Fart Gags and Prudish Machines: Laughter in Human-agent Interactions

We explore how laughter functions in in-the-wild human–Alexa interactions recorded in domestic settings. To do so, we analysed all instances of laughter in a corpus containing audio recordings from six households where an Alexa device had been newly acquired. The participants had no or very limited prior experience with voice assistants. Their interactions with Alexa were recorded over the first seven to ten weeks of use. Unlike previous HRI studies that primarily focus on dyadic, task-based exchanges, our analysis reveals that laughter in these real-world settings often emerges in multiparty interactions and serves a range of social functions beyond direct responses to the device. These observations highlight not only the need for ecologically grounded models of laughter in human–robot interaction, but also the value of linguistic and interactional analysis in uncovering the nuanced communicative roles laughter plays in everyday technology use. Such an approach allows us to identify how laughter signals both matches and mismatches in communication by marking alignment, managing breakdowns, and negotiating social meaning in interactions that often involve more than just the user and the device.