Microsoft’s Bing search engine is leaning further in to artificial intelligence, with a new test feature that summarizes search results for users. The move comes months after Google pumped the brakes on its AI summaries after the feature spread conspiracy theories and dangerous health advice.
The new feature, which Microsoft is calling Bing generative search, uses generative AI technology to create new results pages that provide summarized answers that are “easy to read and understand,” with links and sources listed alongside the generated text.
Read more: OpenAI Just Announced a Search Engine, SearchGPT, but There’s a Catch
“We’ve refined our methods to optimize accuracy in Bing,” Microsoft said while announcing the new feature on Wednesday and emphasizing that it’s still in testing. “We are slowly rolling this out and will take our time, garner feedback, test and learn, and work to create a great experience before making this more broadly available.”
There’s controversy around AI searching
Microsoft’s new Bing AI search feature arrives several months after Google faced widespread criticism over how it attempted to integrate a similar technology into its search results. In theory, summarizing answers to search queries makes sense, but it doesn’t always work as expected. After Google’s AI summaries launched in May, some users quickly noticed that the feature failed to discern fact from racist conspiracy theories about Barack Obama’s religion and birthplace. Google’s summaries also took as truth articles satirizing health advice.
Microsoft appears to have taken these issues into account with its new Bing AI feature, emphasizing that each of its summaries includes a source link people can use to click through to confirm and to learn more information. Microsoft has also said it’s making the results available to only a “small percentage” of user query results at first.
Though the future of Microsoft’s search function is still uncertain, the stakes for getting it right are clear. AI-powered tools and content are flooding the internet and finding their way into everything from emails and text messages to documents and presentations. Social networks including Meta’s Facebook and Instagram, ByteDance’s TikTok, and Google’s YouTube have begun building systems to detect and label posts that are made using AI tools. The tools don’t just force transparency on creators. These labels are also becoming an increasingly important anchor to help the rest of us maintain our shared understanding of reality.
There’s a long way to go. NewsGuard found that AI chatbots failed to provide “accurate information” nearly 57% of the time a week after the assassination attempt on former president Donald Trump. Ultimately, NewsGuard’s AI misinformation tracker found that AIs fell “far short in dealing with the wave of conspiracy theories quickly launched by critics and supporters of Trump, as well as by hostile foreign state actors.”
