Recognising the Importance of Preference Change: A Call for a Coordinated Multidisciplinary Research Effort in the Age of AI

Abstract

Knowing and controlling how AI systems affect our lives and choices is crucial as AI becomes stronger and more pervasive in
everyday life. As an example, personalized recommender systems learn users’ preferences to provide suggestions that alter online
behavior; this is only one example of how modern ML systems often influence user behavior. Changes in preference are an
externality of behavior modification. To better understand how AI systems alter preferences, this paper argues for the creation
of a multidisciplinary effort called Preference Science. We provide a conceptual framework for the evolution of preferences,
explain the significance of meta-preferences and preference-change preferences, and operationalize preference in terms of integrated
ideas from other fields. We differentiate between changing one’s preferences, changing one’s preferences in a way that
is authorized, and manipulating preferences in an explicit manner. This paradigm benefits from the unique insights provided by
a variety of disciplines.