Motivation
Generative AI tools are reshaping access to knowledge by lowering information costs and simplifying complex content. However, women in low-income settings remain at the margins of these advances due to limited internet access, low smartphone ownership, literacy constraints, and restrictive social norms. In Tanzania, for instance, only 24% of women report having mobile internet access, compared to 37% of men. These structural divides prevent women from benefitting from emerging AI-driven opportunities. Voice-based generative AI presents a promising approach to bridging these gaps. Building on earlier collaborations with Viamo, this study investigates whether such technology can expand access to critical health and economic information while strengthening women’s role in decision-making.
Objective
This randomized controlled trial tests the adoption and impact of Ask Viamo Anything (AVA), a voice-based AI assistant that delivers real-time responses in Kiswahili over basic mobile phones. A sample of 3,750 women platform users in Tanzania will be randomly assigned to one of five groups: (1) digital outreach emphasizing AVA’s value for health information, (2) the same health-related outreach plus encouragement to refer peers to Viamo, (3) digital outreach with an economic empowerment framing, (4) the same empowerment-related outreach plus encouragement to refer peers to Viamo, and (5) a control group. This design enables assessment of four hypotheses: whether digital campaigns drive uptake; whether health framing leads to stronger engagement than economic framing; whether referrals accelerate diffusion compared to word-of-mouth; and whether encouraging referrals reinforces use among referrers through social learning. Ultimately, this study provides one of the first experimental tests of how generative AI can be responsibly scaled in low-connectivity environments to support women’s empowerment.
Proposed Impact
This project will generate evidence on the promise and limits of voice-based generative AI in advancing women’s meaningful connectivity and agency. Findings will inform digital service providers, governments, and development partners on how targeted framing and peer referrals can support adoption and sustained use of AI tools amongst women. By documenting both usage dynamics and health-related outcomes, the study will contribute to academic literature on technology diffusion, gendered digital divides, and social learning, while also offering practical guidance to implementers seeking to deploy AI responsibly in low-income contexts.
