicted As Stupid
The research indicates that AI models are more likely to perceive AAVE speakers as “stupid” and “lazy”
The researchers reported that the outcome could potentially lead to biased outcomes such as lower employability and harsher legal judgments.
While ethical guidelines have been implemented to regulate AI behavior, experts warn that these measures only mask the underlying issue.
