Before You Share Medical Information with AI: The Dangers You Must Know

Millions consult AI for health advice, but a KFF survey shows a third of US adults have done so. OpenAI expanded the option to upload medical records to ChatGPT. Experts warn of biases, inaccuracy, and privacy issues, and studies found errors in diagnosing cancer and life-threatening conditions. The recommendation: use it only as a tool, not a substitute for a doctor.

In the technological age, millions of people daily use artificial intelligence tools as a substitute for experts and doctors. A survey conducted in March by the KFF organization found that one in three US adults has consulted AI about their health. ChatGPT is used by 44% of American adults, and OpenAI reported that more than 300 million people use it weekly for medical questions. Last week, OpenAI announced the expansion of the option for US users to upload medical records to ChatGPT. However, many are concerned about privacy, and 41% of those who asked medical questions uploaded personal information. Experts warn of racial and gender biases, noting that the system is built to provide a pleasant answer even when wrong. A study in JAMA Oncology found that ChatGPT provided inappropriate cancer treatment recommendations in more than a third of cases, and a study in Nature Medicine showed that the system underestimated the severity of life-threatening conditions. The overarching recommendation: use the technology only as an assisting tool, not as a substitute for a doctor, and seek immediate medical care in an emergency.