Tech giant Google and maternal health nonprofit Jacaranda Health have announced a fellowship aimed at transforming the potential of Large Language Models (LLMs) for low-resourced communities across Sub Saharan Africa. The fellowship program builds on a recent $1.4 million grant from Google.
Six Google technical experts have been chosen for the opportunity. It is a six-month program focused on tackling real-world AI challenges by applying their skills in research, machine learning, and software engineering.
“Having this support from Google is critical for our innovation journey at Jacaranda. The work we have planned with support from these Fellows will give us the springboard to significantly improve maternal and newborn health outcomes through the power of AI,” said Jay Patel, Director of Technology, Jacaranda Health.
From April to October, the six fellows will collaborate with Jacaranda’s teams to enhance Jacaranda’s technology to better serve mothers across Kenya and pave the way for expansion into new African countries.
This includes testing how to leverage AI and large, anonymized datasets to offer personalized support for high-risk mothers and babies, designing frameworks to manage biases related to gender and economics in Large Language Models, building the infrastructure to rapidly adapt PROMPTS to new contexts. Lastly, they will work on helping other implementers ‘plug and play’ the AI models in other sectors and use cases.
Jacaranda’s PROMPTS program, uses AI to deliver personalized support to Kenyan mothers via SMS. Recently, leveraging the learnings from this project, Jacaranda has created open-source AI models. This not only allows for scaling PROMPTS across Africa but also offers valuable tools for other organizations to improve their AI-driven services focused on maternal health.
Despite a rapidly-evolving landscape of AI technologies, few have been adequately adapted to low-resourced settings, limiting the potential of lifesaving AI-driven initiatives in sectors like health and education. Jacaranda is on a mission to change this by developing and deploying multi-language LLMs that small local teams across Africa can easily customize to their contexts and use cases.
