The AI Labor Legitimacy Paradox: A Governance Framework for ICT and Platform Firms

This is the authors' abstract. We don't add key points for this paper.

As ICT and platform firms deploy generative AI at scale, software-intensive work, machine-readable performance records, and network visibility make both productivity claims and workforce effects unusually observable. Capital markets may reward credible AI-attributed productivity, while labor, regulators, and communities may penalize visible AI-attributed displacement. We conceptualize this bounded contradiction as the AI Labor Legitimacy Paradox, an empirical instance of a stakeholder-induced legitimacy paradox that extends paradox and institutional-complexity research. Attribution asymmetry, skill hollowing, and talent-flight signaling sustain the tension, while a signed visibility gap describes the early excess of displacement attention over productivity attention. We derive a 2×2 response typology (Replacer, Quiet Cutter, Reskiller, Augmenter), five falsifiable hypothesis families, and an operational measurement design. Korean cases illustrate classification rather than causal effects. An analytical model and reproducible power calculations clarify comparative statics and empirical feasibility constraints. The framework predicts stronger effects in software- and platform-intensive settings with extensive algorithmic recording, network spillovers, employment protection, and evolving disclosure obligations. Technological capability alone may therefore be insufficient; firms must also maintain a social license to automate.