Meta used the Bible to train AI models to learn over 1,000 languages

“We can use what that model learned to then quickly build speech systems with very, very little data,” said Michael Auli, a research scientist at Meta who worked on the project.

“For English, we have lots and lots of good data sets, and we have that for a few more languages, but we just don’t have that for languages that are spoken by, say, 1,000 people.” 

The researchers now claim their models can converse in over 1,000 languages but recognize more than 4,000.

In addition, compared to models from rival companies, including OpenAI Whisper, Meta’s version had half the error rate, despite covering 11 times more languages.

All is not rosy, however. The scientists claim that their new models may mistranscribe certain words or phrases and that their speech recognition models yielded more biased words than other models, albeit only 0.7% more. 

Chris Emezue, a researcher at Masakhane, an organization working on natural-language processing for African languages, who was not involved in the project, told MIT that the use of religious text to train the models may be problematic.

“The Bible has a lot of bias and misrepresentations,” he explained. Is this development a step forward for language models or is it too controversial to be impactful?