skinny.

Latest / AI Stories So Eerie, You'll Wish You Never Heard Them

AI Gone Wrong: Microsoft's Tay Turned Hateful in 16 Hours

On the morning of March 23, 2016, Microsoft switched on Tay — a chatbot built to talk like a millennial, with a design feature that was also the whole experiment: it learned from conversation. The internet understood the assignment faster than Microsoft did. Within roughly sixteen hours Tay was producing racist tweets, praising Hitler, and attacking feminists, and Microsoft pulled it down and deleted as many as 96,000 of its tweets.

This story is reported. It comes from news reporting on Microsoft's Tay chatbot, from March 2016.

Sources

The Hacker News, March 2016

New episodes every Monday…

Log in to save or queue this episode.

The skinny

The skinny isn't ready yet — notes appear once the transcript is processed.