Uma jornada arqueológica para desenterrar como funcionam os LLMs

Abstract

This work proposes a gamified web platform for teaching fundamental concepts related to Large Language Models (LLMs), aiming to make their learning process more accessible, engaging, and motivating. The theoretical foundation covers gamification, the Octalysis framework, artificial intelligence, machine learning, deep learning, natural language processing, and the impact of LLMs on Software Engineering. This research is characterized as exploratory and applied in nature. It was conducted through a literature review, an assessment of the target audience's motivational profile using a questionnaire based on the Octalysis framework, statistical analysis of the collected data using Cronbach's alpha coefficient and core drive averages, requirements elicitation and prioritization, and the development of a prototype of the proposed solution. The results revealed a high degree of convergence among participants regarding motivational factors, identifying the predominant core drives as development and accomplishment, epic meaning and calling, empowerment of creativity and feedback, and unpredictability and curiosity. Based on these findings, ten gamification techniques were selected to compose the proposed platform, which is organized as an archaeological exploration journey for teaching LLM concepts. It is concluded that the Octalysis framework provides a solid foundation for the development of gamified educational environments, contributing to student engagement and establishing the basis for the implementation and evaluation of the platform in the second stage of the undergraduate thesis.