Abstract: Magnetic confinement fusion aims to harness the energy released by fusing light nuclei in a plasma hotter than the core of the Sun. To make fusion a practical energy source, these plasmas must remain confined and stable for extended periods. However, their collective dynamics can give rise to magnetohydrodynamic instabilities that degrade confinement and, in some cases, abruptly terminate the plasma. In this colloquium, I will focus on Neoclassical Tearing Modes: instabilities that tear and reconnect the confining magnetic field, forming magnetic islands that degrade confinement and can lead to plasma termination and intense heat and particle fluxes to the tokamak wall. I will introduce the plasma physics of magnetic islands, then describe how analyses of a DIII-D experimental database identified the origin of these instabilities in the ITER baseline scenario and how recent experiments led to a reproducible approach for avoiding them under those conditions. I will also introduce my graduate student research team at UC Irvine and discuss our efforts to understand, predict, and control tearing modes. These efforts include using machine learning to automate the demonstrated control approach and developing strategies that can be applied across a broader range of plasma conditions.