Microsoft considers more limits for its new AI chatbot

Publishing it — despite potential imperfections — was a critical example of Microsoft’s “frenzied pace” to incorporate generative AI into its products, he said. At a press conference at the Microsoft campus in Redmond, Washington, executives repeatedly said it was time to get the tool out of the “lab” and into the hands of the public.

“I feel like especially in the West there’s a lot more stuff like, ‘Oh my God, what’s going to happen because of this AI? ‘” Mr. Nadella said. “And it’s better to kind of say, ‘Hey, listen, is this really helping you or not?’”

Oren Etzioni, professor emeritus at the University of Washington and founding chief executive of the Allen Institute for AI, a top lab in Seattle, said Microsoft “took a calculated risk, trying to control the technology as much as she could be.”

He added that many of the most troubling cases involved pushing technology beyond ordinary behavior. “It can be very surprising how cunning people are to get inappropriate responses from chatbots,” he said. Referring to Microsoft officials, he continued, “I don’t think they expected some of the answers to be wrong when the chatbot was prompted in this way.”

To guard against problems, Microsoft only gave access to the new Bing to a few thousand users, although it said it plans to expand to millions more by the end. of the month. To address concerns about accuracy, he provided hyperlinks and references in his answers so users could verify the results.

The caution was informed by the company’s experience nearly seven years ago when it introduced a chatbot named Tay. Users almost immediately found ways to make her spew racist, sexist, and other offensive language. The company shot Tay in a day, never to release him again.

Much of the training on the new chatbot focused on protecting against this type of harmful response, or scenarios that invoked violence, such as planning an attack on a school.