Artificial intelligence (AI) is reshaping scientific research by accelerating discovery and enabling the analysis of complex data that traditional methods struggle to handle. This review examines over 310,000 journal articles and patents from the CAS Content Collection (2015–2025), with a focus on, biomedical research, and materials science. Our data-supported trend analysis finds that since 2020, AI-related publications have surged across both academic and industrial domains. Traditional machine learning (ML) methods, such as decision trees, random forests, and support vector machines, remain foundational, particularly in molecular property prediction and classification tasks. However, there is an increased adoption of advanced neural network architectures, including graph neural networks for molecular representation and generative models for de novo compound design. Hybrid and ensemble approaches that integrate traditional algorithms with deep learning are increasingly favored for their robustness and adaptability. Large Language Models (LLMs) like GPT, BERT, and GEMINI are transforming literature mining, hypothesis generation, and experimental planning. Domain-specific models such as AlphaFold and chemical language models enable breakthroughs in protein structure prediction and reaction outcome forecasting. Industrial chemistry and chemical engineering are increasingly contributing to AI-driven academic publications, while biochemistry patents have grown eightfold, reflecting AI’s dual impact on discovery and commercialization. Despite these advances, challenges remain, particularly in data privacy, quality, interpretability of black-box models, and uncertainty quantification. These findings highlight AI’s transition from descriptor-based approaches to end-to-end learning systems, offering essential insights for researchers and policymakers navigating its transformative role in science.
Baranwal et al. (Mon,) studied this question.