Why AI health chatbots won’t make you better at diagnosing yourself – new research
Why AI health chatbots won’t make you better at diagnosing yourself – new research
Publish Date: 2026-03-31 10:52:00
Source Domain: theconversation.com
- Millions of people are using AI chatbots for various types of advice, including health-related questions.
- The UK’s chief medical officer has warned against relying on chatbots for medical decisions due to potential risks.
- A study tested the effectiveness of large language model (LLM) chatbots in addressing common health problems, finding they are not ready to act as doctors.
- Participants using chatbots were less likely to identify the correct condition and determine appropriate care compared to those without chatbot interaction.
- Although LLMs can pass medical exams, their real-world performance in patient consultations is limited due to communication issues between humans and machines.
- The study emphasizes the complexity of medical consultations, which involve interpreting patients’ stories and fostering trust, skills not easily replicated by AI.
- AI’s current practical applications in healthcare are more akin to that of a medical secretary, managing tasks like organizing information and summarizing documents.
- The role of AI in medicine is expected to be supportive rather than transformative in the near future, with machines not yet ready to diagnose or direct patient care.
