Click the star to add/remove an item to/from your individual schedule.
You need to be logged in to avail of this functionality.

Accepted Paper:

Epistemic decolonization: towards different sociotechnical visions of AI in Africa  
Yousif Hassan (University of Michigan - Ann Arbor)

Send message to Author

Short abstract:

I discuss sociotechnical visions of African scientists and practitioners working on NLP models of local Indigenous languages. I examine their discourses and practices and highlight how histories of epistemological decolonization are informing some of local AI development approaches in the continent.

Long abstract:

With the global push for technological diffusion of AI, machine learning, and big data to the Global South/Majority World, Africa finds herself once again pushed into contested discourses and fragmented practices of playing catchup with the West. These narratives subscribe to dominant conceptions of progress and development and assumptions of human and technological lacking that long described the continent. However, a growing number of African scholars and scientists are asking different questions that challenge old orthodoxies of technological innovation and economic and social development. They are attempting to rethink AI technology in the local context while reimagining the ways in which technology can contribute to the prosperity of their local communities.

In this presentation, I explore their sociotechnical visions of AI development, drawing on multi-sited ethnographic work across multiple African countries. I follow African scientists and practitioners working on natural language processing models for African Indigenous languages and examine their discourses and practices through the analytical lens of decoloniality in Africa.

Historically, decolonization meant the complete overthrow of colonial structures, institutions, and ideas of Western modernity in post-colonial Africa. I highlight how these histories are informing their AI local development efforts and inspiring different AI approaches using epistemological underpinnings based on African communal practices and Indigenous ways of knowing and being in the World. I argue for the need to look at histories of decolonization in the continent to restore African knowledges and cultural heritage in the current moment of technoscientific capitalism and global AI development.

Traditional Open Panel P195
Making and doing AI from Africa: critical insights on AI and data science
  Session 2 Tuesday 16 July, 2024, -