By Theresa Moses
Highly viral AI-generated content accounted for 48.80 per cent of analysts’ recommendations to unpublish content, compared with 44.52 per cent of recommendations to publish such content, according to data from the #TrackingNaija2027Elections project.
The tracking covered 376 content items, comprising 292 recommendations to publish and 84 recommendations to unpublish, with analysts assessing content across high, moderate and low virality categories.
The data shows that high-virality content represented the largest share of unpublish recommendations, at 48.80 per cent. It also accounted for 44.52 per cent of publish recommendations.
The difference between the two groups was 4.28 percentage points. The pattern was reversed for content with low and moderate virality.
Low-virality content accounted for 27.38 per cent of unpublish recommendations, compared with 28.42 per cent of publish recommendations, a difference of 1.04 percentage points.
For moderately viral content, the share was 23.80 per cent among unpublish recommendations and 27.05 per cent among publish recommendations, a difference of 3.25 percentage points.
The figures indicate that high-virality content occupied a larger share of unpublish recommendations, while low- and moderate-virality content accounted for larger shares of publish recommendations.
According to the tracking team, much of the highly viral content recommended for unpublishing demeans individuals and institutions and undermines the conduct of credible and peaceful elections.
The team attributed the spread of AI-generated content partly to social media algorithms and said removing content identified as harmful would require significant cooperation from platform owners, alongside a commitment to credible and peaceful elections.
However, the figures show how recommendations were distributed across virality categories; they do not establish that highly viral content was more likely to be recommended for removal overall. That conclusion would require the total number of items assessed in each virality category and the number recommended for publication or removal within each category.
The findings form part of TrackingNaija2027Elections, a collaborative initiative involving CEREDEMS-AFRICA, the University of Ilorin and the Institute of Geopolitical Communications at RUDN University.
GATMASH NEWS
…News That Drives Impact
For advert placements and business inquiries, kindly reach us via:
📧 Email: [email protected]
📞 Tel/WhatsApp: +234 (0) 807 202 2024
We value your voice! Share your stories, eyewitness reports, and contributions with us anytime:
📲 SMS/WhatsApp: +234 (0) 807 202 2024
📧 Email: [email protected]
Visit us online: https://gatmash.com
