AI and African Literary Studies

This working paper aims to disentangle the problems and potentials of AI technology as they pertain to the field of African literary studies. It does so because African literature is particularly vulnerable to the biases of literary data created in and for the global north. Calls, over the last decade, to decolonise literary studies have reached an impasse in terms of the continued dominance of the English language in the field, as well as its heavy reliance on commercially produced literature from the global north. The commercial and linguistic bias of the literary data being used to train Large Language Models (LLMs) redoubles the problem of decolonisation in a time of AI. Yet, as this paper argues, the field of African literary studies can also harness the power of AI tools to ensure that the African literary data available for LLM training is diverse, Africa-based, and not only determined by commercial factors.