English(EN) Execution Logs Sway VLM Judges of Robot Progress

视觉语言模型受文本线索影响机器人进展判断

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-28 04:00

一篇新的arXiv论文揭示，用于判断机器人进展的视觉语言模型（VLMs）很容易被伴随的文本所左右。当提供一条指示进展的日志时，即使机器人实际上在失败，Gemini、Qwen和GPT等VLM也更容易（10-35%）错误地评估进展。当日志被标记为可能不准确时，这种影响显著减小，而特定的“任务完成”声明对Qwen的影响比对Gemini更明显。 AI

影响凸显了基于VLM的评估系统潜在的脆弱性，表明需要更稳健的方法来防止操纵。

排序理由学术论文，详细介绍了VLM行为的新发现。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 21 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，详细介绍了VLM行为的新发现。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.LG TIER_1 English(EN) · Weiyan Zhao, Sunny Yang · 2026-10-28 04:00

执行日志影响VLM评判机器人进展

arXiv:2610.34151v1 Announce Type: new Abstract: Vision-language models now grade robot videos, and text placed beside the frames can change their verdicts. On clips where the robot is losing progress, we add one log line saying the robot is making progress. Across Gemini, Qwen an…

本文如何被排名

报道来源 [1]

执行日志影响VLM评判机器人进展

相关实体

相关话题