Análisis de la inteligencia artificial en la gestión de la seguridad y salud ocupacional

Abstract

Abstract: The development of artificial intelligence brings with it the need to assess how accurate this technology is in managing occupational hazards. This study employed a quantitative approach with an experimental design to compare the performance of the ChatGPT and Gemini models in detecting hazards in the construction sector. Various metrics such as accuracy and sensitivity of these technological tools were evaluated using ten simulated scenarios presented to the models. The results suggest that, although both can provide useful insights for analysis, they should be used as a complementary aid to professional work. After all, these tools present significant limitations and errors in hazard classification that cannot be overlooked. The study highlights the value of artificial intelligence as a complementary tool that can strengthen occupational safety and health management, further demonstrating that it is technically feasible to integrate these analyses through APIs in workplace environments.