GENERATIVE AI AGENTS FOR MULTILINGUAL ACCESS TO GOVERNMENT AND COMMUNITY SERVICES

Abstract

Intelligent and adaptive communication technologies are needed to ensure equal access to government and community services by the increasing number of people who require to speak more than one language. The architecture demonstrated in the work is generative AI-based and is supposed to ensure a seamless process regarding both text and speech interaction. The algorithm corrects and sanitizes what users spell out, identifies what language the user is typing and within a few seconds, the text based speech query is translated into text. It extracts relevant information based on orderly data sets and indexed government portal and applies the information to generate relevant responses based on the fit of the situation. To build an LLC-fine-tuned and multilingual structural payload, a context builder employs information that is available to it, including user purpose, facts received, and profile information. Therefore, the generative agent produces brief, sequential instructions which are supported by source references which can be located. Text-to-speech synthesis allows providing the information in any language and postprocessing ensures that the answers are safe, facts are consistent, and linguistic and ethical standards are observed. They make the system more inclusive and trustworthy by having optional analytics modules that monitor the performance of the system, pattern of queries, and language distribution. This design is multilingual and creatively intelligent to ensure that all people have equal access to the tools they require, the tools are made more open and the language barriers are eliminated.