Reactance to Human versus Artificial Intelligence: Why Positive and Negative Information from Human and Artificial Agents Leads to Different Responses

Abstract

People increasingly receive information from artificial agents, such as robo-advisors and artificial intelligence, be it on retirement planning, health insurance or social benefits. Sometimes, this information is negative and undesirable for the person receiving it. Little research has examined the interplay of who is giving the information and what information is provided. Across two experiments and one field study, we show that people accept positive information more readily than negative information when the source is a human agent, while they are equally accepting of both when the source is an artificial agent. Drawing from reactance theory, we show that these differences arise because threatening information arouses less reactance when provided by a non-social rather than a social source, which, in turn, reduces reactance-style responses. These results have important implications for organizations in the public and private sector that need to confront people with information they may not like to hear.