We study the effects of collective learning, communication, group memory, and distributed computation on the efficiency and accuracy of collective behavior and decisions in groups of animals, collections of artifical agents, and ensembles of deep neural networks.