“A piece of cake”: sustainable education practices with transformer models

Abstract

New advances in Artificial Intelligence (AI), particularly transformer language models, have disruptively changed the current education and training practices. The latter advancements have forced a shift towards a more practical and sustainable learning process with the ultimate objective of creating active, high-quality, and continuous educational services. However, human intervention is still essential to keep control over the outcomes of AI-based solutions. Thus, this work contributes with a fully functional and well-designed Conversational Assistant for Knowledge Enhancement (CAKE), which integrates a transformer model in a chatbot platform with empathetic capabilities. The proposed environment is fully immersive and incorporates gamification. Natural Language Processing techniques are exploited through prompts. The most relevant features of CAKE are its flexibility regarding language, areas and topics covered, and the education level of the end users. Results obtained in two evaluation scenarios endorse the performance of the proposed solution and motivate us to continue this research line toward real-time monitoring of smart learning.