Generative artificial intelligence encompasses techniques for automatically generating content such as image, music, and text. A common application of these techniques, especially involving text generation, is the construction of conversational agents. In education, generative artificial intelligence has the potential to transform the learning process and enable new forms of teaching, such as the practical application of concepts studied in class and the acquisition of new knowledge through the interaction with computational tools. In Brazil, literature teachers often complain about students' resistance to the content and insufficient reading practice. To address these challenges, this work presents a serious game in which students can chat with Machado de Assis, an important writer of the Brazilian literature. The game was developed by integrating with ChatGPT. The validation was conducted with 15 high school students for them to interpret the book "Dom Casmurro". The MEEGA+ instrument was used to evaluate the usability and the experience with the game. The main findings were that students better understood the book, found the game easy to use, and felt immersed in the conversation with Machado de Assis, although they assessed the texts as long and tiring. In a world dominated by faster and visually stimulating forms of entertainment, such as video games and social networks, students may have different expectations regarding the reading experience. If they are not used to reading denser and more introspective texts, they may find it difficult to maintain interest when chatting with Machado de Assis. Finally, another implicit point in the game dynamics is observed: the (re)signification of the methodology to be adopted by the teacher, who will perceive the need to complement interactions with the writer in class according to what he can or should answer, tending to responses that stimulate students to study more about what is within the realm of knowledge about Machado de Assis.
Read full abstract