We know that digital voice assistants can make it easier to shop or conduct business with companies. But can Siri or Alexa influence consumer behavior in other ways too? Specifically, how might the consumer tendency to anthropomorphize (or turn human) such assistants affect their decision-making? Researchers Iman Paul, Smaraki Mohanty and Samuel Bond conducted five related experiments in total to find out. They wanted to know how digital assistants’ “social presence” might affect consumer willingness to take part in prosocial behaviors.

Participants were recruited via Amazon’s Mechanical Turk (MTurk). They were presented, in the first experiment, with a scenario. They had to place a food order using a virtual voice assistant named “Erica” or by using the restaurant’s website. Subjects using the voice assistant heard Erica say at checkout, “Hope you enjoy your food! Would you like to add a tip?” Those who heard this message were more likely to tip than those who saw the same exact message via text from the restaurant, once their online order was complete.

A follow-up lab experiment sought to identify the mechanism underlying this differential. Here, subjects were asked to order groceries online. They could either do so by interacting with Erica or by using the website (via text). These participants rated their experience using measures of social presence and the salience of social norms. Those who spoke to the digital voice assistant were more willing to “round up” their purchase amount to the nearest dollar to contribute to charity. This effect was fully mediated by their greater perception of social presence, resulting in somewhat more salient social norms.

A third MTurk experiment confirmed that consumers interacting with a digital voice assistant were more likely to purchase more expensive brands associated with a charitable cause. However, this effect was mitigated for subjects interacting on a conventional website when they were reminded about the upcoming Christmas season, thus evoking the social norm of giving.

A fourth experiment demonstrated that perceptions of social presence when using a voice assistant are driven by both its “voice” and its other defining characteristics, such as its name and personality.  The final experiment found that subjects’ food shopping with a digital voice assistant were more willing to donate a portion of their compensation to “Food for the Poor,” than were those interacting with a conventional website.

Read the full working paper.