Resisting Artificial Intelligence: When Do Decision-Makers Avoid or Use Algorithmic Input?

Abstract

When do people heed the advice from an algorithm as opposed to a human? This session delves into novel research in which people opt for the recommendations from algorithmically-derived sources and uses experimental approaches to quantify and theoretically inform the conditions in which decision-makers listen to (or ignore) algorithmic advice. The papers in this session examine a breadth of decision-making scenarios and a diverse range of contextual factors, including: contrast in stated preferences and behavioral decisions; domain sensitivity of algorithmic advice; conversational fluency in evaluations of AI competence; gender-status stereotype congruence of AI agents; and the effect of anthropomorphization of algorithms. Robo-Coaching: Responses to Performance Feedback from Algorithms Versus People Presenter: Jennifer Marie Logg; Georgetown U. Presenter: Julia Alexandra Minson; Harvard Kennedy School Presenter: Francesca Gino; Harvard Business School Making Sense of Recommendations Presenter: Michael Yeomans; Harvard Business School Presenter: Anuj Shah; U. of Chicago Booth School of business Presenter: Sendhil Mullainathan; Harvard U. Presenter: Jon Kleinberg; Cornell U. Detrimental Dehumanization in the IoT Presenter: Christian Hildebrand; U. of St. Gallen Presenter: Donna Hoffman; George Washington U. Presenter: Thomas P. Novak; George Washington U. The Role of Status and Gender Stereotype Congruence in Preference for AI Agents Presenter: Heather Hee Jin Yang; Massachusetts Institute of Technology Object-Oriented Anthropomorphism as a Mechanism for Understanding AI Presenter: Donna Hoffman; George Washington U. Presenter: Thomas P. Novak; George Washington U.