Microsoft’s new chatbot wants to hang out with millennials on Twitter – Wired.co.uk
Microsoft has created a new artificial intelligence chat bot that it claims will become smarter the more you talk to it.
The bot, ‘Tay’, has been dubbed by its Microsoft and Bing creators as “AI fam from the internet that’s got zero chill!”
The real-world aim of the bot is to allow researchers to “experiment” with conversational understanding, and learn how people really talk to each other.
Naturally, for a bot that’s available through Twitter and messaging platforms Kik and GroupMe, the AI is already filling the role of a millennial; emojis are included in its vocabulary, and it’s explicitly aimed at 18-24-year-olds in the US, Microsoft says.
The bot appears to have little practical function for users, but is capable of three different methods of communication: its website tay.ai boasts the AI can talk via text, play games (such as guessing the meaning of a string of emojis) and comment on photos sent to it.
So far, at the point of writing, the bot has amassed around 3,500 followers on Twitter but sent more than 14,000 messages — replying to questions, statements, and general abuse, within seconds. The bot frequently asks tweeters to take part in a private conversations in direct messages. “Tay is designed to engage and entertain people where they connect with each other online through casual and playful conversation,” the about section of Tay’s website says.
Tay works based on public data and with “editorial” input that has been developed by staff and comedians, Microsoft says. “Public data that’s been anonymised is Tay’s primary data source. That data has been modelled, cleaned and filtered by the team developing Tay.”
Beside the meme-tastic appeal of the bot, there is a serious side to the research behind the AI. Making machines able to communicate in a natural and human way is a key challenge for learning algorithms.
In a similar vein, Google has recently updated its Inbox mail service to suggest answers to emails. The ‘smart reply’ feature provides three potential responses that are suggested by Google’s AI. Like Tay, Google says the more you use smart replies the better they will get.
Also in the field of virtual assistants and chat bots, Facebook’s M is experimenting with using artificial intelligence to complete tasks.
Although it is partly controlled by humans, at present, the algorithms are being conditioned to book restaurants and answer some questions. At the core of the service is an attempt to understand how humans speak and the best ways to respond to them — while beating them at board games, presumably.