A new research paper explores how generative AI, particularly large language models, can influence collective opinion formation on online platforms. The study found that LLMs introduce directional biases when editing user-generated content on controversial topics, such as favoring gun control or opposing atheism. Researchers developed a mathematical model demonstrating that these AI-introduced biases can be amplified through social networks, ultimately shifting collective opinion. An audit of X's 'Explain this post' feature revealed a pro-life bias in Grok's outputs on abortion-related content, linked to specific design choices. AI
IMPACT AI's ability to subtly alter online discourse and collective opinion highlights the need for transparency and control mechanisms on digital platforms.
RANK_REASON Publication of an academic paper detailing AI's influence on collective opinion formation. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
