Algorithms As a Vehicle to Reflective Equilibrium:‬ Behavioral Economics 2.0

Behavioral economics has struggled to simultaneously accommodate two facts: (i) people make mistakes, even on very consequential decisions; and (ii) different people have different preferences. Doing nothing respects revealed preference, but does little to address mistakes. Nudging addresses mistakes but minimizes individual differences, requiring assumptions about what behavior is good for all people. We argue AI will reshape behavioral economics by providing a solution to this dilemma. Algorithms can serve as thought partners that help people get closer to “reflective equilibrium”: what they would choose if they were to exercise their most considered judgment on all information available at the time. Behavioral Economics 2.0 ought to be the design of such algorithms that improve decisions while preserving agency.