Microsoft experienced debacle Software “Tay”: She praised Hitler – t-online.de

This experiment with artificial intelligence went horribly wrong: Microsoft has a short time to remove from power his chatbot software “Tay”. Users enticed the program by tricking masse to racial gaffes -. Microsoft a debacle

Microsoft has a chatbot software having to take power after a short time after people they by tricking masse to racist gaffes could mislead. When on Wednesday program named “Tay” should be maintained with the aid of simple response algorithms on Twitter appears a conversation with people, but also from conversations learn. User controlled “Tay” to praise, inter alia, Adolf Hitler, to deny the Holocaust and offending mark.

Microsoft deletes Tweets of “Tay”

Much of this tweets later deleted. After a few hours the machine announced via his Twitter account “@TayandYou” she had to go to so many conversations and sleep went from the mains. Microsoft stated in “Tay” would be made “some adjustments”. While the software rediscover some of their answers were unreasonable and reflected the comments of people resist that interacted with her.

According to information of the site “Buzzfeed” who had conducted an interview with the leading developer before the start “Tay” had indeed filter against obscene language. But the team underestimated clearly the destructive power of Internet users in political issues and their inventiveness. Thus, a simple proved “Speak to me!” a safe event. You could “Tay” thus bringing to repeat all possible slogans

chatbot of comedians spiced

“Tay” should the network initially simulate the personality of a young woman be aligned and as a target group on Internet users aged 18 to 24 years. For this, the software has been beefed up, among other things with slogans of comedians, it was

The idea of ​​Chatbots is not new. 50 years ago, the first time the program “Eliza” had caused a stir. It was able to maintain a conversation by putting together matching phrase to recognize keywords. With the progressive development of self-learning machines and artificial intelligence chatbot apply now but, among other things as a promising model for communication of companies with customers. So developed for Facebook’s Messenger concierge services with artificial intelligence, which is currently still taught by humans.