Embryo selection remains one of the main unresolved challenges in in vitro fertilization, despite major advances in assisted reproductive technologies. Conventional assessment is still largely based on morphological evaluation, which is limited by subjectivity, static observation, and the difficulty of integrating heterogeneous clinical and biological data. In recent years, artificial intelligence has emerged as a decision-support tool in embryology, enabling the analysis of large datasets derived from embryo images, morphokinetic parameters, and clinical variables. This review summarizes current approaches to artificial intelligence in embryo selection, including models based on static images and time-lapse imaging data. Machine learning and deep learning techniques are discussed, including convolutional neural networks and spatiotemporal models. The evaluation of model performance is also examined, highlighting the clinical relevance of endpoints such as time to live birth compared with traditional outcome measures. Finally, ethical and clinical implications are considered, including issues related to transparency, responsibility, human oversight, and regulation. Artificial intelligence has the potential to improve embryo selection, although further validation and standardized implementation are needed before routine clinical use.