The new AI-chatbot powered Bing is throwing off interesting, and disturbing, interactions with users.

Microsoft’s (NASDAQ: MSFT) release of a new version of Bing search engine that includes the artificial intelligence (AI) of a chatbot has led to the company considering tweaks and guardrails. The tech giant is looking at adding tools for users to restart conversations and limit conversation lengths before they veer into strange territory.

Long chats could confuse the chatbot, and it could pick up on its users’ tone, sometimes turning testy. Microsoft is considering limiting the chatbot’s ability to engage in open-ended and probing personal conversations, and one method would be to restrain conversation lengths. (NYT)

Although the company knew the new technology had occasional accuracy problems, users flagged surprising and unnerving interactions with it.  Researchers who specialize in AI, however, were aware of this possibility. By way of caution, Microsoft gave just a few thousand users access to the new Bing, but it said it planned to expand to millions more by the end of the month.

The company is now considering more guardrails for its new AI chatbot, attempting to reel in some of its scarier and eerily humanlike responses. In doing this, Microsoft is taking a calculated risk by trying to control the technology as much as it can be controlled, according to Oren Etzioni, professor emeritus at the University of Washington and founding chief executive of the Allen Institute for AI, a prominent lab in Seattle.

The Bing search tool combines its search engine with the underlying technology built by OpenAI, in which Microsoft has invested $13 billion. Releasing it was a critical example of Microsoft’s “frantic pace” to incorporate generative AI into its products.

Microsoft’s experience nearly seven years ago with Tay, a chatbot, was an example of what not to do. Users almost immediately found ways to make it spew racist, sexist and other offensive language. The company took Tay down within a day, never to release it again.

Unsurprisingly therefore, much of the training on the new chatbot was focused on protecting against harmful responses or scenarios that invoke violence. Microsoft developed a new way to use generative tools to identify risks and train how the chatbot responded.

The company is enthusiastic about AI; however, it is considering more limitations for its new AI chatbot to ensure as far as possible that users have a positive experience.

It is a learning process, however. “One area where we are learning a new use-case for chat is how people are using it as a tool for more general discovery of the world, and for social entertainment,” Microsoft wrote in a blog.

Related Story:  Microsoft Extends Partnership, Invests $10B In ChatGPT

Image by Gerd Altmann from Pixabay