According to a new Poynter/University of Minnesota study about how the American public feels about generative AI in journalism, 49% of respondents said they aren’t at all interested in using chatbots to find information from news organizations, while 39% said they’d only use it if the responses were verified by the newsroom’s editors for accuracy.

The study, which surveyed 1,128 American adults in March, was presented at last week’s Poynter and AP Summit on AI, Ethics and Journalism in New York. “The data suggests if you build it, do not expect overwhelming demand for it,” Benjamin Toff, associate professor at the Hubbard School of Journalism, director of the Minnesota Journalism Center, who led the research, said at the Summit. (On LinkedIn, he said the University of Minnesota will release a more detailed report soon.)

Forty percent of respondents said they “never” use AI tools or chatbots for any purpose, and survey respondents were generally skeptical about the use of AI in news. Fifty-eight percent said newsrooms should only use AI with clearly established ethical guidelines and policies, while 20% said newsrooms should avoid AI completely. About 70% of respondents said it’s somewhat important or very important to disclose AI use in editing photos, writing article texts, translating stories, and creating infographics.

More than half of respondents (66%) already think news outlets use AI to write their articles at least “sometimes.”