"Whether it's virtual assistants on our phones or chatbots on government websites, the large language models (LLMs) that power AI tools such as ChatGPT are almost everywhere online. But growing evidence suggests these LLMs are judging dialect speakers harshly. In 2024 researchers from the University of California, Berkely, tested ChatGPT's responses to several varieties of English dialects from places like India, Ireland, and Nigeria."
"In July 2025, an AI assistant used by the Derby City Council struggled to understand a radio presenter's Derbyshire dialect when she used words like mardy (complaining) and duck (dear/love) during a call she'd made on air to test the AI assistant. Other dialect speakers have experienced much worse effects. As businesses and governments use more AI in their services, researchers are getting worried. AI developers, however, see an opportunity to provide tailored LLMs for dialect speakers."
"They then presented the models with texts in either standard German or one of seven German dialects, such as Bavarian, North Frisian and Kolsch. The models were asked to describe the speakers of these texts with personal attributes, and to then assign individuals in different scenarios. For example, the models were asked who should be hired for low-education work or where they think the speakers lived. In nearly all tests the models attached stereotypes to dialect speakers."
LLMs power many virtual assistants and chatbots but show bias against dialect speakers, producing more stereotyping, demeaning, and condescending responses. A 2024 UC Berkeley test found ChatGPT responded to nonstandard English varieties with 18% more stereotyping, 25% more demeaning content, and 15% more condescension than for American or British English. Some models fail to comprehend dialects at all, as seen when a Derby City Council AI assistant misunderstood Derbyshire words like mardy and duck during a 2025 on-air test. A German evaluation of ten LLMs found pervasive stereotypes linking dialects to low education and farm work.
Read at www.dw.com
Unable to calculate read time
Collection
[
|
...
]