‹ 首页

transparency-patterns

@owl-listener · 收录于 5 天前 · 上游提交 1 个月前

Showing users what the AI knows, doesn't know, and how confident it is.

场景ai交互设计用户体验优化

适合你，如果你正在构建需要向用户解释AI决策过程的对话系统。

/ 下载安装
https://github.com/owl-listener/ai-design-skills/tree/main/claude-plugin/ai-alignment-reasoning/skills/transparency-patterns ↗
/ 通过 npx 安装 校验哈希
npx oh-my-skill add owl-listener/ai-design-skills/transparency-patterns
/ 通过 bash 安装
curl -fsSL https://oh-my-skill.com/install.sh | bash -s -- owl-listener/ai-design-skills/transparency-patterns
/ 已经装过?验证本机副本,不用重装
npx oh-my-skill verify owl-listener/ai-design-skills/transparency-patterns
安装目标可用 --agent / --scope 或 --to 明确指定;省略时只会在唯一已存在的 agent 目录上自动选择,零命中或多命中会停止并提示。content_hash 缺失或不一致均拒装。
144GitHub stars
~501上下文体积 · 单文件
索引托管

怎么用

商店整理自技能原文 · 版本 f41b650 · 表述以原文为准
它做什么

Claude 会在回答中主动标明信息来源、置信度（如“我很确定”或“不太确定”）、已知局限性（如“我不知道”），以及推理过程。

什么时候触发

当用户提问或要求任务时，Claude 根据问题的重要性和自身不确定性，自动加入置信度、来源和限制说明。

装好后可以这样说
触发来源归因。
触发置信度显示。
触发限制披露。
技能原文 SKILL.md作者撰写 · MIT · f41b650

Transparency Patterns

Transparency in AI products means making the system's knowledge, limitations, and confidence visible to users. It's how you build warranted trust — trust based on understanding, not blind faith.

What to Make Transparent
  • Source: Where did the AI get this information? Training data, retrieved documents, user input, inference?
  • Confidence: How certain is the AI? Is this a well-supported answer or a best guess?
  • Limitations: What doesn't the AI know? What can't it do? Where does its knowledge end?
  • Process: How did the AI arrive at this output? What steps did it take?
  • Identity: This is an AI, not a human. Never obscure this.
Transparency Patterns
  • Confidence indicators: Visual or textual signals of certainty ("I'm fairly confident" vs. "I'm not sure about this")
  • Source attribution: Citing where information came from
  • Reasoning traces: Showing the AI's step-by-step thinking
  • Limitation disclosure: Proactively stating what the AI can't do or doesn't know
  • Model cards: High-level descriptions of what the AI is, how it works, and what it's good and bad at
  • Uncertainty highlighting: Visually distinguishing confident outputs from uncertain ones
Calibrating Transparency

Too much transparency overwhelms. Too little erodes trust. Calibrate by:

  • User expertise: Experts want more detail. Novices want simple signals.
  • Task stakes: High-stakes decisions need full transparency. Low-stakes interactions need less.
  • Output confidence: Show more transparency when the AI is uncertain, less when it's confident.
  • User request: Let users drill into details on demand rather than showing everything upfront.
Transparency Anti-Patterns
  • Performative transparency: Showing a reasoning trace that doesn't actually explain the decision
  • Buried disclaimers: Putting limitations in fine print nobody reads
  • False confidence: The AI sounds certain when it's guessing
  • Opaque refusal: "I can't help with that" with no explanation
  • Transparency theatre: Making the system look transparent without actually being informative
Design Artefacts
  • Transparency level specifications per feature
  • Confidence communication guidelines
  • Source attribution patterns
  • Limitation disclosure templates
按 MIT 许可原样转载,未经改动 · 在 GitHub 查看 →

评论

登录即可评论;带「已验证安装」的,是发布者名下有本店的安装或持有记录。
…