Egalitarian Response to AI

In the AI Story

Every risk perception has a characteristic sensitivity and a characteristic blindness. The egalitarian is sensitive to the way AI reproduces and amplifies existing inequalities: the training data that encodes historical bias, the compute infrastructure concentrated among a handful of firms, the English-language hegemony of frontier models. These are real concerns, and the egalitarian is often right to raise them first and loudest. The democratization of capability that the Orange Pill celebrates is partial, and the egalitarian names the partiality.

The blindness that matches the sensitivity is the tendency to locate all agency in structure and little in the individuals who navigate it. The egalitarian risk portfolio makes it difficult to acknowledge that the same technology that concentrates power can also redistribute it, that the developer in Lagos with a Claude subscription has more leverage than the developer in Lagos without one, even if both are disadvantaged relative to San Francisco. The distribution problem is real, but it is not identical to the concentration problem, and the egalitarian reading often collapses the two.

The Luddite chapter of the Orange Pill is recognizably egalitarian in its analysis: the factory owners captured the productivity gains while the weavers lost their livelihoods. Wildavsky's reading of the Luddites confirms the distributional diagnosis but rejects the strategic conclusion. Machine-breaking did not produce the institutional structures that eventually distributed the gains more broadly; it produced criminalization. The distributive victory, when it came, came through institutional construction — labor movements, voting rights, welfare states — not through refusal of the technology.

The egalitarian response to AI is currently ascendant in European regulation, visible in the EU AI Act's focus on high-risk applications and algorithmic accountability. Whether this framework will produce distributive benefits or institutional sclerosis is the empirical question that the next decade will answer. Wildavsky would have predicted mixed results, sensitive to which feedback mechanisms the regulation supported and which it suppressed.

Origin

The egalitarian position is the cultural home of environmental, civil rights, and labor movements. Applied to technology, it produced the concerns about automation that animated New Left critiques in the 1960s and 1970s, and that have been revitalized in the AI discourse since 2015.

Key voices in the contemporary egalitarian reading of AI include Timnit Gebru, Emily Bender, Safiya Noble, and Kate Crawford — each of whom has emphasized distributional and representational harms that purely technical framings of AI safety systematically miss.

Key Ideas

Power concentration is the primary risk. AI is diagnosed through its effects on who gets to build, who gets to benefit, and who gets to decide.

Structure over individual. Outcomes are explained by institutional arrangements rather than by choices of individuals operating within them.

Representation matters. Training data, model evaluators, and governance bodies that exclude affected communities produce systematically biased technologies.

Redistribution as remedy. Antitrust, data rights, and democratic governance are the preferred interventions.

The Luddite diagnosis was right. The technology did concentrate power; the failure was in the strategic response, not the risk perception.

Debates & Critiques

The sharpest internal debate among egalitarians concerns whether AI should be resisted, regulated, or appropriated. Resistance (the neo-Luddite position) holds that the technology is irredeemable; regulation (the EU position) holds that it can be made compatible with democratic values through institutional constraint; appropriation (the cooperative position) holds that the technology itself should be brought under democratic ownership. The three strategies produce very different political programs.