Can AI be suspicious?

Abstract

The current research focuses on how consumers respond differently to the message from a sales assistant, depending on whether a sales assistant is artificial intelligence (AI) or a human. A message from an AI sales assistant is more credible than a human in a highly suspicious sales situation. However, in a less suspicious situation, a message delivered by a human is more credible than an AI. Compared with a human, an AI is perceived to have a low suspicion of ulterior motives. We also demonstrate moderating effects—first, a format of recommendations moderates the relationship between sales assistant type and message credibility. A message with a complex syntax reduces the suspicion of ulterior motives of a human, which in turn increases message credibility in a suspicious sales situation. Second, we present the moderating role of customized AI. A customized AI is considered a proxy for firms and strengthens the suspicion of ulterior motives, so the message credibility of customized AI is attenuated.