Extensive teaching of simple tasks through watching and repetition, then practiced several at a time on randomly generated, rich environments, virtual or not, allows to evaluate and refine the creation of “policies” that become instantly ready for real-life use by the learner. Are we talking machines or humans?