Artificial Intelligence (AI) Onto-norms and Gender Equality: Unveiling the Invisible Gender Norms in AI Ecosystems in the Context of Africa.

Abstract

Populations are impacted differently by Artificial Intelligence (AI), due to different privileges and missing voices in STEM space. Continuation of biased gender norms is exhibited through data and propagated by the AI algorithmic activity in different sites. Specifically, women of colour continue to be underprivileged in relation to AI innovations. This chapter seeks to engage with invisible and elemental ways in which AI is shaping the lives of women and girls in Africa. Building on Annemarie Mol’s reflections about onto-norms, this chapter utilized informal sessions, participant observation, digital content analysis, and AI model character analysis, to identify the gender norms that shape and are shaped by different AI social actors and algorithms in different social ontologies using Kenya and Ghana as case studies. The study examines how onto-norms propagate certain gender practices in digital spaces through character and the norms of spaces that shape AI design, training and use. Additionally the study explores the different user behaviours and practices regarding whether, how, when, and why different gender groups engage in and with AI-driven spaces. By examining how data and content can knowingly or unknowingly be used to drive certain social norms in the AI ecosystems, this study argues that onto-norms shape how AI engages with the content that relates to women. Onto-norms specifically shape the image, behaviour, and other media, including how gender identities and perspectives are intentionally or otherwise, included, missed, or misrepresented in building and training AI systems. To address these African women related AI biases, we propose a framework for building intentionality within the AI systems, to ensure articulation of women’s original intentions for data, hence abet the use of personal data to perpetuate further gender biases in AI systems.