Cervical cancer is one of the leading causes of death in women globally, and early detection is the key to the advance in survival. Colposcopy is a common method for visual inspection of the cervix, but it is highly dependent on the clinician’s skill for accurate diagnosis and is inconsistent in interpretation. The recent advances in artificial intelligence (AI), particularly in machine learning (ML) and deep learning (DL), have enabled the progression of computer-assisted diagnostic systems that enhance consistency in the detection of cervical cancer through colposcopic image analysis. Traditional ML techniques utilize hand-crafted features such as shape, color, and texture descriptors as well as classifiers like Support Vector Machines, Random Forests, and Gaussian Mixture Models. Conversely, DL models, i.e., Convolutional Neural Networks and more recent Transformer-based ones, learn discriminative features automatically directly from raw image data with much better diagnostic performance. Extensive ML and DL techniques utilized for the diagnosis of cervical cancer from colposcopic images are presented in this study, encompassing preprocessing methods, feature extraction methods, classification methods, and new neural architectures. Several deep learning-based models have been applied to cervical cancer detection with varying degrees of success. CNN based methods had outputs that ranged from 82.6 to a remarkable 92.31%. But the highest recorded accuracy was from a GAN model. Reported results demonstrate accuracies exceeding 90% in several studies, with hybrid and DL models, such as DeepCervix and FSOD-GAN, achieving near-perfect performance. Despite these developments, complications such as data scarcity, variability in image quality, and the need for clinically interpretable models remain unresolved. This paper highlights key research directions to bridge the gap between algorithmic development and real-world clinical deployment.