Story perspectives
AI Language Divide: English Dominates, Minorities Left Behind
9/3/2025
49 8
1 of 1
Story summary
- Johns Hopkins researchers identified that AI tools like ChatGPT create a digital language divide, favoring English over minority languages.
- Their study showed that large language models (LLMs) produce biased information based on query language, neglecting lower-resource languages.
- Switzerland introduced Apertus, an open-source language model focused on transparency and data protection compliance.
- This initiative aims to ensure equitable AI access and may encourage similar developments in other EU countries.
- Ukrainian researchers evaluated LLMs for pro-Ukrainian sentiment, finding Canadian models most supportive and Chinese models least favorable.
