ChatGPT-4o Shows Language Gaps in Medical Translation, Study Finds
A comprehensive evaluation of AI-powered medical translation reveals significant performance gaps for languages like Armenian and Somali. Researchers found that human-in-the-loop approaches substantially improved translation quality while cutting completion times. The study highlights ongoing challenges in achieving equitable machine translation for healthcare communications.
AI Translation Shows Promise and Pitfalls in Healthcare
New research examining machine translation of patient discharge instructions reveals both the potential and limitations of current AI systems in clinical settings. According to findings published in npj Digital Medicine, ChatGPT-4o demonstrated variable performance across six languages, with particularly concerning results for digitally underrepresented languages including Armenian and Somali.