**Researchers Selected for Gates Foundation’s Program to Address Healthcare and Social Causes**
The Gates Foundation has announced the selection of 48 research groups to participate in its new program, which aims to develop AI-based applications to address crucial healthcare and social causes in low-and-middle-income countries. Each of the selected groups will receive a grant of $100,000 to support their projects.
**Diverse Applications for AI-Based Solutions**
The grantees are working on a wide range of issues, highlighting the versatility of AI in tackling various challenges. Some examples include researchers developing a ChatGPT-based chatbot to create and manage electronic medical records for maternal healthcare workers in Pakistan, and an entrepreneur working on an AI tutoring tool to provide specialized education to students in Kenya.
**Focusing on Healthcare and Local Issues**
While many of the grantees are utilizing AI for healthcare issues such as HIV risk assessment, prenatal care, and antibiotic prescriptions, there are also projects that aim to address local issues. For instance, one group plans to use AI to provide farmers in Uganda with information about crop diseases in their native language. Another researcher in Vietnam is developing a chatbot to provide advice to residents affected by saltwater intrusion, using data in Vietnamese. Additionally, a non-profit organization in Brazil aims to use AI to develop a support bot for psychologists and lawyers assisting women who have experienced gender-based violence.
**Expanding AI Development Globally**
The initiative by the Gates Foundation seeks to promote the development of generative AI across the world, as the majority of major AI companies are currently located in the Global North. The goal is to ensure that people from all regions can benefit from AI technology, mitigating existing patterns of discrimination, inequality, and bias.
**Improving Access and Addressing Biases**
One challenge faced by the researchers is making generative AI more accessible to a diverse population. Chatbot applications that rely on text-based prompts can exclude non-English speakers and those without smartphones. To address this, some researchers plan to develop a feature that converts spoken language into text, making AI more accessible to a wider audience.
Additionally, the researchers are aware of the limitations and potential biases of models like ChatGPT, which are trained on vast amounts of public data. To tackle this issue, the Gates Foundation has created a support hub consisting of global AI experts who will assist the grantees in evaluating and addressing potential risks.
**Selection Process and Project Details**
The selection process involved reviewing approximately 1,300 proposals from researchers, nonprofit organizations, and private companies across 103 countries. Critical criteria included conducting the work in a low-and-middle-income country, focusing on a societal issue, and utilizing a large language model (LLM) for the application. The projects are primarily utilizing and fine-tuning OpenAI’s GPT-4 and GPT 3.5, with a few projects incorporating other LLMs such as Google’s LaMDA and Bert, as well as a multilingual text-to-text model called mT5. The recipients have a deadline of three months to complete their projects.
**Building an Evidence Base and Implementation**
The Gates Foundation’s ultimate aim is not only to develop AI solutions but also to create an “evidence base” of generative AI use cases, challenges, and lessons learned. By doing so, the foundation hopes to determine how AI can effectively serve low-income communities and contribute to decision-making and implementation processes.