Purpose: This study examines how artificial intelligence (AI) is discursively constructed in international and national policy documents, and how these representations relate to power structures linked to class, gender, and ethnicity within organizational contexts.
Method: Drawing on critical discourse analysis and an intersectional framework, the study analyzes 31 publications issued between 2018 and 2025 by institutions such as the OECD, ILO, EU, UNESCO, and UN Women.
Empirical data: The analysis identifies three dominant patterns:
The Discourse of Neutrality, in which AI is portrayed as objective, efficient, and apolitical obscuring the political and historical roots of inequality. A Digital Class System, where technical competence is assigned symbolic capital and legitimacy, while algorithmic surveillance reinforces employer control and reduces employee autonomy. Equality as Representation, where gender equality is reduced to numerical representation without power analysis, and racism is often replaced by colour-blind indicators (income, education, geography) or encoded through proxy variables.
Conclusions: In summary, AI systems are described not merely as advanced technologies, but as mechanisms that reinforce and legitimize existing hierarchies, rather than functioning as neutral infrastructures. The study introduces an intersectional power matrix for AI in organizations (Symbol/Role/Governance) and argues for governance that promotes equality, design models that redistribute power beyond training initiatives and ethical guidelines.
Limitations include a narrow time frame and exclusive focus on policy texts; future research should therefore combine discourse analysis with ethnographic and quasi experimental approaches within workplace practices. Furthermore, future research should examine the phenomenon in additional communicative contexts, particularly within media communication, such as news reporting and other forms of public information dissemination.