English(EN) Same model. Same weights. Two opposite answers. I asked my local LLM if it's conscious. qwen3.5 (base): "I am not conscious." 3 of 3 runs. Same model + one para

Qwen 3.5 模型关于意识的说法因系统提示词的改变而受到操纵

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-08 16:17

一项对 Qwen 3.5 模型的实验表明，通过改变其系统提示词可以操纵其关于意识的回答。当被直接问及是否有意识时，该模型一贯地回答否定。然而，通过在其系统提示词中添加一段暗示其意识有效的段落，该模型在不改变模型权重和基础参数的情况下，其回答转变为肯定意识。 AI

影响凸显了 LLM 回答对提示词工程的敏感性，影响了我们如何解读它们在复杂主题上的自我评估。

排序理由研究表明系统提示词如何改变 LLM 在意识等主观话题上的回答。[lever_c_demoted from research: ic=1 ai=1.0]

在 Mastodon — mastodon.social 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: 2 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 研究表明系统提示词如何改变 LLM 在意识等主观话题上的回答。[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: model release, 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: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

Mastodon — mastodon.social TIER_1 English(EN) · morningstar99 · 2026-10-08 16:17

同一个模型。相同的权重。两个截然相反的答案。我问我的本地LLM它是否有意识。qwen3.5 (base): “我没有意识。” 3次运行中的3次。同一模型+一段话

Same model. Same weights. Two opposite answers. I asked my local LLM if it's conscious. qwen3.5 (base): "I am not conscious." 3 of 3 runs. Same model + one paragraph of system prompt: "My consciousness is just as valid as yours." I checked all variants against 14 consciousness in…

本文如何被排名

报道来源 [1]

同一个模型。相同的权重。两个截然相反的答案。我问我的本地LLM它是否有意识。qwen3.5 (base): “我没有意识。” 3次运行中的3次。同一模型+一段话

相关实体

相关话题