Post-Truth in the Age of Post-Human: Comparative Assessment of Truth, Trust and Credibility of Generative AI and Nigerian Newspaper Reports

Nigerian survey respondents report some of the world’s highest trust in news and use AI chatbots for news heavily, yet evidence on generative AI’s news accuracy comes almost entirely from outside Africa. The purpose of this study was to compare the verified accuracy and the perceived truthfulness, trust and credibility of generative AI reports and Nigerian newspaper reports on the same events, and to test whether audience judgements track accuracy. A convergent parallel mixed-methods design combined a blind expert content analysis of 331 reports on 40 Nigerian events, from four national dailies and three generative systems, with a 2 × 2 (source × authorship disclosure) between-subjects experiment among 400 online news users quota-sampled across Nigeria’s six geopolitical zones. Significant issues appeared in 43.8% of AI reports and 11.6% of newspaper reports, mainly through weaker sourcing. Participants rated newspaper reports as more truthful, trustworthy and credible than AI reports. Disclosure sharply lowered the credibility of AI reports; concealment narrowed the gap by about two-thirds but did not close it. Positive machine-heuristic endorsement did not moderate the disclosure penalty. Accuracy and perception diverged in the reverse of the predicted direction: participants discounted unlabeled AI text more than its accuracy deficit warranted, although most could not identify its source. Exploratory genre analyses located the accuracy gap mainly in health and economy. The study gives newsrooms and media-literacy designers an African accuracy baseline, shows that AI labels carried a large credibility cost, and identifies a possible boundary condition on the machine heuristic.