Over half have engaged with AI technologies for diverse reasons; 14% have utilized AI for health information
The AI phenomenon hasn’t skipped over older individuals, a recent survey indicates, yet those over 50 exhibit both intrigue and doubt regarding artificial intelligence, influenced by their usage experience.
Regardless of their familiarity with AI, a significant majority of individuals aged 50 and above (92%) express a desire to know when the information they encounter—be it read, viewed, or heard—is generated by AI, according to the latest results from the University of Michigan National Poll on Healthy Aging.
In total, slightly over half (55%) of those aged 50 and above have interacted with an AI technology they communicated with via speech or text for various purposes on different devices.
Among participants who engaged with voice assistants like Amazon Alexa, Apple’s Siri, or Google Nest during the past year, 80% reported that these devices are advantageous for assisting them in living independently and securely at home, including 28% who claimed they are very helpful for this purpose.
Additionally, 35% have employed AI-enhanced home security systems; nearly all respondents indicated they felt these tools were beneficial in helping them maintain independence and safety in their residences.
Overall, 14% of seniors surveyed stated they have turned to AI technologies for health-related information. Of those, nearly half (47%) believed that a personal interaction—either face-to-face or via phone—would be more effective for obtaining such information, while 26% felt that AI and human interaction were of similar quality.
The survey is conducted by the U-M Institute for Healthcare Policy and Innovation and is backed by Michigan Medicine, U-M’s academic medical institution. The polling team also assessed data from adults aged 50 and older in Michigan, with support from the Michigan Health Endowment Fund; those findings can be found at https://michmed.org/KDBJR, with an interactive data dashboard available at michmed.org/GQkmw.
“AI is here to stay. Many older adults appear to recognize its advantages, yet a majority desire more information about the possible risks associated with AI technologies,” stated Robin Brewer, assistant professor at the U-M School of Information and a health AI researcher who collaborated with the polling team on the report. “The almost universal interest in clearly identifying AI-generated information should be acknowledged by policymakers and the AI sector.”
She noted that numerous state legislatures have recently enacted bipartisan laws to regulate AI usage, particularly in political advertising and mental health care.
Interest, attitudes, and confidence levels related to AI
In addition to inquiring about their AI experiences thus far, the poll sought to understand older adults’ interest in using AI and their awareness of its risks, their trust in AI-generated content, and their confidence in recognizing incorrect AI-generated information.
Slightly over a third of adults aged 50 and older (35%) expressed interest in utilizing AI in their everyday lives. More than half (58%) wish to learn more about the potential advantages of employing AI, but an even larger proportion (81%) indicated a desire for greater awareness of the risks involved.
Older adults are split on whether they believe AI will produce more harm than benefit, with 53% thinking it will do more harm, while the remainder disagrees.
Nearly half of those surveyed reported having some level of trust in AI-generated information (49%), though only 4% expressed a high level of trust, and the remaining 47% indicated they have little to no trust. Those who had experience using AI were more inclined to show some degree of trust.
Furthermore, half of older adults indicated they are not confident in their ability to recognize when AI-generated content is incorrect. Notable instances involve “hallucinations,” where an AI program generates plausible-sounding but false information, or the “deepfake” videos and images that have emerged over the past year. Scammers utilizing AI-driven voice technology have also been reported attempting to defraud older adults via phone calls.
“This challenge in differentiating verified information from misinformation—whether it appears as text, image, video, or audio—is particularly critical concerning health information,” remarked poll director Jeffrey Kullgren, a primary care physician at the VA Ann Arbor Healthcare System and associate professor of internal medicine at U-M.
“Our recent data shows that individuals who describe their physical or mental health as fair or poor are more likely to feel uncertain in their capacity to identify inaccurate AI-generated information and display lower trust in AI-generated health information than those in good to excellent health. Clinicians, aging-focused organizations, and healthcare providers should take note.”
Increased outreach and educational initiatives by these organizations and government entities could be beneficial, noted Brewer, who examines how technology can be designed to accommodate the needs of older adults and individuals with vision impairments.
“These insights provide avenues for older adults and their families to understand ‘healthy’ AI usage, including when it is appropriate to utilize AI instead of a search engine, or how to leverage AI for supporting healthy aging, like accessing non-urgent health information,” she stated.
The University of Michigan’s Michigan Online platform offers free and paid online educational opportunities regarding artificial intelligence, which include free mini-courses on generative AI and ChatGPT. More online educational content about AI is currently being developed; anyone can subscribe to receive updates.
Discover more about healthcare AI research and innovation conducted by U-M faculty who are affiliated with IHPI.
The survey results derive from a nationally representative study executed by NORC at the University of Chicago for IHPI, administered online and via phone in February 2025, encompassing 2,883 adults ages 50 to 97 throughout the U.S. The sample was subsequently adjusted to reflect the U.S. population. Read previous National Poll on Healthy Aging reports and learn about the poll methodology.