Michael researches ways of bringing smarter game-playing AI components to real-time strategy games. His PhD will focus on the study of hierarchical structure in algorithmic decision making and different means of choosing optimal paths despite a lack of information. An important question of this research also revolves around whether the goal of capable agents can be achieved through more accessible solutions with respect to computational requirements and complexity.
Michael about his research motivation: "My motivation for this field of research comes from my admiration of seeing AI strategies for systems of such complexity that even humans are forced to fully commit if they want to develop a solid comprehension, let alone mastery. Games also provide noise-free, controlled environments, which we can explore with new approaches for decision-making and commonly with as much data as needed. Against this background, I believe, a better understanding of how to reliably choose optimal decisions can be gained, paving the way for the application of AI in areas truly beneficial to society. I hope my research can provide impactful progress on decision-making especially in the face of uncertainty for which humans often fail to properly account. At the end of the day, this research might just contribute to an AI that helps future students decide whether they, too, should pursue a PhD in their game of life."