Multi-Method Analysis of Locally Built Artificial Intelligence Models Designed for Journalists in an African Country

Artificial intelligence (AI) models for journalism have been built within the Global North context, hence deepening the concerns of the “AI divide” between the North and South. This has compelled organizations in various areas of the Global South to develop journalism-focused AI models to address the specific challenges peculiar to journalism practice within the region. Nigeria stands out within the Global South context because organizations are developing local AI models to empower local journalists. However, the extent to which these AI products can address real-life journalistic challenges is unknown; hence, this study assesses how well locally developed AI models are positioned to support journalists in Nigeria, using a purposive sample of four homegrown tools: ChatJourno, Dubawa Chatbot, ChatVE, and MyAIFactChecker. Employing a mixed-methods design, we conducted walkthrough usability tests by engaging 15 trained journalists who provided rigorous assessments of the functionalities of each tool. We then conducted semi-structured interviews with the developers of each of the tools to assess each platform’s vision and directly cross-check those claims against the test-users’ evidence. Findings show the tools generally aim to complement Nigerian newsroom needs, such as improving news drafting and accelerating verification, but their realized value is uneven. While some of the platforms offer usable interfaces and services, some common limitations include gaps in up-to-date data and unusable features. We conclude that the tools must reliably achieve their promised functions because usability strongly drives user acceptance, and mass newsroom adoption will follow only when tools demonstrably enhance content standards in news practice.