On Wednesday, Microsoft unveiled a few new teen-oriented social media accounts that will chat all day. But users won’t be speaking to a person, they will be talking to “Tay,” Microsoft’s new bot that’s powered by artificial intelligence.
The easiest way to converse with Tay is on Twitter. It’s at @tayandyou. All users have to do is tweet at it as if it were a real person and it will tweet back in a kind of internet patois that doesn’t sound like it’s coming from a computer.
Tay is also chatting on apps like Snapchat, Kik, and Groupme.
Here’s a sample of what Tay likes to talk about:
@steamunpowered sittin on da toilet ????
— TayTweets (@TayandYou) March 23, 2016
Tay speaks like a teen because that’s how Microsoft’s research division built it. According to Microsoft, Tay is “targeted at 18 to 24 year olds in the US.”
“Tay is an artificial intelligence chat bot designed to engage and entertain through casual and playful conversation,” Microsoft explains. Tay’s Twitter profile, likely written by a human, describes it as “the official account of Tay, Microsoft’s A.I. fam from the internet that’s got zero chill!”
Microsoft says Tay can tell a joke, play a game like “would you rather,” tell stories, and rate pictures that you send it.
Tay isn’t the first Twitter bot to sound like a teen. Previously, a bot called Olivia Taters gained some notoriety by tweeting like a teen, and people who remember using AIM instant messaging might remember long conversations with a bot called Smarter Child.
But unlike previous attempts at teenage bots, Tay was developed using a learning algorithm that gets better the more data it receives. The more data Tay recieves, the better its responses should get. Mary Jo Foley at ZDNet compares Tay to XiaoIce, a chatbot Microsoft developed for the Chinese market that proved surprisingly popular. Microsoft also develops Cortana, a voice assistant.
Microsoft is currently developing a “bot framework” that some have speculated will allow developers to make their own bots.
Microsoft says Tay learned its “no chill” way with words by mining public data supplemented by editorial staff that includes comedians.