St. Louis, Missouri
Tech-Powered
Cultural Diversity
St. Louis, home to 2.8 million people, is experiencing significant demographic shifts, with marked increases in its Asian and Hispanic populations as well as influxes of new residents from around the world. Accordingly, the city of St. Louis is continually improving its service provision in order to meet the needs of its evolving population. In light of this growing diversity, the city has established the Office of New Americans to streamline resources for new residents. However, the city has limited staff and systems to promptly handle the bureaucratic process needed to serve large numbers of new migrants. One of the most salient ways that these resources can be bolstered is through increasing culturally responsive translation services for city offices. In the following brief, a technology-powered approach to translation is explored. Given the breadth of language diversity in St. Louis, resources such as Artificial Intelligence (AI) and Large Language Models (LLMs) have the capacity to provide greatly expanded capacity and accuracy. Supported by human checkers, the city can develop a standardized approach to contextually accurate and responsible translation.
The team initiated the project by conducting a thorough literature review, focusing on AI ethics frameworks, government procurement standards, and policies related to AI use in public services, specifically for language translation. This research identified critical areas such as transparency, accountability, and bias mitigation, which were crucial for St. Louis's context. In collaboration with AI policy experts, city officials, and technologists, the team held biweekly meetings to refine these insights into a detailed policy framework tailored to the city's needs. This framework emphasized ethical AI procurement, implementation guidelines aligned with the San Jose GovAI Coalition and the White House’s Blueprint for an AI Bill of Rights, and robust maintenance protocols, including human-in-the-loop oversight. Additionally, the team developed a proof-of-concept for a legal document translation service using fine-tuned large language models, which provided practical insights into the challenges and opportunities of AI-driven translation services in St. Louis.
• Procurement Policy Guidelines: A concise set of guidelines for St. Louis to ethically procure AI systems, prioritizing vendors with U.S. federal government experience or comprehensive safety and ethics documentation and including a checklist to assess risks like data security and transparency.
• Implementation Policy Manual: A policy manual aligned with the San Jose GovAI Coalition and the White House’s AI Bill of Rights, detailing principles of fairness, accountability, and transparency for AI use and necessary training materials for city staff on ethical AI use, emphasizing transparency and responsible implementation.
• Maintenance and Monitoring Plan: A user-friendly feedback system for reporting AI issues, combined with human-in-the-loop protocols for continuous oversight. This plan includes quarterly audits to detect and address biases.
• Proof of Concept: A report on the development and testing of an AI-powered legal document translation service, showcasing practical applications and challenges. It includes recommendations for future improvements, such as using advanced language models and expanding the dataset.
• Final Internal AI Policy Checklist: A comprehensive checklist covering all critical steps for ethical AI procurement, implementation, and maintenance. This ensures that AI systems deployed in St. Louis are aligned with best practices and legal requirements.
The implementation of AI-powered language access services in St. Louis holds significant potential to transform how the city supports its immigrant communities. By integrating AI, the Office of New Americans (ONA) can streamline communication and provide more culturally sensitive, equitable, and scalable services. The immediate impact will be seen in enhanced accessibility for non-English speaking residents, empowering them to navigate city services more effectively.
Future work on language access AI in St. Louis can involve several key areas of development. The city could expand the range of languages supported by AI systems to better serve its diverse immigrant communities. Further research and pilot programs can explore advanced AI models tailored specifically for legal and administrative language translation, enhancing accuracy and cultural sensitivity. Continuous improvement of AI governance, including regular updates to policies and training programs, will be necessary to adapt to new ethical challenges and technological advancements. Additionally, the city could engage in partnerships with other municipalities and organizations to share best practices and refine AI deployment strategies. Expanding the scope of AI tools beyond translation, such as incorporating AI-driven community outreach and engagement initiatives, could further bridge language barriers and create a more welcoming and inclusive environment for all residents.
