Reliability engineering has traditionally relied on probabilistic modeling, limit-state formulations, and stochastic process theory to quantify system safety and failure risk. While these classical methods provide strong mathematical rigor and interpretability, they often depend on restrictive assumptions and limited data representations, making them less adaptable to complex, sensor-rich cyber-physical systems. The rapid advancement of industrial Internet-of-Things infrastructures, high-frequency sensing technologies, and computational intelligence has catalyzed the emergence of data-driven and artificial intelligence (AI) approaches for reliability assessment. This paper presents a comprehensive synthesis of classical reliability theory and modern AI-based paradigms, proposing a unified taxonomy that integrates probabilistic methods, machine learning models, deep learning architectures, physics-informed neural networks, digital twin frameworks, and uncertainty quantification strategies. A structured comparative analysis is developed to evaluate these paradigms across data requirements, interpretability, scalability, adaptability, and uncertainty handling capabilities. The study further identifies critical research gaps, including the need for explainable AI in safety-critical reliability applications, hybrid stochastic–deep modeling frameworks, reliability assessment of AI components, federated predictive maintenance, and robustness under adversarial data conditions. Building upon these insights, a forward-looking research roadmap is articulated, emphasizing causal reliability modeling, uncertainty calibration in AI systems, real-time adaptive reliability, distributed learning architectures, and quantum-enhanced reliability simulation. The findings demonstrate that future reliability engineering must evolve toward hybrid, mathematically unified frameworks that combine theoretical rigor with adaptive intelligence, thereby enabling resilient and trustworthy operation of next-generation autonomous systems.

Keywords: Reliability Engineering, Probabilistic Modeling, Physics-Informed Neural Networks, Digital Twins, Cyber-Physical Systems, Uncertainty Quantification, Predictive Maintenance, Explainable Artificial Intelligence, Federated Learning, Causal Modeling, Adaptive Reliability, Quantum Simulation