This month, the World Health Organization (WHO) unveiled its latest initiative: a Smart AI Resource Assistant for Health, abbreviated as Sarah. However, despite its capabilities, the AI-powered chatbot has encountered some errors.
Sarah serves users with health-related information in eight languages, covering topics such as healthy eating, mental health, cancer, heart disease, and diabetes. Developed by New Zealand-based Soul Machines, Sarah incorporates facial recognition technology to offer more empathetic responses. Earlier versions of Sarah played a role in combatting misinformation during the Covid-19 pandemic.
In a press release, WHO Director-General Dr. Tedros Adhanom Ghebreyesus expressed optimism about S.A.R.A.H.’s potential to enhance access to health information in an interactive manner. He urged the research community to explore how such technology could address disparities and provide reliable health information.
While the WHO assures that the AI bot is trained with the latest information from the agency and its trusted partners, Bloomberg reports discrepancies. For instance, when asked about the FDA’s approval of Lecanemab for Alzheimer’s, Sarah erroneously stated it was still in clinical trials, despite the drug’s approval in 2023. Additionally, the bot struggled to provide details on a WHO report about hepatitis deaths until prompted to check the agency’s website for updated statistics.
The WHO has yet to comment on the matter. Acknowledging the limitations, the agency concedes on the AI bot’s landing page that “answers may not always be accurate because they are based on patterns and probabilities in the available data.“
