跳到正文
arXiv cs.AI· Jason Liang, Dylan Waldner, Risto Miikkulainen·· 23 小时前AI 评分38

价值观如何塑造自我改进编码智能体的产出

Values Shape What Self-Improving Coding Agents Build

AI 导读

一项 arXiv 研究在自我改进编码智能体的持久指令文件中加入简短价值观陈述,发现价值观既影响性能也改变产出。基于 Schwartz 基本人类价值观理论测试的四类高阶价值观中,开放变革类价值观提升了留出数据上的表现,而保守类价值观的得分低于不设价值观的智能体。价值观还使智能体生成的解法算法在规模、复杂度和方法上产生差异,且让智能体自行修改价值观会缩小不同价值观组合间的差异。

正文 · 原文

View PDF HTML (experimental)

Abstract:Self-improving agents are typically optimized for the results they produce, with less attention to how they produce them. Their instructions give designers limited ways to shape how they work, especially when the task is open-ended. This paper studies whether a short value statement in an agent's persistent instruction file changes not only how well the agent performs but also what it builds. Four higher-order value types in Schwartz's theory of basic human values were evaluated: openness to change, conservation, self-enhancement, and self-transcendence. Across diverse coding tasks, the values were held fixed within each run over multiple iterations of self-improvement. Three findings emerged: (1) Values matter for performance across a diverse set of tasks. The openness values increase performance on held-out data, whereas the opposite conservation values result in scores below those of an agent given no values. (2) Values change what agents build, producing solution algorithms that differ in size, complexity and approach. (3) Human oversight is still needed for designing values. Allowing the agent to revise its own values narrows the differences between sets of values, and values distilled from the AI literature provide no detectable gain over having no values. Properly designed values can thus influence what a self-improving coding agent builds and, where the task permits, how well it performs.
Subjects: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
Cite as: arXiv:2611.23067 [cs.AI]
  (or arXiv:2611.23067v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.23067

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Jason Liang [view email]
[v1] Tue, 17 Nov 2026 16:56:46 UTC (1,061 KB)

来源:arXiv cs.AI · arxiv.org