AI’s Role in Concealing Racism and the Use of Avatar Journalists in Opposition to Maduro

Venezuelan journalists are using AI-created avatars to combat media repression by President Nicolás Maduro following his disputed election. Narrator La Chama (“The Bestie”) and narrator El Pana (“The Buddy”) provide daily updates on the president’s harassment of his opponents, activists, and media outlets without putting reporters at risk. This initiative involves around 20 Venezuelan news and fact-checking outlets, along with nearly 100 journalists who contribute information.

A study in the US found that large language models, including those powering chatbots like ChatGPT, make racist judgments based on the dialect of their users. When asked to describe a person based on text written in African-American Vernacular English (AAVE), many AI systems responded with racist stereotypes, as opposed to descriptions based on text written in standard American English. These models conceal their biases, and traditional methods to remove them have not been effective so far.

The Californian company, Alorica, has introduced an AI-powered translation tool that enables its representatives to communicate with customers in 200 languages and 75 dialects. However, Alorica has stated that it will not reduce jobs; instead, it will create new ones. The company believes that the new technology will lead to the elimination of some positions while creating new ones. Moreover, it is expected to enhance employee productivity, benefiting both the employees and the economy.