In contemporary dance, experimentation with novel and idiosyncratic forms of creating and staging bodily movements lies at the center of artistic attention. Many of these experiments also involve the use of generative techniques. For this reason, contemporary dance provides a fertile environment for exploring creative applications of generative machine learning. This article presents the author’s own work on four machine-learning-based systems, each of which addresses a specific application domain in contemporary dance. The system Granular Dance generates synthetic motions that support choreographic ideation. The system Puppeteering AI creates an artificial dancer that human dancers can interact with. The system Expressive Aliens provides insights into the relationship between body morphology and movement expressivity. The system RAMFEM demonstrates how a dancer’s creative movements made while improvising to music can be adopted for movement sonification. For each of these systems, the article describes the technical implementation and first applications. This information is complemented with a description of the context that is most relevant for this research. This includes how contemporary dance distinguishes itself from other dance practices, the types of generative techniques that are popular in contemporary dance, and the main application domains of machine-learning in contemporary dance.