Rudaz2025c
| Rudaz2025c | |
|---|---|
| BibType | ARTICLE |
| Key | Rudaz2025c |
| Author(s) | Damien Rudaz |
| Title | Social robots as designed artifacts: the impact of programming tools on “human–robot interaction” |
| Editor(s) | |
| Tag(s) | EMCA, Programming tools, Material agency, Social robots, Linguistic practices, Human-robot interaction, Turn-taking models, AI Reference List, In press |
| Publisher | |
| Year | 2025 |
| Language | English |
| City | |
| Month | |
| Journal | AI & Society |
| Volume | |
| Number | |
| Pages | |
| URL | Link |
| DOI | 10.1007/s00146-025-02743-7 |
| ISBN | |
| Organization | |
| Institution | |
| School | |
| Type | |
| Edition | |
| Series | |
| Howpublished | |
| Book title | |
| Chapter | |
Abstract
This work documents how roboticists’ skilled conversational expertise is unavoidably constrained and guided as it is translated into rule-based robots through specific programming tools. To shed light on the social construction of “social” robots, it documents the (material) agency of tools used to script or program robots within the technology company Aldebaran (formerly Softbank Robotics Europe), which specialized in manufacturing and programming the social robots Pepper and NAO and their conversational AI. Conducted over the course of several years among roboticists, this ethnography investigates how programming tools—their variables, publicly exposed events, hierarchy of information, available data streams, and documentation—facilitated the crafting of specific forms of talk (among many) for Pepper. Two types of constraints are examined: those encountered by roboticists when crafting disfluencies and filled pauses (“uhh”) in Pepper’s speech, and those faced when enabling the robot to “continue speaking” after a silence. These constraints produced “paths of least resistance” toward specific designs for the robot, which, in turn, impacted its competence in situated interaction with “end users”. Crucially, these constraints were not inevitable technical limits, but rather resulted from specific assumptions about language and, in particular, what “conversing with a robot” entails. In other words, roboticists’ tools led to the embedding of linguistic practices and a specific organization of talk into their robot. The study concludes with the continued significance of these “paths of least resistance” (and their underlying assumptions about “what conversing is”) in the era of chatbots based on generative AI.
Notes
