The third generation of the CRISPR/Cas9-mediated genome fixed-point editing technology has been widely used in the field of gene editing and gene expression regulation. How to improve the on-target efficiency and specificity of this system, as well as reduce its off-target effects are always the bottleneck in its development. Machine learning provides novel methods to the problems of the CRISPR/Cas9 system, and CRISPR/Cas9-based machine learning has recently become a very hot research topic. In this review, we firstly outline the mechanism of the CRISPR/Cas9 system. Subsequently, we elaborate the current issues of CRISPR/Cas9, including low efficiency and potential off-target effects, and sequence-recognizing limitation from protospacer adjacent motif (PAM). Finally, we summarize the applications of methods within the machine learning framework for optimizing the CRISPR/Cas9 system, such as optimized single-guide RNA (sgRNA) design, CRISPR/Cas9 cleavage efficiency prediction, off-target effects evaluation, gene knock-out as well as high-throughput functional genetic screening and prospects for development.