Time to Hire a Robot Psychologist? Evaluating a Corporate RAG Application

In this work, our primary focus was on assessing the effectiveness, in terms of quality, of an AI Retrieval-augmented Generation application. After constructing this AI web application, we designed and generated a synthetic dataset consisting of questions and answers based on a set of documents. We then compared the performance of two distinct language models powering the application according to a predefined set of metrics when responding to the dataset's questions. In this project, we opted to utilize open Large Language Models and we run them locally without relying on any cloudbased service.