Machine learning techniques make computers teachable by example, much like people -but faster and cheaper. Already widespread in industry, machine learning is currently revolutionizing astronomy, where it encounters successes while also arising some skepticism. I will present a few applications of supervised and unsupervised machine learning from my research: finding black holes in star clusters, grouping stars based on chemical similarity, automatically calculating parameters from color-magnitude diagrams, and measuring the spectral index of turbulence in molecular clouds from images. Throughout, I will touch on the topic of interpretability, which I consider crucial for the adoption of machine learning in science.