Assisted Counterspeech Writing at the Crossroads of Hate Speech and Misinformation
Researchers have developed methods using Large Language Models (LLMs) to assist in writing counterspeech against online hate speech and misinformation. The study explored three strategies, including prompting LLMs with fact-checking and NGO guidelines, and a mixed approach combining both. While LLMs generated adequate counterspeech in 40% of cases, expert revisions significantly improved the output's quality and adherence to guidelines. AI
IMPACT Provides a framework for using LLMs to combat online toxicity, potentially improving moderation and reducing polarization.