Health Questions Need Health-Specific Chatbots

AI chatbots are changing how people search for health information, but when the topic is health, a confident answer is not always a safe one. In this episode of AUB@Work, Imad Elhajj, professor of electrical and computer engineering at the American University of Beirut, and Aline Germani, director of AUB’s Center for Public Health Practice, discuss why general-purpose AI tools can fall short when answering health questions and what it takes to build safer, more trustworthy health information chatbots.

The conversation focuses on HIBA, a health information chatbot developed at AUB with carefully defined limits, curated sources, and ongoing public health review. Elhajj and Germani explain the difference between general health information and medical advice, why AI tools must know when not to answer, and how developers can design chatbots that better serve real users without replacing clinicians.

In this episode:

• Why general-purpose chatbots like ChatGPT or Gemini are not ideal for health-related questions
• The difference between health information and medical advice
• How misleading chatbot answers can contribute to confusion, anxiety, and health misinformation
• Why health chatbots need clear guardrails, curated sources, and ongoing testing
• How AUB developed and evaluated HIBA for a local audience
• What users should ask before trusting an AI tool with a health question
• Why AI literacy should begin early, especially when answers sound fluent and authoritative

Featured guests:

Imad Elhajj

Professor of Electrical and Computer Engineering, American University of Beirut

Aline Germani

Director, Center for Public Health Practice, American University of Beirut

Key takeaway:

AI can help people better understand health topics, but it should not be used to diagnose, treat, or decide whether someone needs care. The safest tools are the ones designed for health, grounded in reliable sources, and clear about their own limits.