The burden of the pandemic has forced seismic changes in human behavior, causing a major transformation in two specific machine learning models. The first is canonical machine learning (CML), representing traditional approaches in pattern recognition, derived from highly structured and labeled data through computational statistics. The second is reinforcement machine learning (RML), which deploys a fundamentally different modeling paradigm as it self-adjusts individual actions to optimize a collective outcome, and operates much more autonomously than CML. We’ll discuss the evolution of both CML and RML in three critical time periods: 1) the time before the pandemic (BP), 2) the time during the pandemic (DP) and 3) the time after the pandemic (AP).