Key facts
- University of Washington research found AI models rated US and UK accents highest and Chinese and Nigerian accents lowest.
- In the scripted tests, every speaker said the same words, according to the research.
- The AI models transcribed the accented speech accurately, meaning they understood the accents, the research said.
University of Washington research found that some of the most advanced AI models rated English speakers with different accents differently, both on what they said and how they said it, according to a report published on November 19, 2026.
In the scripted tests, every speaker said the same words, and the AI models transcribed the accented speech accurately, meaning the models understood the accents, the research said. The models nonetheless scored US and UK accents highest and Chinese and Nigerian accents lowest, according to the research.
The research examined how the models rated speakers when asked to grade a job interview or test answers, finding that a person's accent may change the score.
