# 研究显示仅提供不必要工具即可让 LLM 作答率从 98.2% 降至 63.5%

- 来源：DAIR.AI (@dair_ai)
- 发布时间：2026-09-16 12:20
- AIHOT 分数：56
- AIHOT 链接：https://aihot.news/items/cmu3lpuje06gproo4o0ghfn6h
- 原文链接：https://x.com/dair_ai/status/2100077223252512956

## AI 摘要

一篇论文发现，仅仅让相关工具可用就会使大语言模型放弃回答本可直接答的问题，六个模型的作答率从 98.2% 降至 63.5%，即使工具从未被调用。基准含 10 个领域 500 组查询对并设无工具对照；前序工具调用会挽回部分丢失答案但引入新损失，一句说明工具用途的系统提示词最多挽回 45.6 个百分点。

## 正文

Giving an assistant extra tools seems harmless.

Surprisingly, this paper shows it can make the assistant stop answering questions it knows.

Across six models, the answer rate on questions that need no tool fell from 98.2% to 63.5% when a related tool was merely available.

So the drop comes from the tool being present, even when it goes unused.

The benchmark has 500 query pairs across 10 domains, each with a tool-unavailable control. A preceding tool call recovered some lost answers and caused new losses.

A one-sentence system instruction that tells the model what the tool is for recovered up to 45.6 points.

Paper: https://academy.dair.ai/papers/when-tools-get-in-the-way-the-effect-of-unnecessary-tool-availability-on-llm-ans-2609.14157