SAMANTHA: IA APLICADA AO ESTUDO DA RELAÇÃO ENTRE ATENÇÃO E APRENDIZAGEM

Abstract

This study presents a generative Artificial Intelligence (AI) named Samantha, developed to investigate the relationship between attention and learning. The research focuses on the predictive capability of attention in educational conflicts and classroom performance. A psychometric application called iGnosi was created, classifying attention into four categories: Focused, Alert, Divided, and Alternating. Psychometric data were collected from 144 undergraduate students. For Samantha's development, the base model GPT-3.5 turbo 125 was used, trained with a dataset containing 577 entries for three roles: System, User, and Assistant. The training involved 307,266 tokens over 3 epochs, resulting in a loss of 0.0318. In initial tests, Samantha demonstrated the ability to interpret and guide researchers and educators based on the results, proving to be inspiring and motivating. Her feedback was careful, even in low-performance scenarios of the students. It is concluded that the use of AI in educational research can predict classroom problems and that the generative use of AI can help bridge the gap between scientific culture and educators.