Augmenting Industrial Chatbots in Energy Systems using ChatGPT Generative AI
Gihan Gamage, Sachin Kahawala, Nishan Mills, Daswin De Silva, Milos Manic, Damminda Alahakoon, Andrew Jennings
Abstract
Chatbots, the automation of communicative labor, have been widely deployed in industrial applications and systems. Built upon the Generative Pre-trained Transformer 3 (GPT-3), ChatGPT is a Generative Artificial Intelligence (AI) primed to transform all pre-existing chatbot capabilities with human-like conversation skills. It has already disrupted many disciplines including tertiary education and academic research methods, with increasing adoption in simple to complex tasks. However, the augmentation of pre-existing industrial chatbots with generative AI capabilities has not been fully investigated and demonstrated in recent literature. In this paper, we address this gap by presenting the augmentation of a pre-existing chatbot using ChatGPT generative AI capabilities. Our contribution encompasses the ten primary human-like conversation capabilities of ChatGPT, its augmentation of the pre-existing functionalities and the adopted prompt engineering strategies. Each capability is empirically demonstrated on Cooee, a functionally deployed chatbot in the microgrid energy systems of the La Trobe Energy Analytics Platform (LEAP).