This paper focuses on the application of Artificial Intelligence (AI) networks for fault diagnosis in large-capacity electric motors, which are used in industries such as manufacturing, energy, and mining in Vietnam. The study highlights the potential of AI in automatic fault diagnosis through the analysis of key signals, such as motor vibrations. It uses the Short-Time Fourier Transform (STFT) to analyze vibration signals in the time domain. The output of the STFT is a spectrogram, also known as an image spectrum. The paper also evaluates three popular AI and machine learning models, namely You Only Look Once (YOLO), Support Vector Machine (SVM), and Residual Neural Network (ResNet), to assess their accuracy based on a self-constructed database. The results from these tests help determine the ability to identify and classify electric motor faults, providing potential solutions to improve efficiency and accuracy in the inspection and maintenance of industrial equipment.