Improving User Experience for AI Chatbots through LLM Models

This paper introduces an innovative strategy for enhancing the user experience via an AI-powered FAQ chatbot for the Ecole Supérieure en Sciences et Technologies de l'Informatique et du Numérique (ESTIN). By combining multiple Large Language Models (LLMs) such as BART, GPT, and BERT, the standard static FAQ system is converted into a dynamic, interactive interface. Our hybrid approach aims to expose the unique strengths of each model through a multi-phase training process. Initial experimentations have revealed that BERT outperforms other models in handling ESTIN's FAQ dataset, providing better context comprehension and response relevance. We intend to additionally refine the chatbot by studying the effect of different model training sequences. The expected outcome is a scalable and adaptable chatbot that significantly improves user engagement.