C.A.L.I.X: Leveraging Hybrid RAG for Responsible AI in the Medical Field

Conversational AI memory systems work much like human memory. They record, maintain, and retrieve contextual information from earlier interactions, just as human memory draws on past experiences and learned knowledge to shape understanding and responses. However, individuals often face challenges with long-term recall of specific details from past events and short-term working memory, such as retaining lists, which can limit their ability to provide well-informed and nuanced responses. This research aims at implementing a simple Retrieve-Augment-Generate (RAG) system to implement a memory augmentation system with a particular focus on a use case in the medical field. This paper introduces C.A.L.I.X (Cognitive Archive for Learning and Information Exchange), an AI powered conversational memory assistant designed in alignment with Responsible AI principles. CA.L.I.X has been engineered to monitor conversations and extract data using voice interactions, ensuring that the information gathered is reliable. Future efforts will be focused on enhancing the accuracy of memory retrieval, streamlining the control of memory decay, and elevating the overall user experience.