Agentic Web 中推荐系统向谁推荐？
原标题：Who Are We Recommending To? Recommender Systems in the Agentic Web
AI 摘要
这篇 arXiv cs.IR 立场论文指出，随着 LLM 驱动的 AI 代理在 Agentic Web 中代替用户浏览、比较、谈判和交易，推荐系统的核心假设——推荐由人类直接消费——正在被打破。作者提出推荐范式正在分叉：在可委托场景（如常规购物、旅行）中，代理成为推荐的主要操作消费者，需要新的优化目标、交互协议和评估标准；在体验型场景（如娱乐、艺术）中，人类仍是最终判断者。论文引入「委托光谱」刻画不同推荐情境，并勾勒出代理偏好建模、双受众优化和代理注意力经济等研究议程。
正文节选
Who Are We Recommending To? Recommender Systems in the Agentic Web Abstract. For two decades, recommender systems have been designed under the assumption that a human directly consumes each recommendation: receiving, interpreting, and acting upon it. The emergence of AI agents powered by large language models challenges this assumption. In the emerging Agentic Web (Yang et al., 2025), autonomous agents increasingly act on behalf of users, e.g., browsing, comparing, negotiating, and executing tra