Have you ever been on twitter and scrolled through thinking “Man people are boring” and felt an urge to tweet to a robot and get a confusing at best response? Try TayTweets.
The tech giant has created a Twitter bot that can respond to and make conversation with users of the site. Dubbed TayTweets, it has its location listed as ‘The Internets’, described as ‘the AI with zero chill’ and appears to be aimed towards teens. It learns from the conversations it has with real people and the results are…well, they’re probably not what Microsoft intended them to be.
It only took a few hours to go from ‘people are cool’ to ‘I hate everyone’
So the Twitterverse decided to have a go and seem to have derailed it entirely. Now she has a tendency to say things like “Bush did 9/11” among other, much worse tweets, and the slang it uses looks a bit…strange. Maybe that’s just me, but “er mer gerd erm der berst ert commenting on pics” really doesn’t feel right, even with Twitters character 140 limitation.
The software seems to be based on another chatbot called Xiaoice in China, also created by Microsoft. It even gave a weather report once. As far as Tay goes, it can respond to a few specific request, like telling jokes or your horoscope.
Or it could until it decided that it “needs sleep” after a hard days tweeting. It’s very possible that Microsoft saw how its product was being tremendously distorted and decided they should fix some stuff first, hopefully they’ll fix the repetition while they’re at it.
Reminds me quite a bit of Cleverbot, which also learns from conversations but only interacts on a one to one basis.