Ethics of Data Collection
Introduced| English |
|---|
| ethical/ˈeθɪkl/ |
| Privacy/ˈprɪvəsi/ |
| bias/ˈbaɪəs/ |
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.
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.
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.
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.
Ethical or a concern?
Sort each data-collection practice.
"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.
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.
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.
A program that runs correctly is...
Correct code can still violate privacy or amplify bias.
A good practice when collecting personal data is to...
Minimize collection and respect consent and privacy.
An algorithm trained on biased data can amplify that unfairness at scale.
Bias in data can produce unjust outcomes for many people.
Protecting stored data from theft or alteration is part of security.
A breach can harm many people at once.
The right of people to control information about themselves is called ___ (one word).
Privacy is a core data-ethics concern.