Algorithmic Ethnocentrism: The Challenge of AI Bias in Cultural and Kinship Modeling

Recent studies in anthropology have highlighted a growing concern termed 'algorithmic ethnocentrism.' This refers to the systemic bias within Artificial Intelligence (AI) models that fail to recognize or accurately represent non-Western kinship systems. While Western societies predominantly follow patrilineal or bilateral descent and descriptive kinship terminologies, many indigenous and non-Western cultures utilize matrilineal descent, double descent, or classificatory terminologies. The core of the issue lies in the training data and the underlying logic of AI algorithms, which are often rooted in Western cultural frameworks. For instance, AI models frequently struggle to categorize 'classificatory' systems where a single term (like 'mother') might apply to a group of female relatives, or where descent is traced exclusively through the female line. By defaulting to Western norms, these technologies effectively 'erase' the social complexities of indigenous groups, violating the anthropological principle of cultural relativism—the idea that a culture should be understood on its own terms rather than judged by the standards of another.