Microsoft still set on AI bots after Tay fiasco

A screengrab of the Twitter account of Tay, which interacted with Twitter users and learnt from their responses.
A screengrab of the Twitter account of Tay, which interacted with Twitter users and learnt from their responses. PHOTO: SCREENGRAB FROM TAY TWEETS/ TWITTER

SEATTLE • Microsoft chief executive officer Satya Nadella said his company is committed to the emerging market for artificial intelligence-based chat software, one week after the company's first Internet chatbot in the United States was so manipulated by users that it had to be shut down.

"It's a simple concept, yet it's very powerful," Mr Nadella said about the strategy the company has dubbed Conversations as a Platform. "It is about taking the power of human language and applying it more pervasively in all our computing."

In a keynote speech on Wednesday at the company's Build conference for software developers in San Francisco, Mr Nadella outlined Microsoft's strategy for conversation-based products working with bots - software that uses artificial intelligence to automate tasks - as well as tools for developers to create their own.

Mr Nadella has been trying to expand Microsoft's reach and influence in artificial intelligence, emphasising how the technology can augment mobile and cloud-computing products.

He views the bot strategy as a key way to bring the benefits of artificial intelligence to businesses, consumers and developers. He also predicts that such programs will dominate the next generation of computing, replacing apps in some scenarios.

On March 23, Microsoft released Tay, an online AI chatbot designed to mimic the personality of a teenager, to attract millennial users. However, the company had to yank the program after Internet users taught it to spew racist, sexist and pornographic remarks in what the company called a "coordinated attack" that took advantage of a "critical oversight".

In his keynote, Mr Nadella said the company is taking a principled approach to artificial intelligence in the wake of Tay's misbehaviour. "We want to build tech so it gets the best of humanity, not the worst," he said.

Microsoft is not the only company pursuing bots. Tencent's WeChat, one of the inspirations for Microsoft's strategy, has been an early leader. Facebook is also opening up its Messenger app to bots that assist with shopping and appointments, and the social media leader is expected to expand that effort in the coming months.

The Wall Street Journal reported in December that Google is working on bots, and the artificial intelligence programs are rampant on messaging services like Slack and Telegram, among others.

Microsoft also showed a video of an artificial intelligence program for visually impaired customers that answered questions about food and prices on a restaurant menu using a mobile phone camera to read and voice software to speak.

Still, the Tay debacle loomed large over the proceedings. Though Microsoft plans to recommission the bot once it has been made safer from exploitation, Gartner analyst Chris Howard said the ease and speed of Tay's public re-programming may delay corporate adoption.


A version of this article appeared in the print edition of The Straits Times on April 01, 2016, with the headline 'Microsoft still set on AI bots after Tay fiasco'. Subscribe