大模型“眼睛雪亮” 谣言减少三分之二

Core Insights - Douyin has implemented AI large model technology for rumor governance, resulting in a 67% reduction in rumor exposure over the past quarter [1][2] - The AI model actively reviews trending information, analyzes content, and determines its authenticity, significantly reducing the processing time from days to hours [1] - The "AI Douyin Truth-Seeking" feature has been upgraded to provide users with links to comprehensive information on potentially misleading content [1][2] Group 1: AI Implementation and Effectiveness - Douyin's AI model can identify the credibility of information sources, including authoritative media and historical rumor databases [1] - The accuracy rate of the AI model in identifying rumors is currently at 85% and is expected to improve further [1] - The "AI Douyin Truth-Seeking" feature is estimated to achieve nearly 100 million exposures per day [2] Group 2: Rumor Management Strategies - Douyin has established a rumor governance ecosystem through various initiatives, including hot topic operations and collaboration with media [2] - The rumor operation team actively verifies the authenticity of information related to rumors, collaborating with local authorities to address specific cases [2] - Despite advancements, challenges remain in completely eliminating rumors due to subjective fabrication and the dynamic nature of information [3]