Advances in artificial intelligence (AI) will make mass customization (individualization) of learning increasingly cost effective for many learning settings. However, taking full advantage of AI will require that we bridge several critical gaps in how individuals learn. Much of what we know about effective learning has been developed through the lenses of course-level pedagogy, instructors’ perspectives, and aggregated learner outcomes (Snow, 1991). The idiosyncrasies of individual learners often fall in the error term of our scholarship. But it may be these idiosyncrasies that can unlock how best to leverage AI to mass customize learning. In this study we examine the behavior of individual learners as they face a new and unfamiliar, online game environment. We capture highly detailed records of individual behavior across game trials revealing the idiosyncratic behavioral paths of individual learners. Following an abductive approach, we examine these behaviors to identify patterns and idiosyncrasies against which the potential of AI might be leveraged.