Two open-ended probes using a Wizard of Oz technique tested how users define boundaries in personalized machine learning. Apple Machine Learning Research examined whether giving users design power prevents systems from limiting their imagination. The study shifts focus from technical feasibility to ontological negotiation.