AI Drives Sustainable Materials Discovery With Efficiency

A multi-author Perspective published April 2, 2026 examines sustainability challenges and resource costs across AI-driven molecular and materials discovery pipelines. The authors review quantum-mechanical data generation, model training, and automated workflows, and propose efficiency strategies—including general-purpose ML, multi-fidelity approaches, model distillation, active learning, and physics-informed hierarchical workflows—to reduce compute costs while preserving accuracy and encourage open data and reusable workflows for practical impact.