Artificial Intelligence and its Applications in Physical Sciences

Abstract

Artificial Intelligence (AI) has become one of the most influential technological developments in contemporary scientific research. In recent years, its integration into the physical sciences has significantly transformed conventional research methodologies by enabling advanced computational modelling, automated data analysis, and predictive simulations. This systematic review critically examines the role of AI in physical sciences using the PRISMA 2020 framework. Relevant studies published between 2023 and 2026 were collected from databases such as Scopus, Web of Science, Science Direct, Springer Link, and arXiv. Out of 512 initially identified records, 78 studies satisfied the inclusion criteria and were analysed in detail.
The review categorizes AI approaches into machine learning, deep learning, and physics-informed neural networks (PINNs), examining their applications in physics, chemistry, materials science, astrophysics, climate science, and engineering systems. Findings indicate that AI enhances computational efficiency, improves predictive accuracy, and accelerates scientific discovery. Hybrid approaches that combine physical laws with AI models demonstrate particularly promising results in solving complex scientific problems. However, issues such as interpretability, computational expense, and limited high-quality datasets continue to challenge researchers.
The study concludes that AI has the potential to redefine scientific inquiry in physical sciences by supporting intelligent experimentation, autonomous laboratories, and explainable scientific computing. Future developments should focus on ethical AI, interdisciplinary collaboration, and sustainable computational practices.