Artificial intelligence (AI) systems are reshaping industries and governance across the globe, but their trustworthiness is largely studied through Western lenses. A new study submitted on arXiv challenges that narrative by placing African voices at the center of this critical discourse.
Titled “Enriching Moral Perspectives on AI: Concepts of Trust amongst Africans”, the research examines how African professionals and students with educational or professional ties to AI conceptualize trust in AI systems. Spanning respondents from 25 countries, the study delves into how cultural values, education, and transnational experiences shape perceptions of trust and distrust in AI applications.
How trust in AI shifts across application contexts
The survey of 157 participants revealed clear distinctions in levels of trust depending on the context of AI use. Applications perceived as less invasive, such as those in meteorology, industrial operations, transportation, or language translation, garnered higher levels of confidence. Conversely, AI systems involving sensitive personal or biometric data, particularly in finance, government decision-making, law enforcement, and employment screening, generated significant skepticism.
Participants residing in their native countries expressed comparatively higher trust in AI-driven medical systems, such as those used by local doctors or national hospitals. By contrast, Africans working abroad showed heightened caution in trusting healthcare AI, reflecting greater awareness of data bias and discrimination in global healthcare environments.
Education was a strong predictor of how individuals assessed AI trustworthiness. Respondents with postgraduate degrees expressed deeper concern about the ethical and operational risks of AI systems, including data misuse, security vulnerabilities, and opaque decision-making. This divergence suggests that international exposure and advanced education sharpen awareness of the limitations and risks embedded in current AI systems.
Cultural values shape interpretations of trust
A defining insight from the study is the profound role of communal and cultural values in shaping African conceptualizations of trust. Participants overwhelmingly indicated that their sense of trust is rooted in the communities where they grew up, alongside African cultural traditions and religious influences. This orientation frequently prioritized communal responsibility and mutual dependence over individual autonomy, reflecting principles of Afro-relational ethics.
Respondents with transnational mobility often demonstrated a heightened recognition of these communal values, likely due to the contrast experienced when navigating more individualistic societies abroad. However, variations emerged across countries. For instance, professionals from Namibia and South Africa leaned more toward collective responsibility, while those from Nigeria and Zambia exhibited stronger tendencies toward individualist frameworks in their interpretation of trust and autonomy.
This cultural grounding translated directly into how trust constructs were understood. Terms commonly used in global AI ethics discourse, such as accountability, reliability, and explainability, were adapted to local contexts. Reliability, for instance, was often interpreted through the lens of interpersonal relationships, incorporating notions of mutual goodwill, respect, and shared responsibility rather than mere technical performance or consistency.
The study also found that while respondents were aware of global narratives around fairness and bias in AI, they more frequently emphasized privacy, security, and accountability. This suggests that local realities, such as data vulnerabilities and governance gaps, shape priorities differently than in Western contexts.
Redefining global trust frameworks through African lenses
The research highlights the limitations of universal frameworks for operationalizing AI trust. While constructs like transparency, accountability, and fairness dominate policy and industry discussions in WEIRD societies, they often fail to capture the nuanced, relational dynamics that define trust in African contexts.
Open-ended responses collected during the survey revealed that trust was not only associated with system attributes like accuracy and reliability but also with human-like qualities such as honesty, integrity, and mutual respect. Trust was often described as a bidirectional relationship that evolves over time, underscoring the importance of reciprocity and emotional resonance in human-technology interactions.
These insights carry significant implications for AI governance and system design. In sectors such as healthcare, trust was linked to familiarity and long-term relationships with service providers. This dynamic indicates that initiatives seeking to integrate AI into critical services in Africa must prioritize local engagement and co-design with communities to build authentic trust.
Moreover, the study underscores that values surrounding trust are not static but situated within dynamic social and cultural contexts. Participants frequently referenced the “felt” dimensions of trust, including feelings of safety, comfort, and shared belonging, highlighting that trust extends beyond rational assessments of system performance.
Towards context-sensitive AI policies
The authors argue that recognizing these culturally grounded perspectives is essential to developing more inclusive, effective AI policies and systems. By incorporating Afro-relational ethics, global frameworks can move beyond the oversimplification of trust as a purely technical issue and acknowledge the complex interplay of moral, social, and experiential factors that inform how people interact with AI.
The study also calls for deeper, qualitative investigations into how trust is enacted in daily interactions with AI technologies across African societies. Current findings, while groundbreaking, are limited by the scope of the sample and the reliance on English-language surveys, which may have excluded insights from Francophone and Arabic-speaking regions. Expanding research to include local languages and more diverse methodologies, such as ethnographic studies, could provide richer understandings of trust in different African contexts.
For policymakers, developers, and researchers, the message is clear: one-size-fits-all approaches to AI trust are inadequate. Designing systems that align with the lived realities and values of African communities will not only enhance trust but also promote more equitable outcomes in the deployment and governance of AI.
