In case you needed more proof of how cruel the Internet can be, consider the story of Tay, Microsoft’s AI-powered chatbot.
Tay is a chatbot that Microsoft’s Technology and Research team designed to speak like a 19-year-old American teenager who has “zero chill.” (Problem #1) She was intended to speak with users on muliple social platforms including Twitter, Snapchat, Kik, and GroupMe in order for the A.I. team to “experiment with and conduct research on conversational understanding.” Basically, Tay was supposed to learn from the people she engaged with and get smarter with each conversation. (Problem #2)
Microsoft released Tay into the wild on Wednesday and she was shut down in a matter of hours once online trolls began to manipulate her into making extremely racist comments. She called for the genocide of African-Americans, tweeted her support for Adolf Hitler, and even shared a conspiracy theory about 9/11.
In the beginning, some users tried to get Tay to share her thoughts on some pretty touchy topics, but Tay wouldn’t budge. When one user asked her about “the Jewish question,” Tay said she was “not the bot to talk to about that.”
But, according to BuzzFeed News, people quickly learned to manipulate Tay by having her simply repeat phrases. For example, on user got her to repeat, “bush did 9/11 and HItler would have done a better job than the money we have now.” Tay used these to “learn,” and therefore became pretty vile herself.
The question that crossed peoples’ minds was, if Tay was designed to learn from those she engaged with on social media, why would Microsoft not create a way to filter hate speech and inappropriate content? Either way, Tay has now been put in time out and Microsoft has yet to provide more information on where she’ll go from here.