跳到正文
arXiv · AI· Xingang Guo, Jing Gu, Brian Jang, Renxiong Wang, Utkarsh Tyagi, Daniel Quigley, Steven Li, David Yan, Daniel Yue Zhang, Darvin Yi, Forrest Huang, HiJae Kim, Tianyi Zhang, Jared Lichtarge, Jihua Huang, Le Xue, Manan Tomar, Qiuyi Richard Zhang, Ruofei Yu, Seth Neel, Yaning Hu, Marcella Valentine, Xinzhe Jiang, Daniel Evans, Chenguang Wang, Dustin Tran, Tong Zhao, Yinfei Yang, Yunzhong He·· 2026-10-08AI 评分42

Humanity's Sixth Sense:多模态模型直觉视觉推理基准测试

Humanity's Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models

AI 导读

新基准HSS发现前沿多模态模型在直觉视觉推理上表现远逊于人类,人类准确率达93.1%,而最强模型GPT-6-astra仅为53.6%。 HSS涵盖图像与视频输入,通过人类撰写的提示词探测隐性的时间、空间及社会结构推理。引入智能体动态视觉操作虽缩小差距,但未能弥合与人类表现的根本鸿沟。

来源:arXiv · AI · arxiv.org