Cosmology is entering an era of data-driven science, due in part to modern machine learning techniques that enable powerful new data analysis methods. This is a shift in our scientific approach, and requires us to ask an important question: Can we trust the black box? I will present a deep machine learning (ML) approach to constraining cosmological parameters from X-ray surveys of galaxy clusters. The ML approach has two components: an encoder that builds a compressed representation of each galaxy cluster and a flexible convolutional neural network to estimate the cosmological model from a cluster sample. From mock observations, the ML method estimates the amplitude of matter fluctuations, sigma8, at approximately the expected theoretical limit. More importantly, the deep ML approach can be understood and interpreted. The model interpretation led to the discovery of a previously unknown self-calibration mode for flux- and volume-limited cluster surveys. I will describe this new mode, which uses the amplitude and peak of the cluster mass PDF as an anchor for cluster mass calibration.