AI's persuasive power unaffected by perceived national origin, study finds

A study published on arXiv found that the perceived national origin of a large language model does not significantly impact its persuasive power. In an experiment where participants debated a chatbot, actual attitude changes were similar regardless of whether the AI was labeled as American or Chinese. While participants initially showed less trust in the Chinese-labeled AI, this did not affect their willingness to assimilate its arguments. The research suggests that origin labeling alone may offer limited protection against foreign influence operations conducted via conversational AI. AI

IMPACT Suggests that transparency and origin labeling may be insufficient to mitigate foreign influence operations via AI.

RANK_REASON The cluster contains an academic paper detailing experimental research findings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• DiscoveryAI
• GPT-4o
• Jonas R. Kunst PhD
• large language models
• ZhengheAI

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →