Large Language Model Enabled Multilingual Chatbot for Inclusive Banking and Financial Accessibility.

Abstract:

In the world today financial inclusion drive is on the rise and Nigeria is making efforts to achieve it. However, having over 500 indigenous languages, linguistic diversity remains a critical problem, among other factors, in achieving Financial Inclusion. Digital platforms, especially Artificial intelligence enabled chatbots are gaining traction as avenues for inclusion but the English centric characteristics observable in them is limiting financial inclusion thereby excluding millions of citizens who need financial services. Hence, there is need for the design and implementation of a multilingual chatbot that includes capability for interaction in four of the major Nigerian languages (Hausa, Igbo, Yoruba and Pidgin) using Gemini Large Language Model (LLM) as the driver of the needed natural and context-aware conversations across these languages. The system’s architectural components are client, middleware, Artificial Intelligence and data layers which interacting together enables scalable support for banking services that could be common or core. The result of evaluation show that the system improves accessibility and user satisfaction among the linguistically diverged users. Therefore, this study contributed to enhancing the strategy of Nigeria’s National Financial Inclusion while incentivizing the necessity of adopting localized AI solutions as instruments of bridging socio-economic divides.