Skip to content
cccskills
FreePublished on GitHub

statistical-acceptance

应用从 AI for Everyone 蒸馏出的「统计化验收标准 — 不追求100%准确」方法论。

Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan related to statistical-acceptance.

The English summary is being prepared. Showing the original description.

See how to install

Files included (2)

  • SKILL.md4.7 KB
  • test-prompts.json4.0 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Source Metadata

Original cangjie-skill frontmatter from the distillation run:

id: f18
title: "统计化验收标准 — 不追求100%准确"
book: "AI for Everyone"
author: "Andrew Ng(吴恩达)"
source: p16, p24
tags: [project-management, evaluation, acceptance-criteria, communication]
related_skills:
  - slug: train-test-split, relation: depends-on
  - slug: ml-workflow, relation: depends-on
  - slug: iterate-not-perfect, relation: composes-with

R — Reading(原文引用)

"AI系统的一个创新点是性能通常以统计方式定义,与其要求AI系统完美执行,我们更常要求AI系统达到特定准确率" — p16

I — Interpretation(用自己的话重新阐述)

要求AI达到100%准确率是商业界最常见的误区之一。追求完美通常是不可能的,原因有三:

  1. ML有根本性限制:任何统计学习方法都存在不可消除的误差下限。
  2. 数据本身有噪声:标注错误、模糊样本、边界案例——数据本身就不完美。
  3. 人类专家也会分歧:对于模糊案例,不同专家给出的答案可能不同,不存在"标准答案"。

正确做法:将验收标准定义为统计目标(如"在测试集上达到95%准确率"),并认识到一个95%准确率的系统已经能创造巨大价值。关键是提前约定可接受的标准,而非事后要求完美。

A1 — Past Application(过去应用案例)

  • 咖啡杯缺陷检测:95%的准确率意味着5%的缺陷产品会漏检。但这仍然非常有价值——可以配合人工抽检来兜底。如果追求100%,项目可能永远无法上线,而95%的系统已经能拦截绝大多数缺陷,大幅降低质检成本。

A2 — Future Trigger(未来触发条件)

触发场景:

  • 业务方坚持"AI不能出错"
  • 团队因为AI犯了几个错误就否定整个项目
  • 验收时对"什么算合格"产生争议

语言信号:

  • "必须100%准确"
  • "零错误"
  • "AI犯错了"
  • "完美"
  • "100%"
  • "零缺陷"

区分要点:当有人(包括你自己)对AI系统提出"零错误"要求时,使用本框架来重新设定合理预期。

E — Execution(执行步骤)

  1. 与AI团队讨论可达成的准确率:了解在当前数据和技术条件下,合理的准确率目标是什么。
  2. 定义统计化验收标准:将目标量化为具体数字(如"测试集准确率≥95%"),写入项目文档。
  3. 接受错误必然发生:明确告知利益相关者——系统会犯错,这是正常的。
  4. 设计错误处理机制:为那5%的错误规划应对方案(人工复核、降级流程、用户反馈通道)。

完成标准:团队和业务方对"可接受的准确率"达成共识,并建立了错误发生时的应对方案。

B — Boundary(使用边界)

不适用场景:

  • 安全关键系统(医疗诊断、自动驾驶、航空控制——这些领域错误会导致人身伤害,需要远高于95%的标准)
  • 95%确实不够用的场景(如金融反欺诈,1%的漏检可能造成巨大损失)
  • 法律合规要求零容忍的领域

失败模式:

  • 盲目接受95%而不评估错误成本:如果每个错误的代价极高,95%可能远远不够
  • 没有设计错误兜底方案:告诉业务方"会有5%错误"但没有应对措施
  • 用统计标准掩盖技术不足:把"做不到"包装成"不需要做到"

作者盲区:

  • Andrew Ng没有深入讨论如何将AI与人工审核结合来处理那5%的错误。在实际项目中,"AI初筛 + 人工复核"的混合模式往往是最佳实践,但书中对此着墨不多。此外,不同类型的错误(假阳性 vs 假阴性)代价不同,简单的准确率数字可能掩盖重要差异。

相关 skills

  • train-test-split — depends-on:统计化验收标准需要基于独立测试集上的准确率来定义,没有正确的数据分割就无法建立可信的验收标准。
  • ml-workflow — depends-on:验收标准是ML工作流中评估阶段的核心产出,需要嵌入整体工作流中理解。
  • iterate-not-perfect — composes-with:接受统计化验收标准与"迭代而非完美"的理念相辅相成——先达到可接受的准确率上线,再持续迭代改进。

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

当用户想判断一个问题是否能用监督学习解决,或试图把业务问题转化为AI问题时调用。 用户在问"AI能不能做X"、"这算不算机器学习问题"、"我们的业务场景适合用AI吗"时使用。 不适用于: 纯数据分析洞察、无明确输入输出的探索性问题、聚类等非监督学习场景。

The English summary is being prepared. Showing the original description.

kangarooking/ai-for-everyone-skill38Updated Jul 3, 2026

应用从 AI for Everyone 蒸馏出的「AI流水线 — 多组件串联的系统设计模式」方法论。Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan related to ai-pipeline.

The English summary is being prepared. Showing the original description.

kangarooking/ai-for-everyone-skill38Updated Jul 3, 2026

应用从 AI for Everyone 蒸馏出的「AI项目生命周期 — 从立项到落地的完整流程」方法论。Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan related to ai-project-lifecycle.

The English summary is being prepared. Showing the original description.

kangarooking/ai-for-everyone-skill38Updated Jul 3, 2026

应用从 AI for Everyone 蒸馏出的「AI战略护城河 — 用AI构建可持续竞争优势」方法论。Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan related to ai-strategy-moat.

The English summary is being prepared. Showing the original description.

kangarooking/ai-for-everyone-skill38Updated Jul 3, 2026

应用从 AI for Everyone 蒸馏出的「AI团队构建 — 打造高效AI团队的角色与协作」方法论。Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan related to ai-team-building.

The English summary is being prepared. Showing the original description.

kangarooking/ai-for-everyone-skill38Updated Jul 3, 2026

应用从 AI for Everyone 蒸馏出的「AI转型五步指南 — 组织变革路线图」方法论。Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan related to ai-transformation-playbook.

The English summary is being prepared. Showing the original description.

kangarooking/ai-for-everyone-skill38Updated Jul 3, 2026

See all skills by kangarooking

Report a problem with this skill