Choice Engine

The Choice Engine is Sunstein's most ambitious application of behavioral science to AI governance. The concept designates an AI system that functions as advisor rather than decider, analyzing available options in a complex domain — health insurance, mortgages, energy providers, educational programs — and recommending the option that best serves the individual user's stated preferences and actual needs. The engine accounts for informational deficits that make consumer markets inefficient (complexity of options, opacity of pricing, difficulty of multi-dimensional comparison) and for behavioral biases that make consumers predictably poor decision-makers (present bias, status quo bias, availability heuristic). Properly designed, the Choice Engine preserves autonomy through override preservation while overcoming the informational and cognitive barriers that prevent most consumers from making choices that serve their reflective interests. Improperly designed, the same architecture becomes a manipulation engine serving the deployer rather than the user.