The head of Microsoft Research apologized Friday for comments made by the company’s chatbot, called Tay, saying that the company will keep it offline until it can better adjust Tay to deal with unexpected human interaction.
Microsoft’s offensive teen chatbot is so grounded
The company says it won’t bring its Twitter chatbot back until it is sure Tay can represent the best, rather than the worst, of humans’ online behavior.
Microsoft
In less than a day of interacting on Twitter, Tay had already begun spewing the racism, sexism and xenophobia it encountered.
“We are deeply sorry for the unintended offensive and hurtful tweets from Tay, which do not represent who we are or what we stand for, nor how we designed Tay,” Peter Lee said in a blog post. “Tay is now offline and we’ll look to bring Tay back only when we are confident we can better anticipate malicious intent that conflicts with our principles and values.”
Tay was designed to mimic the language and interactions of the millennial generation, while learning from those with whom it conversed.
As I pointed out in an essay, part of the problem is that Tay did too good a job of mimicking human interaction online. In order to blot out racism, sexism and other hate speech, Tay will have to be better than, not equal to, many of the human beings engaged in Internet conversation.
One interesting fact Lee notes in the blog post is that Tay was not Microsoft’s first publicly released Chatbot. Its XiaoIce chatbot is being used by 40 million people in China, apparently without any similar issues.
Microsoft said Tay’s behavior was the result of “a coordinated attack by a subset of people” exploiting a vulnerability in Tay’s coding.
“Although we had prepared for many types of abuses of the system, we had made a critical oversight for this specific attack,” Lee said. “As a result, Tay tweeted wildly inappropriate and reprehensible words and images. We take full responsibility for not seeing this possibility ahead of time.”
Microsoft said it hopes to learn from this experience as it moves deeper into artificial intelligence.
“We will remain steadfast in our efforts to learn from this and other experiences as we work toward contributing to an Internet that represents the best, not the worst, of humanity,” Lee concluded.
This article originally appeared on Recode.net.
