•  
  •  
 

AIS Transactions on Human-Computer Interaction

Abstract

Stuart Russell's human-compatible AI framework proposes that machines should maximize the realization of human preferences—in practice, designing AI systems to defer unconditionally to human operators. While intuitive, this paper argues that total deference introduces its own risks and limitations. We present findings from a workshop panel at the 59th Hawaii International Conference on System Sciences (HICSS) that examined (un)cooperative robots: systems that can push back against or refuse human directives for principled reasons. The panel addressed four questions: 1) the pros and cons of total deference, 2) theories of collaboration underpinning human-robot interaction, 3) how robot morphology shapes the appropriateness of uncooperative behavior, and 4) the ethical implications of (un)cooperative robots. Eleven research themes emerged (authority hierarchies, cross-cultural variation, benevolent uncooperativeness, goal alignment, theory of mind, joint action theory, common ground, morphology, individual agency, ethical frameworks, and AI liability) along with targeted research questions. Together, these themes call on information systems and human- computer interaction scholars to move beyond compliance-centric AI design towards more richly understanding how principled uncooperativeness can enhance safety, ethics, and the quality of human-robot partnerships.

DOI

10.17705/1thci.00251

Download

Share

COinS
 

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.