'Brilliant and dumb': Bryan Johnson, the millionaire who wants to live forever, on why not to trust AI with health

Tech entrepreneur and anti-ageing enthusiast Bryan Johnson has an advice for millions of people who are using artificial intelligence (AI) for health related advice saying it is still too unreliable. “We are in an in-between state where AIs are brilliant and dumb. They make really stupid mistakes all the time. I would never trust my health to an AI,” Johnson told The Axios Show. However, Johnson added that the technology could eventually lead to major breakthroughs in medicine and longevity. Johnson is the founder of the “Don’t Die” longevity movement and is said to have spent millions to extend his lifespan.

Johnson’s warning comes as AI chatbots increasingly become a source of health information for Americans. A Pew Research Center survey found that about one in four Americans say they have used AI chatbots to diagnose health symptoms. Mayo Clinic has also reported that nearly eight in 10 US adults turn to the internet for answers to health questions. At the same time, more than one in five adults globally are reportedly turning directly to AI chatbots such as ChatGPT and Gemini with health-related questions.

Johnson, meanwhile, hoped AI would find a treatment or potentially discover ways to slow or reverse aging in coming times. He was diagnosed with an incurable autoimmune disease earlier this year. The American millionaire is also training AI agents on his brain and personality, to keep some version of himself even after his death but hopes to continue living in his current physical form.

Experts warn over using AI for health advice

Medical experts have warned that using AI for health decisions comes with significant risks. Mayo Clinic says diagnosing and treating illness is too complex to leave entirely to a machine. AI tools may not have access to a person's complete medical history, cannot physically examine a patient or conduct medical tests, and may not be able to explain how they reached a particular conclusion, it warned.

The clinic also warns that AI can produce answers that sound convincing but are completely wrong. These errors, often called “hallucinations,” can occur because AI systems generate responses based on patterns in their training data rather than understanding medical facts in the same way a healthcare professional does.

AI-generated health information can also contain biases because the data used to train these systems may have gaps or may not adequately represent people from different cultures and populations.

Another concern is misinformation. AI systems can reproduce inaccurate or misleading information that exists online, potentially amplifying false health claims.

Mayo Clinic also points to research suggesting that people seeking health information may not provide AI chatbots with enough specific details.

Small differences in how a person describes symptoms can sometimes lead to dramatically different answers. In one example cited by the clinic, two people described similar symptoms using different wording, and the AI gave the correct recommendation to seek emergency care to only one of them.

Mayo Clinic has recommended using AI for general health education rather than diagnosis, cross-checking information with reliable medical sources and avoiding the sharing of sensitive personal information.