Laser-based powder bed fusion (PBF) and directed energy deposition (DED) additive manufacturing have been embraced by much of the aerospace and defense industry for part production and repair. Unfortunately, there still remains considerable uncertainty regarding the causes and effects of many defects types observed in PBF and DED components. Numerous conditions can lead to the formation of defects (i.e. internal discontinuities or undesirable microstructure) that can negatively affect build and part quality. Some of these defects are easily attributable to systematic errors (e.g. poor processing parameters or contamination). However, many others appear stochastic in nature, appearing randomly even under ideal processing conditions. Here, we detail recent work seeking to elucidate the mechanisms by which systemic and stochastic defects form and approaches to detecting both types in- situ. We illustrate that naturally-occurring, stochastics flaws can be emulated via perturbation of processing conditions and can be sensed via illuminated melt pool imaging and observation of the vapor plume produced above the melt. Similar approaches can also be utilized to detect systematic variations in processing parameters. It is also possible to predict and use feed forward control to avoid many defect types. The presented analyses and methodologies present a path forward for mitigation and detection of both systematic and stochastic defects.