Linguistics and Languages

AI generates covertly racist decisions about people based on their dialect

V. Hofmann, P. R. Kalluri, et al.

This groundbreaking research by Valentin Hofmann, Pratyusha Ria Kalluri, Dan Jurafsky, and Sharese King delves into the hidden biases present in language models, specifically targeting dialect prejudice against African American English (AAE). The findings unveil how these models perpetuate negative associations that not only challenge existing stereotypes but lead to serious real-world consequences.