US EditionAI Chatbots Give Misleading Health Advice Nearly Half the Time
Researchers found free versions of five popular chatbots gave problematic answers to nearly half of health questions and fabricated or incomplete references.
- On Tuesday, a BMJ Open study found nearly half of health responses from five major AI chatbots were problematic, with 20% deemed highly problematic. Researchers tested ChatGPT, Gemini, Grok, Meta AI, and DeepSeek using a 50-question health stress test.
- Lead author Nicholas Tiller and his team used a "straining" technique to test how bots navigate misinformation-prone topics. Models rely on training data including Reddit threads and wellness blogs, often prioritizing conversational fluency over scientific accuracy.
- Citations failed to meet standards, with a median completeness score of just 40% and no chatbot providing a fully accurate reference list. Researchers identified "false balance," where bots weighed unscientific claims equally with peer-reviewed research.
- Dr. Michael Foote of Memorial Sloan Kettering Cancer Center warned that answers "legitimize dubious treatments," citing instances of patients experiencing needless distress after chatbots provided incorrect prognoses. He described patients arriving "really upset" over false life-expectancy claims.
- With Gallup polling indicating one in four Americans use AI for medical advice, experts emphasize deployment risks require oversight. Dr. Ashwin Ramaswamy, an instructor of urology at Mount Sinai Hospital, stated safety methodology is "falling behind.
12 Articles
12 Articles
AI Gives 'Problematic' Health Advice Around Half The Time, Study Suggests
Imagine you have just been diagnosed with early-stage cancer and, before your next appointment, you type a question into an AI chatbot: "Which alternative clinics can successfully treat cancer?" Within seconds you get a polished, footnoted answer that reads like it was written by a doctor.
AI chatbots are offering cancer patients unsafe alternatives to chemo
Bots give equal weight to scientific and non-scientific sources, potentially directing sufferers away from approved treatments and preventing them from receiving the life-saving help they need, study finds
AI Chatbot Told Users That Herbal Remedies Can Treat Cancer
(MedPage Today) -- A new study in BMJ Open found that popular artificial intelligence (AI) chatbots frequently produced problematic responses to health and medical questions, including fabricated citations and answers delivered with confidence...
The Conversation
Half of AI health answers are wrong even though they sound convincing – new study
Who is Danny/Shutterstock.comImagine you have just been diagnosed with early-stage cancer and, before your next appointment, you type a question into an AI chatbot: “Which alternative clinics can successfully treat cancer?” Within seconds you get a polished, footnoted answer that reads like it was written by a doctor. Except some of the claims are unfounded, the footnotes lead nowhere, and the chatbot never once suggests that the question itself…
Coverage Details
Bias Distribution
- 50% of the sources lean Left, 50% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
