The Impact of AI-Generated News on News Production and Consumption Patterns on Weibo

Abstract

This study uses 10,876 2023–2024 Weibo data points to explore AI-driven news generation ’s industry impact from a computer science lens. Natural language models (e.g., GPT) replace traditional workflows, shifting journalists to tuning and data annotation, yet fixed templates cause 13% content homogenization. Collaborative filtering exacerbates information bubbles via embedding matching; 18.1% of users feel deceived by unlabeled AI content due to deep learning's black box. Misuse of multimodal tools (e.g., DALL·E) reveals API flaws: 1.5% involve stolen conversation data, 6.8% of public safety risks link to deepfakes. To balance empowerment and risk, the study recommends blockchain content notarization, metadata annotation, and infrastructure-layer ethical filtering modules.

Ähnliche Arbeiten