In this work, four artificial intelligence (AI) techniques, based on Artificial Neural Networks, Support Vector Machine (SVM), and Regression Tree Ensembles, were used to estimate the operating temperature of photovoltaic (PV) modules (TPV). The models' input parameters correspond to experimental measurements of environmental (solar radiation, ambient temperature, relative humidity, wind speed, and wind direction) and operational (power output and tracking system) variables. Several AI models were trained and statistically compared with the measured data using a computational methodology that determines the performance and accuracy of the AI technique. Finally, a global sensitivity analysis was conducted to identify the ability of each technique to reflect the physical coherence of the phenomenon that is under study. It is reported that the four techniques can provide an estimate having a precision of about 93%. On the other hand, the sensitivity analysis demonstrates that all the models cannot correctly interpret the physical interaction of the input parameters with respect to TPV, where the SVM is reported to be the most appropriate. The results indicate that the proposed methodology is a viable alternative for the estimation of TPV by AI techniques. This methodology can be implemented as an alternative tool in the development of smart PV module cooling systems to improve its performance and to reduce its operating costs.