Power and Play: Investigating "License to Critique" in Teams' AI Ethics Discussions

Source: arxiv.org

The Challenge

Organizations increasingly adopt Responsible AI checklists and guidelines to promote ethical practice; however, these formal mechanisms often paradoxically restrict debate. By establishing narrow, compliance-driven criteria, such interventions enact 'discursive closure,' delegitimizing broader critiques regarding core business models, societal impacts, or whether an AI system should be deployed at all. Additionally, power asymmetries, hierarchical team structures, and corporate productivity pressures silence critical perspectives. Little empirical research examines how real-world teams actively negotiate these constraints or what cultural conditions cultivate a genuine 'license to critique' during internal AI ethics deliberations.

Core Findings

Drawing on socio-technical systems and organizational theories of discursive closure (Deetz) and hidden transcripts (Scott), Widder et al. recruited three corporate teams across two tech firms and one activist group to evaluate ethical deliberation. Utilizing a speculative critique game as an empirical probe followed by reflective interviews, the authors assessed team dynamics across hypothetical and workplace contexts. Findings reveal that corporate norms suppress critique through the rhetorical device of labeling broad concerns 'out of scope' to satisfy delivery timelines. Deliberations are heavily bounded by status hierarchies and the presence of managerial authority, which causes employees to compartmentalize ethics into narrow technical fixes. While speculative gaming failed to directly alter product decisions, its hypothetical nature lowered defense mechanisms, providing a psychologically safe space for practitioners to identify ideologically aligned allies for future collective action.

Strategic Takeaway

For digital leaders and socio-technical scholars, this study demonstrates that deploying standard ethics toolkits is insufficient without deliberate structural empowerment. Leaders must recognize how organizational mandates of efficiency, speed, and 'solving' systems inherently suppress fundamental questioning. To prevent ethical critique from being dismissed as 'out of scope,' leadership must actively expand discourse boundaries beyond functional technicalities to business logics. Moreover, acknowledging internal power hierarchies is vital: executives must create counter-hegemonic spaces where practitioners feel empowered to express critical dissent without risking career safety or being labeled unproductive troublemakers.

Deep Dive Q&A

What does 'license to critique' mean in the context of organizational AI ethics?

Coined via organizational theory, 'license to critique' refers to the socially and structurally legitimized freedom of employees to raise fundamental, non-standard, or disruptive ethical concerns regarding a technology, without fear of career reprisal or organizational dismissal.

How do standard Responsible AI toolkits inadvertently cause 'discursive closure'?

Toolkits often predetermine acceptable ethical categories—such as technical fairness, accountability, or transparency metrics. Consequently, critiques that lie outside these predefined technical boundaries (e.g., questioning corporate business models or societal power redistribution) are deemed out of scope and effectively silenced.

Can speculative ethics games drive real product changes in tech companies?

The authors found that speculative games are unlikely to generate direct changes in real-world products due to their detached, hypothetical nature; however, they serve as powerful probes that uncover teammates' critical orientations and help employees identify allies for collective action.

Related Research & Implementation Playbooks

Cross-disciplinary frameworks and practical blueprints connected to this insight.