This paper studies a broad number of AI based studies within spinal surgery, examining the effectiveness of novel AI assisted procedures. The paper encompasses pre-operative modeling, intraoperative navigation, and prediction of postoperative outcomes, all using various AI models and methodology. Preoperative planning encompasses studies in organizing financial planning, preoperative patient selection and outcome prediction, as well as guiding clinical decisions and improving surgical planning. Another major innovation studied was the increased speed and accuracy of designing 3D mechanical models, as well as in imaging practices for segmenting MRI data. From an intraoperative perspective, a machine-vision image-guided system (MvIGS) was developed to assist surgeons during pedicle screw placement in pediatric spine deformity correction surgery, which was able to reduce intraoperative fluoroscopy time by 68% and radiation exposure by 66%. There were also studies that demonstrated how AI could be used for high accuracy screw placement in cervical spine surgeries. The study predicts that these techniques can reduce procedural times and patient morbidity risk by reducing the number of trials of screw trajectories, repeat fluoroscopic imaging, and avoidance of damage to surrounding structures. Finally postoperative studies found that AI could be used to effectively predict patient outcomes and organize patients based on individual traits and distinct pathologies to aid in predictions. The paper finishes with an examination of the limitations of AI, mentioning the issue of dataset homogeneity, which restricts AI generalizability, as well as ethical concerns over patient consent in data collection.