6 Ways to Stay Safe in a Chatbot World

For many adults over 50, artificial intelligence isn’t a futuristic concept anymore but a tool used in daily routines, whether to find health information, make social connections or to find ways to live independently and safely in their homes.

But understanding how to converse with ChatGPT, Claude, Google Gemini, Microsoft Copilot, Perplexity or other chatbots is imperative, and there needs to be more education for older adults, their caregivers and family members, says Robin Brewer, an associate professor of information at the University of Michigan.

Think of chatbots as computer programs that can simulate human conversation with an end user: a.k.a. us.

“Chatbots can be incredibly helpful for a variety of tasks that simplify the more traditional act of performing Google searches or feeding in multiple sources of information and asking it to review and summarize,” said Emily Laird, an AI integration technologist at the University of Wisconsin-Stout. “They have these core peak things that they’re very, very good at.”

Where using chatbots becomes problematic, Laird cautions, is when people fail to use their critical thinking skills and blindly trust the AI responses to be truthful or accurate.

“AI’s are not using reasoning based on human thought processing,” says Camille Banger, an assistant professor at UW-Stout’s College of Science, Technology, Engineering, Mathematics & Management. “Instead, it’s taking the data it already has access to.”

More troublesome is that chatbots may hallucinate — that is, sound very confident and convincing, even if an answer is wrong.

When using a chatbot, these simple tips can make a difference

Determine if you are communicating with a chatbot or a human. “Chatbots are sycophantic by nature,” says Laird, noting that they’re designed to keep you using them. The biggest giveaway: Chatbots tend to give a long-winded response to a simple question, and it can be full of “overly embellished flowery language” and “more descriptive words.”

“AI tends to explain what you’ve just said back to you before responding because it’s working through that natural language processing,” she said. Chatbots don’t follow the normal cadence of human response with short, simple answers or natural pauses, she added.

Vet AI-generated responses. Chatbots won’t always pull from a reputable source and, at times, will even make up a source to support their answer, Brewer says.

Always double-check answers and verify sources, especially when they relate to health, finance or legal matters. Consider AI responses as a starting point.

That is even more important if you’re a caregiver for a family member who may be experiencing cognitive decline, Brewer says, adding that you may also need to schedule weekly check-ins to review the chatbot’s browser history.