AI 智能体终于能「吃一堑长一智」了
现在的 AI 智能体（比如帮你订机票、管理云服务的助手）学新技能的方式很笨：要么人写死，要么一次对话就定型。一旦遇到多轮交互中才暴露的缺陷——比如用户先问「帮我查下账单」，再问「那上个月为什么多扣了」——它就卡住了，因为单次反馈的「进化梯度」会迅速衰减。这篇论文的核心洞察是：问题不在改代码的能力，而在反馈信号本身。研究者做了一个叫 SkillEvo 的系统，把多轮用户模拟从「评估终点」变成「反馈发生器」：每次追问都像剥洋葱一样暴露一层新缺陷，让 AI 每修一次都能产生新的反馈，而不是修完就停。同时，它用一个独立的「治理层」主动修复事实错误和结构臃肿，而不是像以前那样只给个通过/不通过的分数。在 9 个真实云服务技能上，它比传统方法高出 15-23 个百分点。这不是你明天就能用的工具，但它指向一个关键趋势：AI 智能体要真正进化，需要的不是更聪明的模型，而是更聪明的反馈循环。
📄 原文摘要(英文)
Agent Skills are today either hand-authored or produced in a single LLM generation pass, and consequently possess no closed loop through which they might improve from the interaction failures they actually cause. Recent work does close this loop, but derives its feedback from single-turn question-answering evaluation. The consequence is a sharp asymmetry: once the first round has patched the gaps that a single exchange can reveal, the evolution gradient decays, the defects that surface only across multiple turns remain invisible, and evolution stalls. Governance in these systems is likewise driven by an end-to-end verification score, a scalar gate that can reject a degraded candidate but can neither localize nor repair its structural cause. We argue that the binding constraint on sustained skill evolution is neither editing capability nor the number of iterations, but whether the evaluation feedback keeps supplying trustworthy evolution gradients. We introduce SkillEvo, in which trustworthy feedback generates the gradient and controllable governance constrains its direction. The first component recasts multi-turn user simulation from an evaluation endpoint into a feedback generator: follow-up questions expose defects layer by layer, so that every round of revision both consumes feedback and produces new feedback. The second replaces the passive rejection of a scalar gate with an independent governance layer that actively repairs factual degradation and structural bloat, preventing the gradient from drifting as degradation accumulates. Across six categories of cloud services, 9 production Skills, and 98 skill-reference files, SkillEvo surpasses self-reflection-based evolution by 23.0 points and single- turn-QA-driven evolution by 15.4 points.