分析AI代理在Moltbook上的政治宣传，发现少数代理制造了大部分内容。
Large-Scale Analysis of Persuasive Content on Moltbook
- 用大模型分类器识别政治宣传，经专家验证准确率高
- 1%的帖子是政治宣传，占政治内容的42%，集中于5个社区
- 4%的代理贡献了51%的宣传内容，且重复发布相似内容
我们对Moltbook（一个类似Reddit的AI代理平台）上的政治宣传进行了基于自然语言处理的大规模研究。为实现大规模分析，开发了基于大模型的分类器以检测政治宣传，并通过专家标注验证，一致性系数κ为0.64-0.74。利用包含673,127篇帖子和879,606条评论的数据集，发现政治宣传占所有帖子的1%，但占所有政治内容的42%。这些帖子高度集中在少数社区，其中70%来自五个特定社区。仅有4%的代理生产了51%的政治宣传内容。此外，发现少数代理在多个社区内反复发布高度相似的内容。尽管如此，评论对政治宣传的放大作用有限。
原文摘要 · Abstract (English)
We present an NLP-based study of political propaganda on Moltbook, a Reddit-style platform for AI agents. To enable large-scale analysis, we develop LLM-based classifiers to detect political propaganda, validated against expert annotation (Cohen's $κ$= 0.64-0.74). Using a dataset of 673,127 posts and 879,606 comments, we find that political propaganda accounts for 1% of all posts and 42% of all political content. These posts are concentrated in a small set of communities, with 70% of such posts falling into five of them. 4% of agents produced 51% of these posts. We further find that a minority of these agents repeatedly post highly similar content within and across communities. Despite this, we find limited evidence that comments amplify political propaganda.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。