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

Power dynamics significantly shape who feels empowered to raise ethical concerns: The study found that within corporate teams, existing hierarchies and power relations determined whether team members felt they had a "license to critique" AI ethics issues, meaning that junior staff or those with less organizational power were often reluctant to voice concerns even when they had them.

Standard AI ethics interventions may actually limit rather than expand discussion: The research demonstrates that pre-designed ethics tools like checklists and toolkits can inadvertently narrow the scope of what counts as legitimate ethical concern, potentially silencing important critiques that fall outside the intervention's defined boundaries.

Games and hypothetical scenarios have limited direct impact on real-world practice: While the game-based intervention successfully triggered broader ethical discussions compared to teams' typical conversations, the authors found that "games are unlikely to enable direct changes to products or practice," suggesting that alternative approaches may be needed to translate ethical discussions into actual organizational change.

"Scope" definitions matter for determining what gets discussed: Teams often operate with implicit or explicit understandings of what issues are "in scope" for their AI ethics discussions, and this framing affects which concerns are considered legitimate to raise—a finding relevant to how organizations design their ethics review processes.

Games may be most valuable for building coalitions rather than immediate change: The study suggests that while games didn't directly change team practices, they were "more likely to allow members to find critically-aligned allies for future collective action," indicating their potential value as a tool for identifying and connecting people who share ethical concerns.