TLDR: A study surveyed 157 AI professionals and academics from 25 African countries to understand their concept of trust in AI. It found that trust varies by application, with less trust for personal data uses. Values are heavily influenced by communal relations and African culture, leading to a nuanced understanding of trust that blends technical reliability with human-like qualities like honesty and goodwill. The research highlights the need for AI development to consider diverse, non-Western perspectives on trust.
The global conversation around Artificial Intelligence (AI) often emphasizes its trustworthiness as crucial for widespread adoption and effective application. However, a significant portion of existing research on trust in AI has been concentrated in Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies. This focus has led to a gap in understanding how trust in AI is conceptualized in other diverse cultural contexts, particularly in Africa, a continent rich in varied perspectives and languages.
Understanding Trust in AI from an African Perspective
A recent exploratory study, titled “Enriching Moral Perspectives on AI: Concepts of Trust amongst Africans,” sought to bridge this knowledge gap. Conducted by Lameck Mbangula Amugongo, Nicola J. Bidwell, and Joseph Mwatukange, the research surveyed 157 individuals with professional or educational interests in AI from 25 African countries. The primary goal was to explore how these professionals and academics conceptualize trust in AI, moving beyond the prevalent Western-centric views.
Key Findings: How Africans View AI Trust
The study revealed that trust in AI systems is not uniform; it varies significantly depending on the specific application. Respondents expressed higher levels of trust in AI used for tasks that ostensibly involve less personal data, such as meteorology, industrial applications, or language translation. Conversely, there was a noticeable decrease in trust when AI systems were involved in scenarios handling personal or private information, such as in finance, government decisions, employment, or crime identification. This suggests a heightened sensitivity and concern when AI directly impacts individual lives, privacy, and societal structures.
The researchers also found that respondents’ educational attainment and their experiences of living and working in different countries (transnational mobility) played a role in shaping their concerns about AI systems. For instance, individuals with higher educational qualifications were more likely to express worries or fears regarding AI. These factors also influenced which principles designed to foster trust were emphasized, indicating a nuanced understanding of trustworthiness beyond universal assumptions.
The Role of Community and Culture
A profound and central finding of the study was the significant influence of community and African culture on the respondents’ values. The majority of participants indicated that their values were guided by the communities in which they grew up, consistently emphasizing communal relations over individual freedoms. This perspective aligns with Afro-relational philosophies, which conceptualize humanness as emerging from interconnected relationships between people.
This Afro-relational view deeply influences how trust is understood. While respondents often applied nuances of Afro-relationalism to constructs prevalent in international AI discourse, such as reliability and reliance, they imbued these terms with a sense of interdependence and moral conduct towards others. For example, reliability might be interpreted not just as consistent performance, but also as a commitment to the community’s well-being.
Beyond Technical Metrics: Human and Moral Dimensions of Trust
When asked to describe what trust means to them, respondents frequently combined system-like constructs (e.g., reliability, consistency, transparency, accountability) with human-like qualities (e.g., honesty, integrity, goodwill). They often linked trust to morality, good intentions, and a felt experience of safety or comfort, which they noted could be formed by shared values and beliefs. Trust was also described as an active, bidirectional relationship that evolves over time through consistent actions and reciprocity.
Also Read:
- Shaping AI’s Future: An African Perspective on Safety Challenges and Solutions
- A New Framework for Evaluating AI Training Data Trustworthiness
Implications for Global AI Development
The insights from this study underscore that a “one-size-fits-all” approach to AI trust, often derived from Western societies, may not be universally applicable or sufficient. The diverse and nuanced ways in which African professionals conceptualize trust highlight the importance of incorporating local values and experiences into AI design, deployment, use, and governance. Understanding these contextual differences is crucial for developing AI systems that are not only technically sound but also culturally resonant, equitable, and truly trustworthy for diverse populations worldwide.
For more in-depth information, you can read the full research paper: Enriching Moral Perspectives on AI: Concepts of Trust amongst Africans.
