Answers are grounded in this paper and cite their sources.
Artificial intelligence is widely framed as transforming journalism through efficiency, automation, and innovation. This paper argues that, for journalists in the Global South, AI is also reshaping whose knowledge is recognized, whose reporting is discoverable, and whose stories become part of the historical record. Drawing on practitioner experience and scholarship across journalism studies, AI ethics, and decolonial theory, it examines how AI journalism tools reproduce structural inequality within the profession. Three interconnected dimensions of exclusion are identified: language, as large language models underrepresent non-English text and non-Western narrative traditions; sourcing, as limitations in named entity recognition and knowledge representation may disadvantage institutions, experts, and knowledge holders outside the Global North; and story, as newsworthiness classifiers trained on Western editorial outputs may encode culturally specific definitions of news value. The paper argues that these failures are structural consequences of AI development conducted without meaningful participation from the journalists most negatively affected, rather than isolated technical shortcomings. The argument is conceptual. It builds on existing evidence and identifies mechanisms that require empirical testing. The underrepresentation of Global South journalism therefore constitutes both a media equity crisis and a threat to the historical record, as stories that algorithmic systems cannot recognize or retrieve risk going unpublished, unindexed, or hard to find in the digital archive. Bringing decolonial theory into conversation with emerging African and Global South scholarship on AI and journalism, the paper proposes a decolonial framework for AI in journalism and calls for a research agenda centered on the experiences of Global South practitioners.
No takes yet. Share an insight, caveat, or question.
Mercy Adhiambo Orengo (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: