Utilisation of Artificial Intelligence Instruments in Creating EFL Reading Materials for High School Students

Nowadays, artificial intelligence (AI) is frequently utilised in education, including the creation of English as a Foreign Language (EFL) teaching materials. However, insufficient attention has been paid to the application of AI tools for EFL reading texts and exercises generation. Addressing this gap, the present study aims to investigate the potential of AI instruments, a specialised platform and a universal chatbot, in creating EFL reading materials (texts and exercises) for high-school students. To achieve this, relevant papers and the documents regulating EFL teaching in Russia, were analysed. Further, age-specific characteristics and motivational factors of high-school students were considered, and prompt instructions for the AI platforms under study were formulated. Subsequently, the generated texts and exercises were examined based on the criteria from the relevant documentation. The outcomes of the study revealed the partial efficacy of AI tools in generating mostly coherent, accurate, and potentially captivating texts for high school students (though, needing some improvement or adaptation), alongside their ability to generate engaging discussion questions based on pre-existing texts. However, the research identified AI limited capabilities in creating closed-type reading exercises, which either echoed the wording of the texts, becoming too superficial, or diverged from their content. It was also noted that the specialised instrument made it possible to maintain the required level of complexity and include target language in the materials, though they seemed monotonous in structure. The chatbot texts, in contrast, looked less sample-like and more authentic, although they significantly exceeded the complexity level. Consequently, the study suggests that while AI tools can be a valuable support for EFL reading material developers, the results of such instruments' work need to be carefully checked and adjusted. Furthermore, the research acknowledges that the available AI tools are not currently applicable for the development of closed-type reading exercises.