Minnesotan among US accents most confusing to AI, study finds

MINNEAPOLIS — Artificial intelligence has issues comprehending some American accents, with Minnesotan among the ones causing the most confusion.

A study published last month by the language resource organization Guide2Fluency pinpointed the top 30 U.S. accents that trip up voice recognition technology in AI-driven devices and apps like Alexa, ChapGPT and Siri.

With a survey size of 3,000 adults — broken down by age, gender and geography — the "Southern" accent was ranked first for the highest error rates in transcription and comprehension. Study authors say AI gets perplexed by the accent's "unique expressions and drawl."

New York and New Jersey accents were ranked second and third, respectively, with Texan and Bostonian rounding out the top five.

Minnesotan is ranked 13th on the list, although study authors didn't elaborate on its tricky features.

Regional slang and distinct vowel sounds are big sources of AI befuddlement, as well as "cultural nuances," like the Cuban accent's blend of Southern American English and Cuban Spanish, or Cajun's inherent French flourishes.

Authors say the study raises concerns about the accessibility, effectiveness and perceived trustworthiness of AI technology.