Artificial intelligence for accelerating material science research and development

The 20th century was a very successful period for the development of basic science and the transfer of knowledge into technological applications. Physics in particular has largely contributed to this effort with the advent of relativistic physics, quantum mechanics and great advances in high energy and particle physics, quantum optics or condensed matter physics. The 21st century is seeing a change in paradigm because of the many societal challenges (climate change, COVID-19 pandemic, green energy production, water, or food shortage) that become more acute and will demand a faster pace in science. The scientific method of discovery and experimentation is dramatically changing thanks to innovations in information technology such as artificial intelligence (AI), hybrid cloud, supercomputing and perhaps quantum computing in the next future. The automation of the scientific process will help accelerating discovery [1] and producing advances in science but also in new business opportunities. In this paper we shall address how AI technologies can accelerate discovery in material science. IBM is working on four technologies, namely Deep Search, a way to ingest 1000-time faster information from unstructured data, Faster Screening in order to understand data with less computing, Generative Models to expand creativity in the design of new molecules and the discovery of new drugs, 100Xfaster synthesis of new materials without ever experimenting in a laboratory [2].