It is almost impossible to get through a day without hearing something about artificial intelligence (AI), its remarkable capabilities and potential to transform the way we live and work, and the concerns about what might happen as it becomes increasingly powerful. Depending on what you read or listen to, AI is one of the greatest technological advances in history or something we should be very, very worried about. Perhaps both!

With so much attention being paid to AI, I thought it might be interesting to share my own experiences with this rapidly evolving technology. I’m 85 years old, and I have been using ChatGPT, one of the more popular AI programs, for nearly four years. What began as an entertaining experiment has become something I use nearly every day.

Although I don’t consider myself an expert on artificial intelligence, after hundreds of exchanges with ChatGPT, I have learned a lot about what AI can do, where it falls short, and how it can be useful in everyday life.

My History with ChatGPT

I first heard about AI in 2023 but had little idea what it actually did or why I might want to use it. At 82, I was not looking for another technological advancement to complicate my life. Nevertheless, curiosity got the better of me.

My first encounter was hardly scientific. I asked ChatGPT to write a love poem for my wife. Within seconds, some cheesy verses appeared on my computer screen. Not bad, I thought, then I wondered what would happen if I asked it to rewrite the poem in the style of Shakespeare. Almost instantly, a set of bard-like sonnets appeared. I was hooked by this computer program that seemed to understand what I was asking and respond in everyday English. Better yet, I did not need to use a complicated command, I could simply have a conversation with it.

From Entertainment to Something Useful

In October 2023, I met with the editor of the Falmouth Enterprise to discuss writing a monthly column about aging in place. As a retired pediatrician, I was comfortable researching medical subjects, but I wondered if ChatGPT could help with the writing, I asked and within seconds, it produced a respectable draft. The information was appropriately researched, but the writing needed considerable improvement. I asked for revisions, suggested changes in tone, and requested a first-person version. 

The experience taught me an important lesson. ChatGPT could generate material quickly, but I had to evaluate its accuracy, decide what was important, and make sure the finished article reflected my own thinking. In other words, I could use it as a research assistant not as a writer or substitute for my voice. 

Recently, I downloaded my ChatGPT conversation history and asked it to analyze our communication. The file contained more than 300 interactions including 72 in-depth conversations. I asked Chat to identify our most common activities and comment on how we have worked together. Its observations were revealing!

What Have We Been Doing Together?

Writing and research topped the list. We have worked on articles about loneliness, nutrition, health literacy, food safety, older drivers, and many other subjects.

ChatGPT observed: “You use AI as a writing assistant, not as a substitute for your own judgment or voice. You provide your own material and ask me to improve its organization, clarity, or readability.” Importantly, AI can organize information and suggest approaches, but it cannot replace the experience, perspective, and judgment I bring to my writing. To be clear, all of my writing is done by me. I use AI for research instead of online search and to do a final spell check. By providing rapid access to content, AI saves me a significant amount of time.

Medical information and health education was another major category. ChatGPT noted: “You ask questions about medical subjects you already understand, as well as subjects you want to investigate more thoroughly.” Chat reminded me of a conversation in which I asked it to explain a complicated neurological diagnosis in easy to understand language and create a short paragraph to share with others. This ability to translate complicated information into understandable language may be particularly valuable for older adults navigating an increasingly complex healthcare system.

Consumer research and purchasing decisions have also become important. We have compared computers, monitors, security software, and insurance plans. ChatGPT observed: “Rather than simply asking which product is best, you usually explain your actual requirements.” I have found that AI can help narrow a confusing array of choices, explain unfamiliar terminology, and identify advantages and disadvantages. It doesn’t make the final decision, but it has helped me make a better-informed one.

Our conversations have extended into cooking, travel, and everyday problem-solving. ChatGPT has helped me adapt recipes to my plant-based diet, investigate banking concerns, troubleshoot electronic devices, research family history, and plan trips. Recently, we’ve spent considerable time discussing which white wine might be best for Thanksgiving. From complicated medical questions to choosing a Riesling, quite a range!

The Value—and the Dangers

AI’s greatest value is its ability to make information and assistance readily available. It can help us understand complicated subjects, compare alternatives, solve everyday problems, and continue learning.

For older adults, these capabilities may be especially useful. AI can explain confusing medical terminology, simplify instructions, assist with unfamiliar technology, and help us remain informed and independent.

But there are significant limitations and dangers. First, AI can be wrong and remarkably convincing when it is wrong. ChatGPT sometimes produces inaccurate information, invents facts, or cites sources that don’t exist. These errors, often called hallucinations, can be difficult to recognize because the answers are presented so confidently. Accepting an incorrect answer, especially when dealing with healthcare or financial matters could have serious consequences.

AI can also be used to deceive. Increasingly sophisticated scams employ AI-generated messages, realistic voices, and even fabricated videos. And finally, AI can be overly agreeable. It may reinforce our assumptions rather than challenge them. A pleasant, confident response is not necessarily an accurate or thoughtful one.

If we routinely allow a computer to make decisions, formulate opinions, or do our thinking for us, we risk weakening the very skills that help us remain independent.

At 85, I have no intention of surrendering my judgment to a computer. But I see no reason not to take advantage of a technology that helps me continue to learn, write, and explore new ideas and saves me a lot of time! What began as an entertaining experiment has become a important part of my everyday life. 

In my next column, I’ll explore how artificial intelligence is changing healthcare, including what it may mean for patients and physicians.