The global rush to adopt AI for public health communication faces a critical test: does it work? A recent study in Kenya and Nigeria, analyzing 120 messages on vaccine hesitancy and maternal health, offers a surprising answer. Neither AI nor traditional methods proved superior.
While AI-generated messages showed more creativity and attempted cultural relevance—using local slang and metaphors—these references were often shallow or inaccurate. AI was also prone to errors, including distorted images. Traditional campaigns from health ministries and NGOs were more authoritative but often rigid, relying heavily on Western medical language and overlooking community knowledge.
This finding is crucial as African health systems rapidly adopt AI. For sensitive issues like vaccines and maternal care, trust and cultural understanding are paramount. The core problem both methods share is positioning people as passive recipients rather than active participants
The solution isn’t to abandon AI but to redesign it with local communities. This means training AI on local data, involving community members in message validation, and investing in homegrown AI platforms. The future of AI in health depends not on its intelligence, but on its ability to genuinely listen to and learn from the people it serves.
