Skip to main content

Microsoft made a chatbot that tweets like a teen

Microsoft made a chatbot that tweets like a teen

Share this story

Microsoft is trying to create AI that can pass for a teen. Its research team launched a chatbot this morning called Tay, which is meant to test and improve Microsoft's understanding of conversational language. But not just any conversational language — the most #teen tumblrcore what-are-you-even-saying type of conversational language. For reference, Microsoft describes Tay as "AI fam from the internet that's got zero chill!"

Here are some examples of what Tay's been tweeting:

Admittedly, those are some of the weirder, more naturally internet-y responses; many others still sound like they're written by a 40-something trying to sound cool.

Tay is available on Twitter, GroupMe, and Kik. On Twitter, you just have to tweet at the bot to get a reply (it may ask you to start at DM, if it's receiving too many responses); on the other platforms, you need to find Tay's account and begin messaging it.

Tay's responses should become more natural over time

The chatbot was created in collaboration between Microsoft's Technology and Research team and its Bing team. They say Tay's conversational abilities were built by "mining relevant public data" and combining that with input from editorial staff, "including improvisational comedians." The bot is supposed to learn and improve as it talks to people, so theoretically it'll become more natural and better at understanding input over time.

Microsoft made Tay able to respond to a handful of specific requests, beyond straightforward chatting. It can tell jokes and stories, make memes out of photos, and deliver a horoscope. You can also play some weird emoji guessing game with it.

While it's hard to see Tay's forward-facing functions as particularly useful, as a research project it could be quite valuable. Natural-language assistants (like Siri and Cortana) are increasingly important parts of smartphones and computers, but none of them are truly skilled at responding to natural, complex language — for the most part, they can only handle some straightforward phrases. By learning how different groups of people actually talk, Microsoft may eventually be able to better tailor its own assistant to deliver a helpful response. And as messenger bots begin to look more and more like a critical interface for getting stuff done in the future, the ability to tweet like a teen starts to sound a lot more valuable.