Difference Without Domination

In the AI Story

The concept synthesizes Allen's engagement with classical republicanism—particularly the ideal of non-domination—and her work on cultural diversity and democratic citizenship in Talking to Strangers (2004). The formulation is deliberately precise. It refuses both the liberal universalism that ignores cultural difference in the name of neutral principles and the identity-based particularism that treats difference itself as the organizing political category. The target is the mechanism of domination, not the fact of difference.

For the AI moment, the principle exposes a specific failure mode of the dominant development paradigm. Current AI systems are built by a culturally homogeneous population—the technology industry of the American West Coast—and embed the values of that population in their defaults, training data, and optimization targets. The aesthetic of smoothness that Byung-Chul Han diagnosed is not a universal human value. It is a particular cultural preference that becomes domination when imposed on communities whose traditions define quality and excellence in different terms.

The principle also supplies the positive standard by which inclusion should be evaluated. Genuine inclusion means that diverse communities can engage with AI on terms that reflect their own governance traditions, their own definitions of quality, their own priorities—without the defaults embedded in the tools' design imposing one community's values on others. The developer in Lagos, the Indigenous governance council in New Zealand, the educator in Mumbai should each be able to use AI in ways that reflect their own understanding of what the technology should be for.

This is why Allen and her collaborators in the GETTING-Plurality network have organized their work around the concept of plurality rather than singularity. The institutional implications extend from training data governance to infrastructure investment to the governance of platform policies—all of which must be redesigned to serve difference without domination rather than homogenization in the name of efficiency.

Origin

Allen developed the concept across Talking to Strangers (2004) and Our Declaration (2014), and applied it directly to AI governance in her 2025 'Roadmap for Governing AI'. The formulation draws on the republican tradition of non-domination while extending it to the specifically cultural and epistemic dimensions of democratic life.

Key Ideas

Difference is irreducible. The democratic project does not require and should not attempt the elimination of cultural, linguistic, or epistemic diversity.

Domination is constructed. The conversion of difference into systematic advantage occurs through institutional choices that could be made otherwise.

Structural prevention. Democratic equality requires institutions designed to prevent specific mechanisms of domination, not merely the absence of overt discrimination.

Configurability requirement. AI systems must be designed so diverse communities can engage with them on terms that reflect their own values.

Governance diversity. Genuine inclusion means communities with diverse governance traditions have authority over how AI operates in their contexts.

Debates & Critiques

Critics from the universalist tradition argue that 'difference without domination' risks a fragmentation that undermines the common standards democratic deliberation requires. Allen's response, developed through her work on cross-difference deliberation, is that genuine common ground is built through engagement across difference rather than imposed through the suppression of difference—and that the alternative, whether explicit or implicit, is the domination of the community whose values happen to be treated as universal.