Aug 2026· Encyclopedia· Vol 6, pp. 182· 0 citations· 63 references
TL;DR
The review proposes a multilevel governance approach combining technical evaluation, documentation, professional and community oversight, appeals, remedies, and public-interest knowledge infrastructure to show why democratic and feminist governance can keep alternative technological futures open.
Abstract
Gender bias in generative artificial intelligence (GenAI) is both a technical and a social phenomenon: it emerges from historically patterned data, model design, and interactions in institutional use, and it cannot be understood by engineering or by social critique alone. This critical integrative review develops a more differentiated account. It connects feminist epistemology, Science and Technology Studies, critical AI scholarship, natural language processing, and governance research to examine five levels: historical knowledge production, technical representation and generation, benchmark evaluation, institutional deployment, and accountability. The review explains tokenization, next-token prediction, transformers, and the transition from static embeddings to contemporary language models before assessing evidence from standard fairness tests—coreference tests (WinoBias), sentence-pair tests (CrowS-Pairs), and stereotype tests (StereoSet)—as well as open-ended generation, multilingual testing, and text-to-image systems. It shows that measured bias varies with task, prompt, language, model version, and metric. What a test records and what that record means are therefore distinct questions: measurements are situated and depend on the instrument, and their interpretation draws on theory rather than following from the numbers alone. Evidence from employment, education, healthcare, and translation further indicates that the relevant unit of analysis is the model-in-context—the model together with the institution and workflow in which its outputs are used. Technical mitigation can reduce specific harms but does not repair unequal criteria, incomplete evidence bases, or weak institutional accountability. The review proposes a multilevel governance approach combining technical evaluation, documentation, professional and community oversight, appeals, remedies, and public-interest knowledge infrastructure. Its distinctive contribution is to connect three observations usually kept apart—how bias is measured, how generative systems concentrate epistemic authority, and how statistical learning is oriented toward past data—and to show why democratic and feminist governance can keep alternative technological futures open.
This study presents a bibliometric analysis of English-language, Scopus-indexed scholarship on cultural bias and ethical concerns in artificial intelligence (AI)-driven communication, covering 1,919 documents published between 2015 and 2025. Using co-citation and co-word analyses conducted in VOSviewer, the study maps the intellectual structure and thematic development of this interdisciplinary field. The findings indicate exponential growth in scholarly output, particularly from 2023 onward, coinciding with the widespread adoption of generative AI tools and large language models. Co-citation analysis identified five thematic clusters: fairness toolkits and justice frameworks; algorithmic bias and word embeddings; explainable AI and algorithmic accountability; critical studies of race and inequality; and philosophical and filtering-based sources of bias. Co-word analysis reveals that while terms such as "artificial intelligence”, “algorithmic bias”, and “machine learning” dominate the literature, the cultural dimensions of bias are frequently addressed in isolation from technical and ethical frameworks, suggesting a fragmented research landscape in which technical optimisation tends to receive more attention than deeper socio-cultural analysis. This study is limited to English-language publications indexed in Scopus and does not include grey literature, non-English scholarship, or other databases; findings should therefore be interpreted as reflecting English-language, Scopus-indexed research rather than the full body of global scholarship on the topic. The study contributes a structured, data-driven map of a rapidly expanding field and identifies gaps that future interdisciplinary research may address, including the need for integrated frameworks that bridge technical, cultural, and ethical dimensions of AI in communication.
V. Muriira, Venoth Nallisamy, J. Gikonyo et al.· Journal of Communication, La...· 0 citations
Through a new formalist reading of English- and Spanish-language short stories generated by ChatGPT-5, it is demonstrated that distributional regularities in language modelling scale upward, producing gendered and culturally normative stylistic patterns.
This critical narrative review synthesises evidence from ten recent publications in Scopus-indexed journals or conference proceedings, three directly relevant articles from Acta Humanitatis, DIALOGICA, and AI, and proposes a transparent protocol centred on identifiable texts, model and prompt documentation, repeated runs, preserved outputs, bilingual evaluation, linguistic evidence, negative cases, and explicit human responsibility.
Generative artificial intelligence (AI) has entered higher education quickly, and it has brought back old sociological questions about how academic misconduct gets defined, spread, and controlled. This paper does not treat AI-assisted writing as just a quicker way to plagiarise. Instead, it argues that generative AI shakes the very norms that academic integrity has always relied on. Using classical sociological theories of deviance — strain theory, neutralization theory, social learning theory, and institutional theory — along with recent research on AI in higher education, the paper looks at how AI blurs the line between help and authorship, how “cheating” is a socially built category applied unevenly, how detection tools carry linguistic and economic bias, and why institutional policies often look better than they work. The paper concludes that academic integrity is not something technology can simply threaten or destroy. It is something built continuously through institutional design, peer relationships, and resources that are not shared equally among students.
Keywords: generative artificial intelligence, academic integrity, sociology of deviance, higher education, academic dishonesty
Teena, Megha Sharma, Seema Boliya· International Journal For Mu...· 0 citations
Generative artificial intelligence (GenAI) is reshaping debates on digital inequality. It shifts attention from simple access gaps to differences in effective use, institutional capacity, and unequal outcomes. This study examines how the “GenAI divide” is constructed in high-visibility discourse on X (formerly Twitter) between 2022 and 2025. The study analyzes a purposively assembled, multilingual but English-dominant corpus of 332 high-engagement posts using an interpretive qualitative design supported by structured descriptive quantification. The posts were retrieved through tool-assisted search, semantic screening, and iterative filtering. Inductive thematic analysis and discourse analysis were used to identify the main dimensions of the debate and the framing, legitimating, and rhetorical strategies through which they circulated. The findings show that high-visibility GenAI divide discourse was organized around three thematic clusters: Access, capability, and distribution; political economy, data, and compute power; and societal and educational impacts. Over time, the discourse shifted from early concerns with paywalls, institutional privilege, and computational concentration toward stronger emphasis on unequal regional availability, labor-market disruption, and broader structural consequences. At the discourse level, early posts relied more heavily on high-certainty and dramatic rhetoric. Later posts increasingly drew on reports, institutional references, and evidence-oriented legitimation. The study suggests that, within high-visibility discourse on X, the GenAI divide is framed not simply as a question of who can access GenAI tools. Instead, it appears as a layered inequality formation shaped by infrastructure, governance, extraction, and unequal capacity to convert access into social and economic advantage.
As so-called artificial intelligence diffuses throughout society, a growing body of critical texts contests its dominant sociotechnical imaginaries. To understand how these critiques are organized, this paper reconstructs the internal grammar of a corpus of 252 English-language critical texts. Each text is assigned a node, the domain its critique is rooted in (society, capital, technology, or the nation-state), and a critique mode, the strategy it argues through (immanent, reframing, or structural). What each mode can and cannot contest is then examined. Across the four nodes, these texts contest the dominant AI imaginary as a self-justifying financial trajectory (capital), an autonomous cognitive trajectory (technology), an irreversible social force (society), and a governable geopolitical technology (state). The findings also indicate strong correlations across the texts examined here. Capital-focused texts lean overwhelmingly toward structural critique, technology-focused texts tend toward immanent critique, state-focused texts favor reframing, and society-focused texts split between reframing and structural approaches. The analysis points toward an asymmetry, indicating that the deeper the challenge to the imaginary's conditions, the less it registers within the vocabulary the imaginary itself recognizes as valid. State institutions sustain this asymmetry by keeping structural claims out of policy, and market culture by recasting them as matters of preference.
Nurullah Karaca· İstanbul Üniversitesi Sosyol...· 0 citations
