Recently, the increased use of artificial intelligence in the sector of health has been observed. The propose of this study is to systematically and methodically synthesize recent and relevant research on the use of artificial intelligence (AI) models for diagnosing temporomandibular joint osteoarthritis (TMJOA) using various types of data. Seven databases were searched to identify articles on TMJOA and AI published between 1991 and 2024. One hundred and thirty-seven papers were identified, of which 24 were retained, representing over than 5,000 patients, between control and TMJ osteoarthritis groups. Several studies have focused on diagnosing temporomandibular joint (TMJ) osteoarthritis from radiographic images (approximately 7,700 orthopantomograms, over than 8,800 cone-beam computed tomography scans, and about 3,600 MRI scans). Other studies have explored the 3D shape of the condylar head and the determination of different disease stages observed on radiographic images, integrating biological, radiomic features, demographic, and clinical data… to identify the highest performing model and the most promising features to diagnose temporomandibular joint osteoarthritis. A wide variety of artificial intelligence models have been used, ranging from Machine Learning (XG-Boost, SVM, Logistic Regression, etc.) to Deep Learning (VGG16, R-CNN, SSD, etc.), thus, a systematic review was conducted to determine the contribution of these models.