Let’s Talk About AI Health Advice – by Dr. Scott Cohen, Chief Medical Informatics and AI Officer

Whenever we pick up our smartphones or log on to a computer we are met with artificial intelligence (AI). Chatbots, search engine result summaries, and other tools can make information and advice more accessible than ever before.

According to OpenAI, the organization that created and operates ChatGPT, more than 40 million people go to that chatbot each day with health-related questions. Just think of how many questions and responses that adds up to each week, month, and year. When you add in what are likely similar numbers for other chatbots, the number of inquiries is nothing short of astronomical.

While this technology can certainly be useful, it's important to keep perspective, view anything generated by AI with a critical eye, and remember this technology has limitations. Children and teenagers - age groups that can lack the life experience and knowledge needed to question what a chatbot tells them - are often adopting AI faster than older people.

Let's talk about how you and your loved ones can responsibly use AI to get medical information while not putting your health at risk by mistaking it as a replacement for a human healthcare provider. If you are a parent or guardian, it is important to have conversations with children about their AI use to ensure they are staying safe and feeling supported by the real people in their lives.

What Chatbots Can Do Well

AI chatbots can summarize information from multiple sources across the internet far quicker than any of us could if we were researching a topic on your own. They also allow users to easily ask follow-up questions to drill deeper and explore a topic fully, in a format that mimics a conversation.

They can be helpful for general questions about a healthcare topic from an educational standpoint. Questions or commands that are grounded in facts, rather than emotions or opinions, are easier for the technology to produce accurate responses for.

You can also use an AI chatbot before a medical appointment to get the most out of your time with your practitioner. For example, if your primary care practitioner has referred you to a specialist and you don't feel prepared to ask questions about a topic, ailment, or condition you are unfamiliar with, spending some time researching with the help of AI might spark some questions that you can bring to your appointment to have a more meaningful conversation. In the same way, AI can be used after a medical appointment. If you don't understand a term or phrase in any test results or messages from your medical team, asking AI can help you understand your practitioner or results better. Feeling confident to actively engage with your provider will always make your appointments better.

The American Medical Association put together a helpful guide containing five prompts that can be used to great effect, to get the information you want.

Where Chatbots Can Fall Short

One of the major downsides to using a chatbot is that the answers you receive will only be as good as the prompts you enter. Even slight changes in phrases (like describing a headache as bad vs. describing it as the worst headache you've ever experienced) can dramatically alter the responses you receive.

In a study published this year in the journal Nature Medicine, researchers acting as if they were average people turned to a chatbot for medical advice to see if it could accurately identify their condition and recommend the correct course of action. The chatbot was only able to correctly identify the condition described by the researcher in about a third of all cases. In less than half of all conversations, the chatbot recommended what a doctor would as the next steps (like whether to go immediately to the emergency room or stay home).

When you bring medical concerns to a healthcare professional, they ask you questions, based on their training, expertise, and professional standards. These questions are guided by what you've told them, the results of any tests they've conducted, and your personal medical history. This conversation helps your practitioner determine your diagnosis or the next steps that should be taken to help you feel better. Your caregivers have the uniquely human ability to apply reason and approach situations with kindness, and empathy - things you can never receive from text generated by AI.

The bottom line: chatbots and search engine results should never be used as a substitute for a visit to your primary care practitioner's office, a Convenient Care location, or emergency room. Delaying or forgoing care can have dangerous, life-altering or even life-ending consequences. As an example, a pastor from Florida recently filed a lawsuit accusing ChatGPT of downplaying early warning signs of what was eventually diagnosed as a life-threatening pulmonary embolism.

Let's Talk About It

Artificial intelligence is evolving quickly. In the short amount of time between this article's publication and the moment when you are reading it, there will no doubt have been new advancements in the technology and more people opening up a chatbot for the very first time.

AI cannot replace human connection, but it can be a powerful tool to help us save time and do more. In fact, at your most recent appointment, your healthcare provider may have mentioned that they are using it themselves to tackle some of the notetaking and other documentation demands of their job so that they can devote their focus more fully to you, the patient.

If you go to a chatbot or type a health-related question into Google and are unsure about the results you receive, come talk to us. Do not treat what you see online or in an app as the be-all and end-all when it comes to your health - your life is too important to trust to an unfeeling algorithm.