Fake or fact? Evaluating chatbots’ performance to help users detect fake news in human-computer communities

Abstract

The rapid proliferation of fake news poses a significant challenge to information ecosystems, particularly in digital and social media environments. This study investigates the effectiveness of chatbot interventions in assisting users with fake news detection within human-computer communities. Grounded in the Heuristic-Systematic Model, the study employs a 6 (fake news type) × 3 (chatbot intervention strategy) mixed design to examine how different chatbot strategies - fact-checking, contextual explanations, and authority endorsements - affect users’ ability to identify various types of fake news. The results show that fact-checking is most effective for detecting fabrication and photo manipulation, contextual explanations enhance recognition of satire and parody-based fake news, and authority endorsements are particularly useful in countering propaganda. These findings highlight the importance of tailoring chatbot interventions to specific fake news types.