Generative AI for Predictive Maintenance for Manufacturing: A Survey

Recent advances in generative artificial intelligence (GenAI)—particularly large language models, generative adversarial networks, and diffusion models—are reshaping the landscape of predictive maintenance (PdM). These technologies offer new capabilities in synthetic data generation, intuitive human-AI interaction, and multi-modal data analysis, thereby addressing long-standing challenges such as data scarcity and enabling multimodal reasoning across sensor, image, and textual data. While efforts to apply GenAI in PdM are emerging, a comprehensive understanding of its unique strengths and synergies with existing methods, such as AI and knowledge systems, remains lacking. This survey systematically reviews the application of GenAI in PdM, highlighting the emerging roles of image and language generative models, evaluating their impacts in PdM, and proposing future directions toward building trustworthy and real-time PdM solutions.