WHO chatbot provides inaccurate health information
A new World Health Organisation chatbot got off to an inauspicious start by getting basic health facts wrong, according to a media investigation. SARAH (or Smart AI Resource Assistant for Health) was trained on OpenAI's ChatGPT 3.5 large language model to provide information across major health topics, including mental health, tobacco and nutrition', in eight languages. However, the bot fails to provide up-to-date information on US-based medical advisories and news events, according to Bloomberg . In one instance, SARAH replied to a prompt by Bloomberg journalists by saying Lecanemab, an Alzheimer’s drug, was still in clinical trails when in reality the US Food and Drug Administration had approved the drug in 2023 . The incident raised concerns about the prospect of a United Nations agency providing health and mental health disinformation. The UN agency acknowledged on the chatbot landing page that 'answers may not always be accurate because they are based on patterns and probabilities in the available data.' System 🤖 Sarah chatbot 🔗 Documents 📃 WHO (2024). Meet S.A.R.A.H. - A Smart AI Resource Assistant for Health WHO (2024). WHO unveils a digital health promoter harnessing generative AI for public health Operator: Wo rld Health Organisation (WHO) Developer: Soul Machines Country: Global Sector: Health Purpose: Provide health information Technology: Chatbot; Facial recognition Issue: Accuracy/reliability; Mis/disinformation; Safety
- Date it happened
- 2024-04-01
- Organisation involved
- World Health Organisation
- Product, system or model
- SARAH
This incident was imported from AIAAIC and is used under CC BY-SA 4.0. Our additions to it — the structured fields, the translation, the checks against other reports — are published under the same licence.
This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.