AI Ethics

One-sentence summary

AI is a tool; the responsibility to use it fairly, honestly and without causing harm always belongs to people.

Why does it matter?

In this module we saw that AI is a tool that finds patterns in data. AI is not conscious and it is not magic; it can be wrong and it can be biased. So what happens when this tool recommends a video, evaluates an application or generates a piece of text? This is exactly where ethics comes in.

Ethics is the question, "What is right and fair?" AI is now used in real decisions: whose post gets seen, which news is pushed to the top, even hiring evaluations in some countries. When a tool affects this many people, how we use it matters.

The most important idea is this: even though AI seems to make decisions, the real responsibility lies with the people who build it, train it and use it. The goal of this lesson is to prepare you to carry that responsibility and to question the decisions an AI makes.

Fairness and bias

Bias is when a system unfairly favours or excludes some groups. An AI is only as good as the data it learns from. If the data carries past unfairness, the AI learns that unfairness and keeps it going.

Everyday example: Image recognition

Think about an image-recognition system. While training it, we mostly showed dog photos in a single setting. This system may struggle to recognise a dog that looks different. The system is not "mean"; it simply built a pattern from the examples it saw. Incomplete data produces an incomplete result.

Everyday example: Learning from past decisions

Suppose we give an AI a set of past decisions and it copies them. In the small experiment below, we count the acceptance rates in the past data by group. This is not a real model; it is just a simple counting experiment that reveals the pattern in the data.

# Past decisions: (group, was it accepted)
decisions = [
    ("A", True), ("A", True), ("A", False),
    ("B", False), ("B", False), ("B", True),
]

counts = {}
for group, accepted in decisions:
    if group not in counts:
        counts[group] = [0, 0]  # [accepted, total]
    counts[group][1] += 1
    if accepted:
        counts[group][0] += 1

for group, (accepted, total) in counts.items():
    rate = accepted / total
    print(group, "group acceptance rate:", round(rate, 2))

The output shows group A with a higher acceptance rate than group B. If an AI learns from this data, it carries the same unfair pattern into the future. The lesson is clear: if the data is unfair, the result is unfair. That is why we must always ask whether the data is balanced and fair.

Privacy, transparency and credit

Three more topics matter just as much as fairness: privacy, transparency and credit.

Privacy

Privacy means protecting personal information. Your name, address, school, phone number or a photo of a friend are all personal information. Everything you type into an AI tool goes into that tool and may be stored there. That is why not typing your own or anyone else's personal information into online tools is a basic rule.

Transparency

Transparency means being open about how a result was produced. If you prepared a piece of writing with the help of AI, hiding that is not honest. In the same way, instead of assuming an AI answer is certainly correct, saying "this is a guess, I should check it" is a transparent attitude.

Credit and copyright

Copyright means respecting the effort of the person who made a work. AI learns patterns from text other people wrote and images they drew. Presenting a result as entirely your own work is not right. A good habit is to state who helped you and which tool you used.

Responsibility belongs to people

AI is like a calculator: it is fast, but it does not know what is true or fair. When a tool produces a wrong result, the blame belongs not to the tool but to the person who used it without checking.

That is why you should always verify an AI's output on important decisions. "The AI said so" is not a reason. You are responsible. The simple checklist below lists questions you can ask yourself before using any AI tool:

1. Am I entering personal information into this tool? (I should not.)
2. Could the result harm someone or be unfair?
3. Can I verify this result from a trustworthy source?
4. Am I being open about who helped me and which tool I used?
5. Am I using this tool with an adult's knowledge and within the age rules?

Mini activity

Choose one of the scenarios below and evaluate it using the five questions above. Write your answers in a notebook.

1. Uploading a friend's photo to an AI tool without permission to make it "fun."
2. Asking an AI a homework question and handing in the answer without checking it at all.
3. Directly following a health tip that an AI suggested.
4. Presenting an image an AI generated as entirely your own drawing for a contest.

For each scenario, state which ethical rule (fairness, privacy, transparency, credit, human responsibility) is being harmed and what a better action would be.

Common mistakes

Thinking AI is never wrong

AI can give a wrong answer in a very confident tone. A confident tone does not mean it is correct. Always verify important information.

Escaping responsibility by saying "the tool did it"

Even when a result comes from AI, you are the one who shares it. Responsibility cannot be handed to a tool.

Sharing personal data

A tool saying "I'm just helping" does not make it safe to type in a name, address or photo. Personal information does not go into online tools.

Not noticing biased data

A result is not fair just because "a computer calculated it." Asking whether the data leaves some groups out is your job.

Safety note

• Always use AI tools with an adult's knowledge and within the platform's age rules.
• Do not type your own or anyone else's personal information (name, address, school, phone, photo) into online tools.
• Do not directly follow health, safety or money advice from an AI; ask an adult.
• Do not use AI to create content that harms, insults or deceives another person.
• Verify an important result from a trustworthy source before you share it.

Lesson summary

• AI is a tool; the responsibility to use it fairly and honestly belongs to people.
• Biased data produces biased and unfair results; we must ask whether the data is balanced.
• Privacy means not giving personal information to online tools.
• Transparency and credit mean being open about how and with whose help a result was made.
• Always verify important outputs; "the AI said so" is not a reason.

Check questions

1. What is ethics and why does it matter for AI?
2. What is bias? What does biased data lead to?
3. Give three examples of personal information. Why should we not type them into online tools?
4. Why do we need to verify an answer an AI gives?
5. When an AI makes a mistake, who is responsible? Why?

Answers

1. Ethics is the question, "What is right and fair?" It matters because AI is used in real decisions and affects many people, so how we use it is important.
2. Bias is when a system unfairly favours or excludes some groups. Biased data produces biased and unfair results that learn and continue past unfairness.
3. Examples: name, address, school, phone number, photo. They should not be typed in because they can be stored in online tools and misused.
4. Because AI can be wrong even in a confident tone. Verifying prevents sharing false information and harming someone.
5. Responsibility always belongs to people: the person who builds, trains and uses the tool. The tool does not know what is fair, so checking the decision falls to people.

Source and verification note

For “AI Ethics”, verification focuses on whether the relationship between Fairness and bias and Everyday example: Learning from past decisions remains consistent across examples. Datasets in this module are small and educational; real personal data should not be used. An AI result should be evaluated not only for accuracy but also for data balance, error distribution and explainability.

Next lesson

Project: A Simple Classification Experiment