Ethics of Data Collection · 数据收集的伦理
| English | 中文 | Pinyin · 拼音 |
|---|---|---|
| ethical/ˈeθɪkl/ | 伦理 | lún lǐ |
| Privacy/ˈprɪvəsi/ | 隐私 | yǐn sī |
| bias/ˈbaɪəs/ | 偏见 | piān jiàn |
Data has consequences
- Programs collect and process data about real people — with real effects.
- How data is gathered, stored, and used raises ethical 伦理 questions.
- A helpful data set can also expose or harm the people in it.
- Responsible programmers think about impact before collecting.
数据有后果
- 程序收集并处理关于真实人的数据——带来真实的影响。
- 数据如何被收集、存储和使用引发伦理问题。
- 一份有用的数据集也可能暴露或伤害其中的人。
- 负责任的程序员在收集之前就思考影响。
Privacy and consent
- Privacy 隐私: people have a right to control information about themselves.
- Collect only what you need, and get consent where appropriate.
- Storing sensitive data creates a duty to protect it from misuse or leaks.
- "Because we can" is not a good reason to collect personal data.
隐私与同意
- **隐私：**人们有权控制关于自己的信息。
- 只收集你需要的，并在适当处获得同意。
- 存储敏感数据带来保护它免受滥用或泄露的责任。
- “因为我们能”不是收集个人数据的好理由。
Bias and fairness
- Data can carry bias 偏见 — patterns that treat some groups unfairly.
- An algorithm trained on biased data can amplify that unfairness at scale.
- Fair systems require checking who the data represents and who it leaves out.
- Correct-looking code can still produce unjust outcomes.
偏见与公平
- 数据可能携带偏见——不公平对待某些群体的模式。
- 用有偏数据训练的算法可能在规模上放大那种不公。
- 公平的系统需要检查数据代表谁、又遗漏了谁。
- 看起来正确的代码仍可能产生不公正的结果。
Ownership and security
- Data and code have ownership — respect licenses, credit sources, don't steal.
- Security: protect stored data so it isn't stolen or altered.
- A breach can harm thousands of people at once.
- Ethics isn't optional polish — it's part of building software responsibly.
所有权与安全
- 数据和代码有所有权——尊重许可、注明来源、不要窃取。
- **安全：**保护存储的数据，使其不被窃取或篡改。
- 一次泄露可能一次性伤害成千上万的人。
- 伦理不是可选的润色——它是负责任地构建软件的一部分。
Ethical or a concern? · 是合乎道德的还是值得担忧的？
Sort each data-collection practice. · 对每种数据收集做法进行排序。
"The program runs correctly" is not the same as "the program is ethical." Code that works can still violate privacy, amplify bias, or expose data through weak security. Before collecting or processing data about people, ask what you actually need, whose consent you have, who the data represents, and how you'll protect it.
“程序运行正确”不等于“程序合乎伦理”。能工作的代码仍可能侵犯隐私、放大偏见，或通过薄弱的安全暴露数据。在收集或处理关于人的数据之前，问一句：你到底需要什么、你有谁的同意、数据代表谁、以及你将如何保护它。
A school app that stores student grades:
- Privacy: collect only grades needed, not unrelated personal details.
- Bias: a "recommendation" feature must not disadvantage any group.
- Security: protect the stored grades so they can't leak or be altered.
一个存储学生成绩的学校应用：
- **隐私：**只收集所需的成绩，而非无关的个人细节。
- **偏见：**一个“推荐”功能不得使任何群体处于不利。
- **安全：**保护存储的成绩，使其不能泄露或被篡改。
Collecting data about people raises ethical duties: protect privacy (collect only what's needed, get consent), guard against bias that can be amplified at scale, respect ownership, and ensure security. Correct code is not automatically ethical — weigh who is affected before you collect or process data.
收集关于人的数据引发伦理责任：保护隐私（只收集所需、获取同意），防范可能在规模上被放大的偏见，尊重所有权，并确保安全。正确的代码不会自动合乎伦理——在收集或处理数据前权衡谁受影响。
A program that runs correctly is... · 正确运行的程序...
Correct code can still violate privacy or amplify bias. · 正确的代码仍可能侵犯隐私或放大偏见。
A good practice when collecting personal data is to... · 收集个人数据时的良好做法是...
Minimize collection and respect consent and privacy. · 最小化收集并尊重同意和隐私。
An algorithm trained on biased data can amplify that unfairness at scale. · 在偏见数据上训练的算法可能会在大范围内放大这种不公平。
Bias in data can produce unjust outcomes for many people. · 数据中的偏见可能会对许多人产生不公正的后果。
Protecting stored data from theft or alteration is part of security. · 保护存储的数据免受盗窃或篡改属于安全范畴。
A breach can harm many people at once. · 数据泄露可能同时伤害许多人。
The right of people to control information about themselves is called ___ (one word). · 人们控制有关自身信息的权利称为___（一个单词）。
Privacy is a core data-ethics concern. · 隐私是核心数据伦理问题。