2025, Vol. 6, Issue 2, Part A
Approaches computational misinformation detection vs. human media literacy: A systematic review of detection capabilities, limitations, and hybrid
Author(s): Rowland O Loveth and Igwenagu Emmanuel
Abstract:
Background: The proliferation of misinformation across digital platforms threatens democratic governance and social cohesion. Two primary response strategies have emerged: computational detection systems and human media literacy education programs.
Objective: This systematic review synthesizes literature comparing computational and human misinformation detection capabilities, examining performance patterns across content types, political contexts, and individual differences.
Methods: Literature search across multiple databases (2015-2024) identified 127 studies meeting inclusion criteria, including 45 experimental comparisons, 38 computational evaluations, 31 media literacy interventions, and 13 hybrid approach investigations.
Results: Computational systems demonstrate superior aggregate accuracy (75-85%) compared to humans (55-65%), with advantages for text-based content and political neutrality. Humans excel at novel misinformation tactics, contextual interpretation, and visual content analysis. Political ideology creates 15-25 percentage point disparities in human performance, while computational systems maintain consistency. Hybrid approaches achieve optimal performance (80-90% accuracy) through strategic integration.
Conclusions: Neither purely computational nor purely educational approaches suffice for comprehensive misinformation mitigation. Optimal strategies require sophisticated integration leveraging computational consistency and scalability while preserving human capabilities for contextual evaluation and novel pattern recognition.DOI: 10.22271/27084450.2025.v6.i2a.120
Pages: 40-47 | Views: 574 | Downloads: 214
Download Full Article: Click Here
