This perspective highlights an emerging framework for understanding evolved biological and designed physical systems as tunable materials. In analogy to supervised learning in artificial neural networks – where weights between nodes are adjusted to optimize a cost function during the training phase – recent work postulates that biological systems may be using similar tuning rules on their adjustable degrees of freedom to learn or perform tasks. Researchers are also using these algorithms to design synthetic physical systems that exhibit exotic mechanical responses or perform tasks like classification and regression. Here, we briefly review the algorithms that enable such behavior and highlight examples in both physical and biological systems, emphasizing major open questions in this nascent field. Our goal is to provide SPLASHY researchers with a jumping off point for considering whether their own systems of interest might be a platform for tunable matter, and spur future work to address open questions.