Perspective: Towards sustainable exploration of chemical spaces with machine learning

2026-03-31

AI Summary

This work addresses the sustainability challenges in AI-driven molecular and materials discovery, which are exacerbated by escalating computational and data demands. The authors propose a hierarchical intelligent workflow that integrates efficient modeling, multi-fidelity learning, active sampling, and physics-informed constraints to jointly optimize energy efficiency, reliability, synthesizability, and multiple performance objectives. By leveraging foundation models, knowledge distillation, and embedded physical priors, the framework substantially reduces the computational cost of quantum simulations and model training while maintaining predictive accuracy, thereby enhancing scientific output per unit of compute. This approach advances the responsible deployment of green AI in drug and materials discovery and promotes open data sharing and reusable pipelines, offering a sustainable pathway for domain-specific AI systems.

Abstract

Artificial intelligence is transforming molecular and materials science, but its growing computational and data demands raise critical sustainability challenges. In this Perspective, we examine resource considerations across the AI-driven discovery pipeline--from quantum-mechanical (QM) data generation and model training to automated, self-driving research workflows--building on discussions from the ``SusML workshop: Towards sustainable exploration of chemical spaces with machine learning'' held in Dresden, Germany. In this context, the availability of large quantum datasets has enabled rigorous benchmarking and rapid methodological progress, while also incurring substantial energy and infrastructure costs. We highlight emerging strategies to enhance efficiency, including general-purpose machine learning (ML) models, multi-fidelity approaches, model distillation, and active learning. Moreover, incorporating physics-based constraints within hierarchical workflows, where fast ML surrogates are applied broadly and high-accuracy QM methods are used selectively, can further optimize resource use without compromising reliability. Equally important is bridging the gap between idealized computational predictions and real-world conditions by accounting for synthesizability and multi-objective design criteria, which is essential for practical impact. Finally, we argue that sustainable progress will rely on open data and models, reusable workflows, and domain-specific AI systems that maximize scientific value per unit of computation, enabling efficient and responsible discovery of technological materials and therapeutics.

Problem

• sustainability
• chemical space exploration
• machine learning
• computational cost
• resource efficiency

Innovation

• sustainable machine learning
• multi-fidelity modeling
• active learning
• physics-informed AI
• chemical space exploration