A comprehensive study published in Nature finds that AI has accelerated the pace of scientific discovery in materials science by a factor of four, measured by the rate of new material discovery and characterisation.
Google DeepMind's GNoME model alone has predicted the stability of 2.2 million new crystal structures — equivalent to 800 years of conventional experimental discovery.
Of these predictions, over 700 new materials have been synthesised and validated in laboratories, including several with promising applications in battery technology, superconductivity, and solar energy.
The acceleration is not limited to materials science. Similar AI-driven speedups have been documented in drug discovery (3x), climate modelling (2.5x), and genomics (5x).