Zoom link:

https://aarhusuniversity.zoom.us/j/61282365222

"Don’t Obey the Robot: Designing for Agency in the Face of Persuasive Machines"

Abstract. In this talk I trace my research trajectory from early work leveraging social cues for effective interaction to more recent efforts focused on designing socially effective robots that preserve human agency - the capacity to interpret, choose, and resist.  

Robots continue to be made increasingly lifelike to be engaging, communicate effectively, and foster trust. Through my team's work, however, I have found that the same design strategies that support interaction can also give robots surprising power to influence people. This power can be asymmetrical and inflexible, emerging as structural consequences of social design, even when unintended. Ever-more polished social robots can create unrealistic user expectations and obscure robot intention and action, reducing transparency and undermining a person's ability to reflect and make informed interaction choices. I argue that social robots should be transparent, should be designed to garner expectations congruent with their ability, and created from the ground up to support reflective choice, resistance, and meaningful user control.

I present projects from my research group tracing a shift from early excitement about social human-robot interaction, to a more critical focus on transparency, expectations, and agency in long-term, real-world deployments. My research is design-driven: I investigate these issues through designing, building, and studying these systems in ecologically valid contexts, using the creation and use process to reflect on interaction power and user autonomy.