How human are our machines? Rethinking how we communicate with social robots

As robots make their way into unstructured contexts, they are required to have social characteristics to effectively communicate with users, posing the question: To what extent do existing theoretical frameworks apply to interaction with modern social robots? Our theoretical understanding of interaction with media agents is derived from the Computers-Are-Social-Actors (CASA) framework. Since its conception, there have been significant advancements to media agents themselves, people’s understanding of media agents, and the nature of interaction between these media agents and people, which has transformed people’s understanding and expectations of media agents from what CASA predicted.
This dissertation conducted two cross-sectional experiments and one longitudinal survey. The first study focused on the effect of interdependency and a robot’s display of augmented cognitive capabilities on human-robot team behaviour and found that when participants perceived the robot as a teammate, they did not experience more team affinity or behaviour than when they did not perceive it as a teammate. The second study found that when being helped by a robot, participants did not reciprocate the helpful behaviour, even with the presence of a single humanlike social cue (a face). The third study was based on data from an eight-week study, in which children interacted with a social robot within their own home. It did not find any robust long-term patterns between social presence and perceived companionship.
These findings indicate the need to update theoretical frameworks such that they can better explain social responses in (long-term) interactions between advanced social robots and experienced users.