Artificial intelligence in advanced manufacturing: a comprehensive review of machine learning, generative AI, and agentic AI systems
Аннотация
Artificial intelligence (AI) is reshaping advanced manufacturing by improving prediction, decision support, and system adaptability across the product and process lifecycle. While traditional machine learning (ML) and deep learning (DL) are already established in applications such as process monitoring, defect detection, predictive maintenance, and scheduling, recent advances in Generative AI and agentic AI are expanding the scope of industrial intelligence. This review provides a structured synthesis of AI in advanced manufacturing through three complementary layers: (i) traditional ML/DL, (ii) Generative AI, and (iii) agentic AI systems. First, the review outlines the digital context of smart factories, including cyber–physical production systems, industrial data infrastructures, and digital twins, and summarizes the role of ML/DL as the predictive backbone of manufacturing intelligence. It then examines how Generative AI supports design, documentation, workflow automation, and synthetic data generation. Next, it defines agentic AI systems in a manufacturing context and discusses emerging architectures for closed-loop rescheduling, quality control, maintenance coordination, and cross-site decision-making. Finally, the review identifies cross-cutting challenges related to safety, explainability, integration, governance, and data infrastructure, and proposes a research roadmap centered on hybrid physics–data–knowledge models, trustworthy AI, federated and edge-native architectures, and human-centered Industry 5.0 deployment.