Language barriers prevent millions from accessing critical services. While AI translation holds promise, generic models often fail marginalized groups. This presentation will demonstrate how human-centered datasets enable effective, responsible machine learning and generative AI applications. Speakers from the New Jersey Department of Labor and U.S. Digital Response will discuss the roles of user research, content strategy, and generative AI in producing accurate, simple unemployment insurance translations in New Jersey, making it easier for Spanish speakers to access benefits while reducing administrative overhead. We’ll share challenges, solutions, and best practices for multilingual data collection, curation, and model training, along with our vision for empowering public servants with emerging technologies to advance language justice in social safety net programs. Attendees will take away strategies for ethical applications of AI, open collaboration, and designing digital services that truly serve everyone.

Emerging Technology + Innovation