AI seems to have trouble with Minnesota accents
AI seems to have trouble with Minnesota accents 02:52
MINNEAPOLIS — Artificial intelligence is having trouble understanding some American English accents, with the Minnesota accent causing the most confusion.
A study published last month by language resource organization Guide2Fluency identified the top 30 U.S. accents that disrupt speech recognition technology in AI-powered devices and apps such as Alexa, ChapGPT and Siri.
The study, conducted among 3,000 adults split across age, gender and region, ranked the Southern accent as the one with the highest transcription and comprehension error rate, with the study authors saying that the AI was stumped by the Southern accent's “unique phrasing and slower speech pattern.”
The New York and New Jersey accents ranked second and third respectively, while Texas and Boston accents rounded out the top five.
Minnesota ranked 13th on the list, but the study authors did not elaborate on its troubling characteristics.
Regional slang and unique vowel pronunciations are major sources of confusion for AI, but “cultural nuances,” such as the Cuban accent being a mix of Southern American English and Cuban Spanish, and the Cajun French accent, are also major sources of confusion for AI.
The authors say the study raises concerns about the perception of accessibility, effectiveness and trust of AI technology.
More from CBS News
Stephen Swanson
