AI Translations Break Down Language Barriers Online

Published 26 October 2022

Author
Pia Benthien
2 min read

Experts claim that the internet has turned the world into a global village. But in 2020, nearly half of all...

German start-up OBTranslate focuses its AI translation services on rural African languages. The company crowdsources language data sets from a community of paid volunteers, which it runs through deep learning programs. This results in automatically generated translations that are then sold to businesses with non-English speaking employees looking to become more competitive in international markets.

Big tech companies like Meta are also investing in translation. Google recently announced an AI-based cloud service marketed to small businesses called Translation Hub, which automatically translates documents in under 30 seconds – no humans required. The resulting translations are basic, but they can be edited after creation.

There are critics of AI-based translations, though – like American cognitive scientist Gary Marcus, who says that AI translation is not actually as ‘intelligent’ as it seems. “What these systems are ultimately doing is mimicry,” he told Undark, an editorially independent digital magazine based in Cambridge, Massachusetts. “When you’re mimicking something, you can do that to a high degree, but it’s still kind of like being a parrot […], because we don’t think parrots actually understand what they’re talking about.”

AI can’t identify nuance or other fundamental aspects of human communication, which is why it’s not always the best tool for solving human problems. But it can still be useful for sorting through quantities of data that humans aren’t equipped to process themselves – like millions of words.

We’re keeping tabs on how the AI industry is addressing accessibility issues as it develops. Read our EmTech Takeaways report for more recent insights.

German start-up OBTranslate focuses its AI translation services on rural African languages. The company crowdsources language data sets from a community of paid volunteers, which it runs through deep learning programs. This results in automatically generated translations that are then sold to businesses with non-English speaking employees looking to become more competitive in international markets.

Big tech companies like Meta are also investing in translation. Google recently announced an AI-based cloud service marketed to small businesses called Translation Hub, which automatically translates documents in under 30 seconds – no humans required. The resulting translations are basic, but they can be edited after creation.

There are critics of AI-based translations, though – like American cognitive scientist Gary Marcus, who says that AI translation is not actually as ‘intelligent’ as it seems. “What these systems are ultimately doing is mimicry,” he told Undark, an editorially independent digital magazine based in Cambridge, Massachusetts. “When you’re mimicking something, you can do that to a high degree, but it’s still kind of like being a parrot […], because we don’t think parrots actually understand what they’re talking about.”

AI can’t identify nuance or other fundamental aspects of human communication, which is why it’s not always the best tool for solving human problems. But it can still be useful for sorting through quantities of data that humans aren’t equipped to process themselves – like millions of words.

We’re keeping tabs on how the AI industry is addressing accessibility issues as it develops. Read our EmTech Takeaways report for more recent insights.