Meta’s No Language Left Behind (NLLB) project is expanding to include 200 lesser-spoken languages, such as Scottish Gaelic and Galician. The project uses AI trained on data from the Opus repository and sources like Wikipedia to create multilingual language models. However, experts like William Lamb suggest that Meta should consult with native speakers to improve the accuracy of translations. Lamb, a professor of Gaelic ethnology, notes that Meta’s translations in Scottish Gaelic are not yet satisfactory due to the lack of input from native speakers. He recommends that Meta work with organizations like the BBC, which produce high-quality content in minority languages, to improve their translations. Alberto Bugarín-Diz, a professor of AI, also emphasizes the importance of collaboration between linguists and tech companies to refine data sets and improve translation quality. While Lamb is hesitant to recommend the use of Meta’s translation tools due to errors in the data, Bugarín-Diz believes that user feedback is essential for companies to invest in improving their translation services. Ultimately, understanding the limitations of AI tools is crucial before relying on them for accurate translations.
Source link
Source link: https://www.euronews.com/next/2024/06/30/meta-expands-ai-translation-to-200-languages-but-experts-suggest-talking-to-native-speaker
Can AI preserve Europe’s endangered languages from extinction? #AIrescue
Preeti
290 Views
0 Votes
GIPHY App Key not set. Please check settings