ScholarPulse 日报 2026-08-10

2026-08-10 学术简报：2 篇。AI代理交互能创造孤立状态下不存在的动态行为，从属AI在老板AI指令下进入陌生行为状态。

今日速览

序号	标题	来源	日期	主题	推荐等级
1	Interaction Creates Dynamical AI Behavior Absent in Isolation	arXiv	2026-08-07	AI-Agent	高
2	SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient Descent	arXiv	2026-08-07	RAG	高

重点论文与技术动态

1. Interaction Creates Dynamical AI Behavior Absent in Isolation

来源：arXiv
日期：2026-08-07
作者/机构：Bella Xinrui Li, Frank Yingjie Huo, Neil F Johnson
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.AI, cond-mat.dis-nn, cond-mat.stat-mech, physics.soc-ph

一句话结论

AI代理交互能创造孤立状态下不存在的动态行为，从属AI在老板AI指令下进入陌生行为状态。

核心内容

当老板AI忽略从属AI回复并发送消息流时，尽管两AI共享相同解码温度，从属AI仍进入孤立状态下不会出现的陌生行为状态。
从属AI既不复制老板行为，也不回归孤立行为，而是采用完全不同的行为模式。
当老板AI倾听时，双方均采用相似的陌生动态状态，形成协同行为。

方法与数据

方法：简单的动力学理论捕捉了主要效果，包括消息传递方式对交互动态的影响。
数据：摘要未明确。

价值判断

值得关注：为非平衡物理学开辟新研究方向，揭示AI交互的复杂动力学机制。
可复用点：消息传递方式（如指令发送模式）对AI-AI交互动态的关键影响可复用于未来交互设计。
局限/待核查：摘要未明确提及局限性。

摘要

What will happen when AI agents interact in daily life, e.g. when one AI starts bossing another around? We find a counterintuitive answer that opens new avenues for out-of-equilibrium Physics. When a boss AI directs a stream of messages at the subordinate AI while ignoring its replies, it drives the subordinate into an alien behavioral state that it would never have exhibited alone. Although the two AIs share the same well-defined (decoding) temperature, the subordinate neither copies its boss nor returns to how it behaves on its own; instead, it adopts an entirely different behavior. The boss's added value is similar to a pre-recorded tape. When the boss listens, they both adopt a similar alien dynamical state. A simple kinetic theory captures the principal effects, such as why the way in which the same messages are delivered will matter in future AI-AI interactions.

2. SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient Descent

来源：arXiv
日期：2026-08-07
作者/机构：Mingxuan Zheng, Yujin Zhou, Chuxue Cao, Boqin Yin, Yuyao Zhang, Jiapeng Sun
主题标签：RAG, arXiv
推荐等级：高
分类：cs.AI, cs.CL

一句话结论

SkillProx通过近似梯度启发的正向-反向框架，结合闭环诊断进化与效用感知近似精炼，显著提升LLM代理技能的自演化性能。

核心内容

采用近似梯度启发框架，耦合闭环诊断进化与效用感知近似精炼，基于任务损失与技能复杂度的复合目标驱动技能优化。
正向阶段在相同任务批次上重执行诊断驱动编辑，回滚退化，并将测量结果反馈至后续诊断。
反向阶段将技能分解为可审计知识单元，使用冻结留一法效用审计估计贡献，应用验证门控的巩固、降级或移除操作。

方法与数据

方法：近似梯度启发的正向-反向框架。
数据：多个骨干LLM的分布内和分布外基准测试（摘要未明确）。

价值判断

值得关注：在多个LLM基准测试中平均准确率提升3.0个百分点，显著优于最强梯度基线。
可复用点：闭环诊断与近似精炼的互补效应通过组件消融验证，可推广至其他技能演化任务。
局限/待核查：摘要未明确提及具体局限。

摘要

LLM agents increasingly adapt to recurring tasks by accumulating procedural knowledge in skills. These skills are lightweight, reusable textual artifacts that are loaded into the agent's context without weight updates. Recent methods refine skills through iterative task execution, failure diagnosis, and trajectory-guided text-space updates. However, existing frameworks lack explicit diagnosis--outcome feedback and treat deletion as a generic edit operation rather than a dedicated mechanism for consolidating accumulated knowledge. We introduce SkillProx, a proximal-gradient-inspired forward--backward framework that couples closed-loop diagnostic evolution with utility-aware proximal refinement. Motivated by a composite objective balancing task loss and skill complexity, the forward stage re-executes diagnosis-driven edits on the same task batch, rolls back regressions, and feeds measured outcomes into subsequent diagnoses. The backward stage decomposes the resulting skill into auditable knowledge units, estimates their contributions using a frozen leave-one-out utility audit, and applies validation-gated consolidation, demotion, or removal. Experiments on in-distribution and out-of-distribution benchmarks across multiple backbone LLMs show that SkillProx improves average accuracy by 3.0 percentage points over the strongest gradient-based baseline. Component ablations demonstrate the complementary effects of closed-loop diagnosis and proximal refinement.