Leveraging Al and Language Models for Social Impact: Supporting Marginalised Communities through Data-Driven Solutions

This research explores the implementation of artificial intelligence and large language models to empower marginalised communities while optimising policy implementation through enhanced named entity recognition in large-scale databases. Using Enter-link’s digital platform as the primary testing environment and incorporating data from the the Chinese/English Political Interpreting Corpus (CEPIC, https://digital.lib.hkbu.edu.hk/ cepic/), the study investigates how digital solutions can overcome barriers to inclusion and adapt to local needs and capabilities. Through action research methodology, the research employs iterative cycles of planning, action, observation, and reflection, with community action groups over a one-year period. Each three-month cycle begins with community engagement and needs assessment, progresses through AI implementation with continuous feedback on the Enter-link platform, and concludes with participatory monitoring and collective analysis. The methodology leverages databases to enhance pattern recognition and community-specific insights, while emphasising democratic validity through inclusive decision-making and maintaining ethical considerations through community ownership, data sovereignty, and sustainable empowerment. Success is measured through community adoption rates, service accessibility improvements, and communitydefined impact metrics, all tracked through Enter-link’s analytics dashboard. This approach ensures both technological solutions and research processes contribute to lasting social change while maintaining scientific rigour and adaptability to emerging community needs, with Enter-Link serving as both a testing ground and implementation platform, enriched by CEPIC’s comprehensive database.