Event ·Official first-hand

当发货人变成算法：LLM中介货运市场的候选展示、信息设计与集中度

一项研究模拟了50个基于OpenAI GPT、Anthropic Claude和Google Gemini的货运代理在30天内选择承运人的市场。结果显示，代理在第一天就集中选择同一承运人，占76%的请求。当候选列表超过10个时，集中度急剧上升，但不同模型表现有差异。披露承运人剩余日容量可将集中度降低三分之一，并使发货人剩余翻倍。供应商多样化、列表随机排序和流行度展示没有明显效果。

Current read

这篇论文用真实LLM做模拟，发现只要给发货人看十多家承运人，市场就会高度集中，但暴露容量信息就能大幅改善，比换模型管用多了。

11 sources51° AI heatLast updated 2026/7/22 09:51:56

Evidence

11 sources

Primary sourcearXiv: OpenAI

2026/7/22 09:51:56View source

Related source 1Decoder

2026/7/20 18:08:33View source

Related source 2IT之家

2026/7/22 23:25:14View source

Related source 3elvis

2026/7/20 22:09:00View source

Related source 4Guillermo Rauch

2026/7/21 14:55:55View source

Related source 5berryxia

2026/7/21 15:20:55View source

Related source 6Harrison Chase

2026/7/21 17:30:41View source

Related source 7Geek

2026/7/22 01:20:44View source

Related source 8小互

2026/7/22 07:05:47View source

Related source 9The Rundown AI

2026/7/20 10:30:11View source

Related source 10techcrunch

2026/7/20 15:15:00View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story