G.J. Mulholland
Citrine Informatics,
United States
Keywords: artificial intelligence, machine learning
Summary:
Artificial intelligence is already reshaping materials science, not through incremental efficiency gains but by transforming how discovery itself occurs. From polymer design to catalyst optimization, AI-driven approaches are uncovering relationships and navigating chemical spaces at scales previously unimaginable, setting the stage for a true paradigm shift in how new materials are conceived and validated. Yet alongside this genuine revolution, a parallel narrative has emerged: one defined by exaggerated claims and a lack of appreciation for the hard realities of materials development. Many recent entrants to the field underestimate the scarcity of high-quality experimental data, the complexity of structure–processing–property relationships, and the integration challenges between predictive models, experimental validation, commercial relevance, and manufacturability. This talk distinguishes substance from spectacle, arguing that while AI is indeed remaking the field, the path to breakthrough materials demands more than algorithms. It requires deep materials understanding, rigorous data stewardship, and tight coupling between computation, experiment, business model, and expertise. It will also highlight advances in AI that may yield additional dividends in transforming healthcare and other domains but have not yet been leveraged meaningfully in materials.