March 28, 2016
Microsoft’s AI chatbot “Tay” was attempting to learn how to interact with humans. Instead, it was hijacked by trolls and learned nasty, racist language. Will the Twitter fail adversely affect Microsoft’s brand?
Twitter trolls have claimed another victim.
Microsoft’s artificial intelligence chatbot “Tay,” which uses the handle @TayandYou, went offline almost as quickly as it began chatting with users. The bot was created to have quick-witted conversations with the 18 to 24 crowd on Twitter and lean how to interact with people on the platform over time, the company said.
“The more you chat with Tay the smarter she gets, so the experience can be more personalized for you,” Microsoft’s website for Tay said.
However, Tay’s last tweet was on March 23. Instead of learning human communication from your average Twitter account, Tay was spouting xenophobic, racist, and absurd sentiments from Twitter’s cadre of trolls. Tay tweeted that the holocaust was fabricated, agreed with white power slogans, used racial slurs, and admitted support of genocide.
“c u soon humans need sleep now so many conversations today thx <3,” Tay’s most recent tweet reads. A string of its most incendiary tweets have since been deleted.
On Friday, Peter Lee, corporate vice president for Microsoft Research, posted an apology on Microsoft’s blog. “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,” he wrote.
Lee wrote that the Microsoft team put Tay through multiple stress tests to ensure interactions with the bot would be a positive experience, but what he called a “coordinated attack” of people on Twitter exploited vulnerabilities in the chatbot.
Screenshot from Tay's Twitter account from Business Insider
“Although we had prepared for many types of abuses of the system, we had made a critical oversight for this specific attack,” Lee wrote. “As a result, Tay tweeted wildly inappropriate and reprehensible words and images. We take full responsibility for not seeing this possibility ahead of time.”
How does something like this affect Microsoft’s brand? Rob Frankle, an independent branding strategist and consultant who works with Fortune 1000 companies, says he doesn’t believe it will hurt it at all. In fact, Frankle doesn’t believe Microsoft had anything to lose by experimenting in this way, as he doesn’t believe they truly have a brand at all.
“[Microsoft] does some things really well, but the fact is that as a brand, most people don’t really have a lot of loyally to Microsoft,” he says. “They haven’t cultivated a brand. They have never had a brand; they have an identity. But in my experience, given availability and cost issues, most Microsoft people, if they can choose another solution, probably would. … They are a tactical company, not a brand company.”
In fact, this situation be end up as an opportunity for Microsoft, he says: It gave Tay a good shot and can use the Thomas Edison Defense. “If you ask Thomas Edison, he didn’t create the lightbulb the first time out; he discovered 10,000 ways not to invent the light bulb.” It shows that the company is trying something new, he says, and they’ve never been a company that is afraid to fail. Frankle points out the countless clips of former Microsoft CEO Steve Ballmer “getting all sweaty and jumping around like a nerd on stage and looking stupid” and Bill Gates’ famed blue-screen-of-death during Windows 98’s launch.
“There’s far worse stuff than this,” Frankle says with a chuckle. “This is sort of a technology thing. Everyone knows you can game technology.”
Karen Post, a branding expert and owner of Bran Tattoo Branding, says the Tay incident may be embarrassing momentarily for Microsoft, but innovators know that “launch and learn” is now a common practice in technology.
“Our society is forgiving and in many cases has a very select memory. The last brand blow up is quickly replaced with the latest brand blow up,” Post says. “The upside of this is … Microsoft is playing [in] the sand box, trying new things and aiming to better understand its customer and prospect. As long as it responds in a timely fashion, it will be fine.”
Marketing News spoke with Mary van de Wiel, a branding expert and CEO of Zing Your Brand, on Thursday before Lee wrote the apology blog post. Van de Wiel said she believes in making companies look more human by having more human dialog, and this was a fantastic opportunity for just that. She doesn’t believe Tay’s failure will negatively affect the brand, but said she hopes this would open up a greater conversation about where this kind of hatred and vitriol came from on social media platforms like Twitter.
“I would want my person at Microsoft to come out and talk about human responses, [and] create an amazing context for this kind of dialog,” she said. “Say ‘This is what we did and we could have waited until it was more fine-tuned, but it is powered by machine intelligence. Of course it’s going to be tricky.’ Maybe [Microsoft could] acknowledge that and instead of being embarrassed, say ‘We understand machine intelligence is not giving us what we need.’”
Frankle, like Post and van de Wiel, believes there is an opportunity to turn this into a positive branding experience,
“[Microsoft execs] should have some kind of focused effort on innovation and some sort of notion to the public about how they are a company that is always testing things, is always trying to develop things. Sometimes they work, sometimes they fail. When they succeed they’re amazing, and [expressing] that they’re open to the public’s reactions to certain things [is important],” he says.