Guy Aridor is a PhD student in Economics at Columbia University whose research primarily focuses on using tools from theoretical computer science to help understand problems in behavioral economics and industrial organization. Previously, he worked as a software engineer at Nutonian and HubSpot as well as studied computer science, mathematics, and economics at Boston University.
What can artificial intelligence teach economics?
While the fields of cognitive science, computational neuroscience, and artificial intelligence have progressively built off each other in the quest of reverse engineering intelligence, economics has largely remained agnostic to this progress. Historically economics was only concerned with observing the choices individuals made and not with the process decision makers utilized in coming to these choices. However, in recent years behavioral economics and neuroeconomics have emerged with the purpose of focusing on models that take into account the decision making process in order to better rationalize economic behavior that previously had been at odds with classic economic decision theory. In this talk we discuss how the direction that behavioral economics and neuroeconomics is taking is similar to the direction taken in parts of the artificial intelligence literature towards striving for designing bounded optimal agents instead of perfectly rational agents. We discuss what insights from this literature can inform the models utilized in economics and, conversely, how the models in economics can inform the artificial intelligence research community. Finally, we present preliminary experimental evidence from a human problem solving experiment as an application of the applicability of the concepts that are discussed.