Earlier today, we told you about a disturbing translation discovery on WeChat, in which certain uses of the term "黑老外" ("hei laowai" or "black foreigner") returned the N-word, a deeply offensive racial epithet.
WeChat has since responded to the news, with a spokesperson telling Sixth Tone: "We’re very sorry for the inappropriate translation. After receiving users’ feedback, we immediately fixed the problem.”
A quick check of the phrases we tested earlier confirmed that the changes to the translations had indeed been made.
The spokesperson also said that the app uses neural machine translation, and noted that it's constantly being tweaked to provide "more accurate, faithful, expressive, and elegant” translation results.
As Sixth Tone points out, many modern translation apps use machine learning. But without a human filter, the artificial intelligence can sometimes let offensive slurs through, as was seen in 2016 when Twitter users taught an account to support Hitler.
“This is the way social change should happen," Shanghai-based theater producer Ann James said of WeChat’s swift rectification of the issue. James first brought the translation to our attention.
0 User Comments