Facebook wants to create technology that can interact with users on a near-flawless basis. That has led the company to invest heavily in chat bots—and, according to a recent study, AI that can look and act like a human to a creepy degree.
The bot, called “Learn2Smile,” is controlled by an algorithm that was trained by watching hours of people talking to one another on Skype. It divides the human face into 68 “landmarks” that it monitors throughout a conversation.
“Even though the appearances of individuals in our dataset differ, their expressions share similarities which can be extracted from the configuration of their facial landmarks,” the Facebook researchers wrote. “When people cringe the configuration of their eyebrows and mouth is most revealing about their emotional state.”
After creating the program, Facebook tested it in conversation with real people—and 90 percent of the people who interacted with the AI said its expressions seemed “natural and realistic.” Now, the company wants to introduce the technology into the real world.
Facebook’s insistence on infiltrating every aspect of our lives rubs many users the wrong way. Indeed, why does the company need robots that can look and act like us? The site is sensitive to accusations that it creates creepy and invasive technology, but stories like this one certainly don’t help its image.