It’s the fifth big tech company forced to admit to the practice this year.
The news: Facebook paid contractors to listen to, and transcribe, audio clips generated by people using its Messenger app, according to Bloomberg. They were not told where the clips were recorded, or how they were obtained, and the conversations were sometimes deeply personal. The contractors were employed to check if Facebook’s artificial intelligence software correctly interpreted the messages, which the company says were not traceable back to users. Facebook suspended the program a little over a week ago, a spokesperson told Bloomberg.
Déjà vu: This story follows almost identical revelations about Amazon in April, Google and Apple in July, and Microsoft earlier this month. In each case, the companies didn’t disclose the practice to users. Most of them have since suspended their operations, pending investigations.
Were you affected? Facebook said the only users affected were those who chose the option in its Messenger app to have their voice chats transcribed, a service it launched in 2015. But there was never any mention of human involvement in that process.
Behind the curtain: The fact that the practice of getting humans to check voice recordings and responses is so ubiquitous shows just how limited AI still is in its ability to recognize words and meaning (though it’s improving). It’s also a reminder that for many AI services, there’s still an army of (usually low-paid) data workers toiling away behind the scenes.
Sign up here for our daily newsletter The Download to get your dose of the latest must-read news from the world of emerging tech.