The impact of GPT, experiential learning, and reinforcement methods on complex problem solving

This study, conducted in the Differential Equations Analysis course (MA1033) at Tecnologico de Monterrey, applied supervised learning, reinforcement learning, and experiential learning methods, along with the GPT model, to solve a complex Bungee Jumping problem. Engineering students developed mathematical models using differential equations, evaluated safety aspects, reviewed real-world cases, and validated results using GPT. The quasi-experimental design, involving an experimental group and two control groups, showed the experimental group achieved an average final grade increase of 11.4% and 9.6% compared to control groups. In this second implementation, students also developed argumentative texts, enhancing critical thinking and communication skills. This methodology fostered problem-solving and transversal competencies such as collaboration and critical evaluation of technologies, moving beyond rote memorization. The integration of GPT and innovative learning methods significantly improved academic performance and motivation, emphasizing AI's transformative potential in preparing students to tackle complex challenges with a practical and interdisciplinary approach. © 2025 IEEE.

The impact of GPT, experiential learning, and reinforcement methods on complex problem solving Academic Article in Scopus