Algoritmos de aprendizaje para predecir el uso de la inteligencia artificial generativa en estudiantes universitarios
Journal
Revista de Ciencias Sociales
Date Issued
2026-07-15
Type
Article
Abstract
The rise of generative artificial intelligence has transformed educational processes in higher education, generating both opportunities and challenges in university training. This study aimed to analyze machine learning algorithms to predict the use of ChatGPT and Gemini among students at the Escuela Superior Politécnica de Chimborazo in Ecuador. A quantitative, non-experimental, cross-sectional, and correlational design was employed, with a sample of 699 students selected through simple random sampling. Data were collected using a statistically validated questionnaire and processed using SPSS and Orange Data Mining. The results showed that sociodemographic variables (sex, age, religion, and academic semester) did not present significant associations with usage preference, indicating low explanatory power. In contrast, the frequency of generative artificial intelligence use was a relevant predictor, highlighting the importance of habit as a determinant of adoption. Furthermore, the Random Forest and AdaBoost algorithms achieved the highest levels of accuracy, confirming the effectiveness of ensemble methods. It is concluded that socio-educational conditions better explain student behavior than demographic factors, consolidating ChatGPT as the most used tool. Licencia de Creative Commons Atribución 4.0 Internacional (CC BY 4.0)