Behavioral Science and Psychology

This book makes the argument that comparing AI to humans leads us to diminish similar outcomes from AI across situations. This may be taking a human’s advice for a restaurant recommendation over an AI’s or believing that AI can’t be as biased as people can when denying loans to others. This machine penalty is caused both by comparing humans and AI in terms of appearance, identity, behavior, mind, and essence, and by situations involving controllable, personal, important, subjective, or moral decisions. It can be applied across many different situations, where we diminish different AI outcomes. We penalize machines’ influence when they give advice, fairness when they evaluate people, blame when they cause harm, value when they produce art, and satisfaction when they provide companionship. The result is immediate consequences in those domains and downstream consequences for society. This monograph brings together diverse research from human-computer interaction, psychology, sociology, and communication including theories such as Computers Are Social Actors, anthropomorphism, mind perception, and algorithm aversion to present an expansive argument and evidence for the machine penalty.