AI chatbots may help older adults with health questions — if they stay cautiously skeptical

(ANN ARBOR, MI) Older adults are often seen as reluctant or inexperienced technology users, but many people age 60 and over are increasingly turning to artificial intelligence for entertainment, travel planning and even health information.

I research how people use computers and technology, especially how older adults interact with voice assistants and AI chatbots such as Amazon Alexa and ChatGPT to find information. Using AI for health questions can be especially tricky: these tools can make medical information easier to understand, but they can also be dangerous if people accept every answer without question.

The best approach, my research suggests, is informed skepticism — staying open to what an AI chatbot offers while also checking its credibility and quality with the tools built into the system.

Why AI is appealing

About one in five adults say they use AI chatbots to ask health-related questions, according to a 2025 survey of 300 adults. Those questions may involve symptoms or advice about diet and exercise. People may use mainstream tools such as Claude, Gemini and ChatGPT, or systems designed for a specific health provider.

There are many reasons people choose these tools. It can feel easier to raise sensitive issues with a chatbot, and the response is almost immediate — much faster than waiting weeks or months for a doctor’s appointment. Chatbots can also rephrase complicated subjects in simpler language and turn a flood of information into a concise summary.

Although older adults may be less likely than others to adopt AI, my own work has found that many use voice assistants like Siri and Alexa to ask health questions. A 2025 poll I conducted with colleagues also found that half of older adults reported using AI chatbots or voice assistants, and about 14% said they were using them to access health information.

Older adults often have more health needs than younger people because of chronic conditions or new diagnoses that come with age, which naturally leads to more questions.

AI is especially attractive for health topics because it can provide answers more quickly than searching across several websites or waiting for a medical visit. Using a chatbot or voice assistant at home can also give people a stronger sense of control over their care, especially when clinical settings leave older adults feeling overlooked or shut out of conversations in favor of caregivers.

What can go wrong?

Even though older adults may use chatbots to supplement or even replace medical professionals, there are real risks.

Chatbots can give wrong advice in emergencies, misidentify conditions, or be overly agreeable by telling users what they seem to want to hear even when it is inaccurate. AI specialists refer to that last behavior as “sycophancy.”

Those mistakes can be hard to spot. AI chatbots rely on huge datasets and pattern matching to predict the next word in a sentence, so their answers may sound convincing while still being incorrect.

Most major chatbots display a warning below the message box saying their responses may include errors, but the notice is usually small or faint and easy to miss. The message is meant to encourage users to verify information with published sources, but many do not. In the 2025 survey of 300 adults who used a health chatbot, four in five said they never checked the answer against another source.

In my research, I found that older adults using voice assistants preferred to begin and finish their search on the same device, which suggests they are less likely to verify chatbot answers if doing so requires switching formats. Younger adults, by contrast, tended to see AI as just one step in the search process and expected to look for more information elsewhere, such as with a search engine. That difference shows why older users need practical guidance for using AI.

Privacy is another concern. Public-facing AI companies often retain user prompts to train future models, so researchers generally advise people not to include personally identifying details when asking a chatbot questions.

Ways to search more safely

Awareness is growing that older adults may use AI differently from other groups and may face unique challenges. A bipartisan bill introduced in the U.S. Senate, the Aging with AI Act, highlights the need for more research into both the benefits and the risks.

My research suggests several practical steps:

Know how to turn off AI. While popular chatbots like ChatGPT get much of the attention, many people encounter AI through search engines, where AI features are often built into results pages. Some search engines, such as Microsoft’s Bing, require users to opt in to AI. Others, including Google Search, expect users to opt out.

Google places an AI-generated summary at the top of search results. To avoid AI content in a Google search, adding “-ai” to the end of the query removes AI-generated material. That may be especially useful when looking for sensitive health information you want to keep private.

Use quality-check features. Most chatbots include tools that let users see where information came from. For instance, Claude, Gemini and ChatGPT provide links to sources used to generate a response. Review those sources to make sure they are legitimate and to find more detail if needed.

You can also switch the model the chatbot uses, which may change how detailed the response is. Gemini, for example, offers “flash-lite,” “flash” and “pro” options in a drop-down menu next to the search box. More advanced models generally produce more thorough answers, which can help users judge the quality of the information.

Users can also flag inaccurate, toxic or harmful content by selecting the thumbs-down icon at the bottom of a response.

Ask sharper questions. Seek objective information that can be checked rather than subjective advice, especially in higher-risk health situations. For example, instead of asking which blood pressure medicine to take, ask about the side effects of a specific medication.

To reduce sycophancy, phrase questions neutrally. For instance, rather than asking whether it is a good idea to get a flu shot, ask about the pros and cons of flu shots.

Do not use AI instead of medical professionals. Health-related chatbot answers are best used alongside advice from a clinician. That could mean saving the chatbot’s response for a future appointment, using it to better understand a diagnosis, or asking it to help generate questions for a doctor visit.

Most important, any health-related guidance that could affect your routine should be confirmed with a licensed healthcare provider. A qualified professional is best able to explain how general information about illness or treatment applies to your specific situation.

Robin Brewer receives funding from the National Science Foundation.