Dispensing Climate Change Information in Mother Tongue Using a GPT-4 Based a ChatBot: A Case with Mauritian Creole

Abstract

As the effects of climate change become more pronounced, residents of low lying coastal areas are experiencing these effects with often devastating results. Being able to dispense information about climate change to the residents of these areas in languages which they understand is important. Mauritius is an island nation in the Indian Ocean and has been in the path of multiple cyclones of increasing strength due to climate change. This paper looks at the possibility of using OpenAI's GPT-4 API connected to a mobile chat app such as Whatsapp in order to dispense information about climate change and adjacent topics such as the built environment in the mother tongue of the residents of Mauritius - namely Mauritian Creole. A chatbot was created using GPT-4 which was configured to be an expert in climate change and associated architectural issues and which communicated textually in Mauritian Creole. This paper reports on the language evaluation of this chatbot.