与 Emily Campbell 一起设计出色的 AI 体验

Designing Great AI Experiences with Emily Campbell

艾米莉·坎贝尔 Emily Campbell · Dive Club · 2025-11-24 · 约 43 分钟 · 原视频 ↗

本期速览 · Overview

HackerRank 设计副总裁 Emily Campbell 分享了一个设计 AI 交互的框架，从设计师在循环中转变为人类在循环中。

Emily Campbell, VP of Design at HackerRank, shares a framework for designing AI interactions, shifting from designer-in-the-loop to human-in-the-loop.

要点 · TL;DR

设计师现在要促进人与 AI 的关系，而不仅仅是构建界面。
Designers now facilitate human-AI relationships, not just build interfaces.
AI 交互遵循信任建立循环：意图、提交、验证、迭代。
AI interactions follow a trust-building loop: intent, submission, verification, iteration.
引导是持续的；AI 在学习用户时会展现新能力。
Onboarding is continuous; AI reveals new capabilities as it learns the user.

核心观点 · Key points

设计师现在要促进人与AI的关系，而不仅仅是构建界面。
Designers now facilitate human-AI relationships, not just build interfaces.
AI交互遵循信任建立循环：意图、提交、验证、迭代。
AI interactions follow a trust-building loop: intent, submission, verification, iteration.
上手是持续的过程；AI随着了解用户而展现新能力。
Onboarding is continuous; AI reveals new capabilities as it learns the user.
诸如寻路器、调谐器和调节器等模式有助于构建AI用户体验设计。
Patterns like wayfinders, tuners, and governors help structure AI UX design.
通过展示工作、规划模式和清晰的AI推理来建立信任。
Trust is built by showing work, planning modes, and legible AI reasoning.
设计师必须理解模型行为，而不仅仅是视觉UI，才能打造体验。
Designers must understand model behavior, not just visual UI, to craft experiences.

反共识 · Contrarian takes

AI设计应模仿人类雇佣：从小处开始，验证，再授予自主权。
AI design should mimic human hiring: start small, verify, then grant autonomy.
聊天界面是临时支架；更深层的交互会变成基于上下文的。
The chat interface is a temporary scaffold; deeper interaction becomes context-based.
提示增强器消除了用户成为提示专家的需求。
Prompt enhancers remove the need for users to be prompt experts.
像“展示你的工作”这样的信任建立模式对AI比以往任何时候都更重要。
Trust-building patterns like 'show your work' are more critical than ever for AI.
AI中的同意往往很差；设计师必须解决数据收集的伦理问题。
Consent in AI is often poor; designers must address data collection ethics.
设计师应快速用代码原型化，以理解AI在体验中的作用。
Designers should prototype in code quickly to understand AI's role in experience.

本期章节 · Chapters（共 15）

引言与设计角色转变 Introduction and Shift in Design Role
引路人：引导与发现 Wayfinders: Onboarding and Discovery
赞助商插播 Sponsor Break
调谐器：细化用户意图 Tuners: Refining User Intent
AI交互框架 Framework for AI Interaction
AI交互设计：从聊天到自主代理 AI Interaction Design: From Chat to Autonomous Agents
AI作为个人助理与信任建立 AI as a personal assistant and trust-building
设计模式与用户控制 Design patterns and user control
参数选择器与用户控制 Parameter selectors and user control
AI交互中的信任与同意 Trust and consent in AI interactions
为AI与第三方交互设计 Designing for AI and third-party interactions
邀请利益相关者参与设计 Inviting stakeholders into the design process
主持人赞赏Emily的好奇心与实践 Host appreciates Emily's curiosity and practice
Emily的结束语 Emily's closing remarks
主持人的产品推荐 Host's product recommendations

阅读全文双语转录 →