 Data-driven science is a new paradigm in materials science that uses data as a resource to discover new or improved materials or phenomena, fueled by factors such as the open science movement, national funding, and information technology advancements. However, challenges such as data veracity, integration of experimental and computational data, data longevity, standardization, and the gap between industrial interests and academic efforts impede progress in this field. The historical development and current state of data-driven material science are discussed, along with key successes and challenges, providing a perspective on the future development of the field. This article was authored by Laurie Himinan, Amber Jertz, Adam Stewart Foster, and others.