Short abstract:

This contribution provides empirical evidence of the amount of human work performed by visitors and store employees to remedy the functional limitations and enact the social character of a robot set up in a department store, then raising methological issues related to the grasp of "interactions".

Long abstract:

Praised as a "champagne expert […] able to answer the most specific questions" by the its sponsors, as an "emotional creature" naturally engaging users in a social interaction by its designers, a robot called Spoon was set up for two months in early 2019 in a Paris luxury department store to provide clients with bilingual information about a special vintage, as part of a broader event mixing cutting-edge technology and luxury branding. This contribution, drawing on lenghty periods of observation, video recordings of human-robot encounters and semi-structured interviews with visitors, focuses on the amount of human work performed by the latter along with store employees to remedy the robot’s "functional limitations and social awkwardness" (Jeon et al., 2020), ranging from mere repetitions to the recruitement of potential users through pre-interactional pathways.

Yet, this maintenance and interactive support was disregarded by the robot’s makers, whose insight on the robot sociality was derived from self-generated analytics (based on data processed by the robot, exclusive of external observational data). Contrasting these analytics with a finer-grained set of ethnographic data, this contribution thus aims at re-inscribe the enactment of robot sociality out of a dualistic framing supported by and supporting these analytics. Finally, it points out a series of methological issues we, as STS scholars, shall confront if they are to embrace the complexity of interactions whose very definition, threshold, temporal boundaries and implicated actors are affected by our spatial standpoint and data collection methods.