Additive manufacturing (AM) has revolutionized modern fabrication by enabling complex geometries, material efficiency, and customized production. However, process variability, material inconsistencies, and defect formation remain critical challenges, limiting scalability and industrial adoption. Machine learning (ML) has emerged as a powerful tool to address these limitations by enabling data-driven optimization, defect detection, material property prediction, and real-time process control. This review provides a comprehensive analysis of ML applications in AM, spanning polymers, metals, ceramics, and carbon-based materials, with a focus on process optimization, quality assurance, and predictive modeling. Specifically, this review examines real-time defect detection through vision-based ML techniques, printing parameter optimization using supervised and reinforcement learning, and predictive modeling of material properties–laying the groundwork for deeper exploration of key methodologies such as deep learning and physics-informed models. Key ML methodologies, including deep learning, reinforcement learning, and hybrid physics-informed models, are explored in the context of enhancing print precision, mechanical performance, and functional properties. Despite significant advancements, challenges such as data scarcity, model generalization, and real-time integration into AM workflows persist. Emerging trends, including multimodal sensor fusion, in situ monitoring, and cloud-based predictive analytics, are discussed as potential pathways toward fully autonomous and intelligent manufacturing. By consolidating recent developments and outlining future directions, this review provides essential insights for researchers, engineers, and industry professionals looking to harness ML in AM, facilitating advancements in process efficiency, quality control, and overall manufacturing reliability.