Source-linked AI summary
Lost in Translation: How Universal Ethical Values Fail to Translate Across Global Contexts
Ozioma C. Oguine, Munachimso B. Oguine, Cesar Cervera, Jenny Yang, Pooja Voladoddi, Mario Rodriguez, Saif Eddin Bani Malhem, Karla Badillo-Urquiola, Daricia Wilkinson
TL;DR
The paper addresses the limited empirical understanding of how supposedly universal AI ethics values are interpreted across diverse cultural contexts. Using qualitative interviews with 14 experts across 10 countries, it finds translation gaps between global frameworks and situated practices, and identifies plural governance pathways grounded in local knowledge and distributed authority.
Problem
Existing AI ethics work often defines universal values or evaluates technical implementations, leaving limited empirical understanding of how experts interpret and adapt those values across contexts.
Method
The study conducts qualitative interviews with 14 experts across 10 countries and uses iterative coding, reflexive thematic analysis, and cross-context comparison.
Results
Experts reinterpret fairness, privacy, and transparency as relational and context-dependent, revealing translation gaps between universal principles and situated practices.
Takeaways & Limitations
AI governance should emphasize local knowledge, participatory processes, and more distributed authority to support context-sensitive systems.
Takeaways & Limitations
The purposive sample of 14 experts is not intended to be generalizable or representative of the cultures in which participants are situated.
Abstract
from arXiv · showhide
AI ethics frameworks treat values such as fairness, transparency, and accountability as universal and uniformly operationalizable across contexts. We examined how 14 experts across 10 countries made sense of AI in practice, reinterpreted core values, and envisioned governance alternatives. We found that AI deployment is characterized by structurally unequal conditions, marked by infrastructural constraints, extractive practices, and a "mystification" of technology, which fundamentally shape perceptions of risks and opportunities. Our findings reveal that experts reinterpret values to fit local moral logics: privacy as collective and relational rather than individual; transparency as trust-building accountability rather than technical disclosure; and fairness as equity in access and representation rather than parity in outcomes. We identify these as translation gaps between encoded global frameworks and situated local practices. Finally, we propose pathways toward plural governance that redistributes epistemic authority and treats ethical negotiation as an ongoing, context-sensitive process rather than a settled technical standard.
Introduction
The paper examines how AI ethics values presented as universal are interpreted and adapted across cultural contexts. It studies experts’ situated experiences and proposes translation-sensitive, plural governance approaches.
- AI ethics frameworks articulate values such as fairness, accountability, transparency, privacy, and safety as broadly applicable guidelines.
- Existing critiques argue that ethical values are interpreted, prioritized, and operationalized differently across communities shaped by historical inequities, cultural norms, and global power asymmetries.
- The study addresses limited empirical understanding of how experts across diverse contexts engage with and adapt dominant AI ethics frameworks in practice.
- Experts are defined as people with substantive AI engagement and deep familiarity with the cultural contexts where AI is designed, deployed, or governed.
- Interviews with 14 experts across 10 countries examine how infrastructural constraints, competing narratives, and power dynamics shape experiences of AI and interpretations of fairness, privacy, and transparency.
- The paper’s central contribution is the AI Ethics Translation Model, which explains how universal principles acquire situated meanings and produce translation gaps affecting governance outcomes.
Related Work
Related work documents the expansion of universal AI ethics frameworks and critiques their concentration of epistemic authority. Alternative scholarship develops culturally grounded approaches that challenge this universalism.
- International and governmental frameworks have proliferated around universal commitments including transparency, fairness, accountability, human rights, dignity, and sustainability.
- Critiques argue that global AI ethics can privilege particular ways of knowing and valuing, producing epistemic injustice and ethics washing.
- Culturally grounded AI ethics scholarship articulates alternative epistemologies rooted in specific traditions, including African philosophies emphasizing interconnectedness and communal flourishing.
- Prior research finds that concentrated governance and context-specific systems or policies can fail to account for diverse social norms, infrastructures, and knowledge systems.
Method
The study uses a qualitative, interpretivist design combining cross-context expert interviews, structured value probes, open-ended questioning, and reflexive thematic analysis. Recruitment and analysis were designed to capture culturally situated interpretations while remaining reflexive about researcher positionality.
- The researchers used a qualitative, interpretivist approach to examine how experts interpret, negotiate, and envision culturally grounded AI ethics.
- Participants were recruited purposively and conveniently based on AI expertise and deep familiarity with a specific cultural context.
- The final sample comprised 14 experts from 10 countries across Europe, North America, South America, Africa, Oceania, Asia, and the Caribbean.
- Semi-structured remote interviews lasted 45–60 minutes and used probes on bias, explainability, fairness, transparency, privacy, risks, opportunities, and culturally grounded governance.
- Analysis Approach: Analysis combined independent transcript review, open coding, iterative code refinement, constant comparison, and reflexive thematic analysis across cases.
- The study received ethics approval, obtained participant consent, and emphasized voluntary participation with the option to pause, skip questions, or stop.
- Positionality Statement: The authors’ diverse cultural backgrounds and reflexive discussions informed interpretation of cultural context, power dynamics, historical inequities, and analytic assumptions.
Findings
Participants described AI deployment as uneven and shaped by infrastructural constraints, extractive practices, power asymmetries, and competing narratives. They also identified context-dependent opportunities and interpreted ethical values through local conditions rather than as fixed universal standards.
- Risks in AI Deployment: AI access was structured by unequal infrastructures, leaving communities without stable electricity, connectivity, or devices peripheral to meaningful engagement.Development and deployment decisions favored already-resourced environments, reproducing inequalities in who benefits from AI.
- Opportunities in AI Deployment: Participants identified four context-dependent opportunities: expanding service and information access, supporting learning, improving everyday efficiency, and enabling community empowerment.These benefits emerged when AI aligned with local needs, constraints, priorities, and opportunities for local participation.
- Risks in AI Deployment: Participants described AI systems as extractive, relying on data and labor practices that prioritize efficiency while undermining consent, accountability, and worker protection.Data annotation and content moderation were portrayed as underpaid, precarious, psychologically harmful, and systematically undervalued.
- Risks in AI Deployment: Representation gaps in languages, accents, and cultural knowledge produced practical exclusion and epistemic harm by prioritizing some communities’ knowledge over others.Participants linked these gaps to broader inequalities in data ownership, labor, power, and representation.
- Opportunities in AI Deployment: Participants described AI as supporting learning when augmenting existing practices, extending essential services through available infrastructures, and improving convenience in everyday tasks.AI was framed as complementary rather than replacing established practices, and practical utility could motivate adoption without deep technical understanding.
- Ethical Values Interpreted Differently Across Cultural Contexts (RQ2): Participants interpreted core ethical values through cultural norms, lived realities, and structural conditions, identifying privacy, transparency, fairness, and explicability as context-sensitive concepts.The analysis identified five themes, including privacy as collective and relational practice, transparency as trust and accountability, and fairness as equity and structural justice.
Discussion
The findings challenge universalist AI ethics by showing that values require contextual translation and that governance must address power, epistemic authority, and local meaning-making. They therefore support plural, adaptive approaches spanning system design, policy, participation, and data governance.
- Universalism and translation: Participants treated ethical values as situated systems of meaning shaped by local communities, rather than uniformly operationalizable principles.Their interpretations included locally rooted values such as community, respect, and collective well-being.
- Universalism and translation: Translation gaps are material as well as interpretive, producing misclassified behaviors, trust breakdowns, and governance that feels externally imposed.These gaps shape how systems are adopted, trusted, and resisted.
- Power-aware governance: AI governance is constrained by concentrated epistemic authority, so participation alone is insufficient without redistributing decision-making power across value systems.Plural governance coordinates diverse perspectives without enforcing a single interpretation of ethical values.
- Design implications: AI systems should support contextual interpretation in use and systemic contestability instead of treating ethical values as fixed inputs.This design implication responds to the persistence of static representations of harm and fairness.
- Governance and policy implications: Governance frameworks should enable local interpretation and operationalization while maintaining broader coordination across levels.The discussion frames subsidiarity as an alternative to one-size-fits-all governance.
- Participation and stakeholder engagement: Meaningful participation requires co-construction and the ability to influence decisions, rather than consultation or inclusion alone.The proposed shift addresses superficial participation in AI governance.
- Data practices and governance: Data governance should treat data as socially and culturally embedded and support data sovereignty against extractive and aggregative practices.This reframes data from neutral input toward relationships and meanings situated in context.
Limitations and Future Work
The study’s qualitative sample and self-reported accounts provide situated, theoretically transferable insights rather than representative or directly generalizable findings. Future research should broaden stakeholders, observe practice over time, and develop concrete plural-governance mechanisms.
- Scope and sample: The purposive sample of 14 experts supports depth but is not intended to produce generalizable or culturally representative findings.The perspectives reflect participants’ professional and lived experiences.
- Evidence base: Self-reported accounts may differ from how AI ethics is negotiated in practice, motivating ethnographic or longitudinal research.Such studies could examine these dynamics over time.
- Future governance research: The paper remains primarily analytical and leaves governance mechanisms for balancing local values with globally shared human-rights commitments for future work.This is identified as a future research priority.
Conclusion
This paper examines how experts across diverse contexts interpret AI ethics and envision culturally grounded governance. It finds that ethical values require translation across contexts, supporting local knowledge, participation, and plural, power-aware governance.
- Conclusion: Experts’ interpretations show that widely cited AI ethics principles, including fairness, are shaped by contextual and relational factors.The conclusion presents this as a finding from examining diverse expert perspectives.
- Conclusion: The central challenge is translating shared values across contexts rather than merely defining them.Translation gaps contribute to misalignment between systems and the communities they serve.
- Conclusion: Participants point toward local knowledge, participatory processes, and distributed governance as pathways for more culturally grounded AI governance.The conclusion reframes governance as ongoing negotiation across diverse contexts.