Counterspeech for Mitigating the Influence of Media Bias: Comparing Human and LLM-Generated Responses

2025-08-20

AI Summary

Media bias and hostile reader comments mutually reinforce each other, exacerbating societal polarization and harming marginalized groups—necessitating interventions that balance free expression with effective mitigation. Method: This work presents the first systematic study of counterspeech in news contexts to suppress bias propagation, introducing the first fine-grained annotated dataset linking media bias, hostile comments, and counterspeech. We propose a large language model–based generation framework integrating news contextual information and few-shot learning to enhance politeness, relevance, and diversity of generated counterspeech. Contribution/Results: Experiments show that over 70% of hostile comments indeed amplify bias; our method effectively curbs their propagation. The approach offers a scalable, interpretable, content-aware pathway for bias mitigation, advancing responsible content governance in digital news ecosystems.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Responsible Web: Mitigating misinformation and disinformation
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Biased news contributes to societal polarization and is often reinforced by hostile reader comments, constituting a vital yet often overlooked aspect of news dissemination. Our study reveals that offensive comments support biased content, amplifying bias and causing harm to targeted groups or individuals. Counterspeech is an effective approach to counter such harmful speech without violating freedom of speech, helping to limit the spread of bias. To the best of our knowledge, this is the first study to explore counterspeech generation in the context of news articles. We introduce a manually annotated dataset linking media bias, offensive comments, and counterspeech. We conduct a detailed analysis showing that over 70% offensive comments support biased articles, amplifying bias and thus highlighting the importance of counterspeech generation. Comparing counterspeech generated by humans and large language models, we find model-generated responses are more polite but lack the novelty and diversity. Finally, we improve generated counterspeech through few-shot learning and integration of news background information, enhancing both diversity and relevance.

Problem

• Counterspeech combats media bias reinforced by offensive comments
• Comparing human versus LLM-generated counterspeech effectiveness
• Improving counterspeech diversity through few-shot learning

Innovation

• Counterspeech generation using large language models
• Few-shot learning for improving response diversity
• Integration of news background information for relevance