ReDDDoT Phase 1: Planning Grant: AI-Enabled Support Services for Transplanted Populations: A Community-Centered Design and Development Approach

Transplanted populations typically find themselves in a new place without appropriate placement matching and without any support for their critical needs. Ensuring the successful integration of these people in their host communities is a profound and critical humanitarian goal. Artificial Intelligence (AI) technologies have the potential of automating many aspects of the integration process, and to assist both the impacted people and the organizations working with them in a variety of tasks from translation services to information gathering and data collection, to targeted matching of people with appropriate resources and services. The responsible design, development, and deployment of such technologies, particularly when dealing with vulnerable populations, requires a community-based approach that integrates iterative feedback from multiple stakeholder groups. The goal of this proposal is the responsible development of an AI-based platform dedicated to improving such support services. Lessons learned as part of this framework's evaluation will be applicable to other domains where AI technologies are considered to address the social and economic needs of vulnerable populations. This project will develop a new community-centered design and development framework leading to the creation of a holistic AI-powered platform that integrates real-time translation services, information extraction capabilities, and a personalized recommendation system. This project has three core objectives: (1) Develop and pilot a process for identifying and engaging stakeholders across a variety of spheres, the organizations that serve such populations, community leaders, local government, and service providers, in a needs assessment and data collection process that will lead to the development of a personalized recommendation system that will match migrants with appropriate services and resources; (2) Develop and pilot a framework for iterative machine learning model design and development that engages stakeholders in the entire data pipeline, including evaluating information extraction, feature selection, parameter optimization, recommendation generation, and platform usability testing; (3) Based on the results of the work performed during the planning grant period, develop a proposal to scale the community-centered framework and fully develop and deploy a recommender system to match these populations with the relevant services. The community-centered design approach utilized in this project will ensure that critical system-level criteria such as group fairness, policy compliance, privacy preservation, transparency, and general usability are prioritized. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.