From code to classroom

a preliminary evaluation of a chatbot prototype for english language teaching

Abstract

This paper presents the development and initial testing of an educational chatbot prototype named Lexi, designed to support English language teaching in the Public School System of Sergipe, Brazil. The research is based on the assumption that the integration of generative artificial intelligence (GenAI) technologies, such as chatbots, still faces resistance in school environments, mainly due to the low level of digital literacies among many teachers. The theoretical framework draws on authors such as Castells (2023), Jenkins (2009) as well as recent studies on AI in education. The adopted methodology is Design Science Research (DSR), which focuses on the construction and evaluation of technological artifacts applied to real-world problems. The chatbot was developed using Python, Flask, Twilio, ngrok, and the OpenAI API, and was integrated with WhatsApp as its communication platform. The tests demonstrated technical feasibility, appropriate responses, and smooth interaction, although they also revealed operational limitations, such as dependence on a local server and the need for prompt optimization. It is concluded that, although still in an initial stage, the chatbot shows potential as a pedagogical tool, provided that its adoption is accompanied by critical teacher training and equitable infrastructure conditions. The study reinforces the importance of aligning technological innovation with educational intentionality and proposes new testing phases with public school teachers, aiming to improve the system and support its future implementation.