Freshwater scarcity has become more widespread on a global scale in recent years. Surface water resources are no longer sufficient to meet the demands of human productivity and survival, and groundwater resources are now being utilized extensively. Therefore, exploration of potential groundwater resources is critical for their rational development and utilization. In this study, four methods of machine learning, ensemble learning, deep learning, and automated machine learning (AutoML) were selected, and their representative models were chosen for comprehensive groundwater potential mapping (GPM) in Hubei Province, China. In total, 812 samples were collected and acquired; about 80% of the samples were selected randomly as training data set and the remaining 20% as test data set. Considering local hydrological, geological and climatic conditions, slope, elevation, curvature, landforms, geology, distance to fault, land type, soils, precipitation, evaporation, topographic wetness index, stream power index, distance to rivers, normalized difference vegetation index, and distance to residential area were selected as factors in this work. Finally, the four models were validated using receiver operating characteristic (ROC)–area under the curve (AUC) and classification reports. The Shap values of each factor were calculated as a measure of contribution of each factor to groundwater potential. The results showed that the ROC–AUC values of random forest, Stacking, convolutional neural network, and AutoML were 0.82, 0.85, 0.87 and 0.88, and the precision values were 0.793, 0.784, 0.807 and 0.844, respectively. These results indicate that the first application of AutoML to GPM in this study was feasible, and it had the best predictive ability and accuracy compared with the other three methods. In addition, the Shap values indicated that the two factors that had the greatest influence on groundwater potential were geology and precipitation. The results obtained in this study can provide technical support to the local government for groundwater exploration and development in Hubei Province, and the newly introduced method of AutoML can provide new ideas to experts and scholars for groundwater potential assessment.