Oxford Study Flags Pitfalls of AI Health Chatbots (Full Transcript)

[00:00:00] Speaker 1: Have you ever used an AI chatbot for health advice? A new study from the University of Oxford says it's risky. Not only because the advice could be wrong, but the study found people found it harder to communicate to AI what the problem was, too. The research split nearly 1,300 online participants into two groups and gave them symptoms such as a severe headache. One group could use AI to help them figure out their condition and what to do next. The other had to search online. Researchers found the people who used AI often didn't know exactly what to ask to get the right results. And the people often found it hard to tell what information given by AI was useful and what wasn't. Major AI firms such as OpenAI and Anthropic have both released versions of their chatbots dedicated to health to try and improve these results. Last year, polling by Mental Health UK suggested more than one in three UK residents now use AI for support.