Computer Science
Grade 7
20 min
Lesson 8: Ethical Considerations in AI: Bias and Fairness
Explore ethical considerations related to AI, such as bias in algorithms and fairness in decision-making.
What you'll learn
- Identify at least three different types of biases that can occur in AI systems, providing a specific example for each.
- Explain in their own words how biased data can lead to unfair outcomes in AI applications, using an example scenario like facial recognition or loan applications.
- Analyze a given scenario involving a potentially biased AI system and suggest at least two ways to mitigate the bias and promote fairness.
- Evaluate the potential consequences of using a biased AI system in a real-world application and propose at least one ethical consideration to address.
Tutorial Preview
1
Introduction & Learning Objectives
Learning Objectives
Define AI bias and fairness in their own words.
Identify at least two sources of bias in an AI training dataset.
Explain how a biased algorithm can lead to unfair outcomes for different groups of people.
Analyze a simple AI scenario and determine if it is fair.
Propose a method to make a biased AI system more fair.
Describe the importance of diversity in data and development teams for creating ethical AI.
Have you ever wondered why your video recommendations sometimes feel like they're for a completely different person? 🤔 Let's explore why that might be happening!
Today, we'll investigate how Artificial Intelligence can sometimes make unfair decisions. We will learn about AI bias, why it happens, and how we, as future computer scientists,...
2
Key Concepts & Vocabulary
TermDefinitionExample
Artificial Intelligence (AI)A field of computer science where we teach computers to perform tasks that normally require human intelligence, like learning, problem-solving, and making decisions.A smart speaker like Alexa or Google Home that understands your voice commands and answers questions.
AlgorithmA set of step-by-step instructions that a computer follows to complete a task or solve a problem.A recipe for baking cookies is an algorithm. In AI, an algorithm might be the set of rules the AI uses to decide if an email is spam.
Training DataThe information (like images, text, or numbers) that we feed to an AI to help it learn patterns.To teach an AI to recognize cats, you would show it thousands of pictures, each labeled 'cat'. These pictures are the train...
3
Core Syntax & Patterns
The Diverse Data Principle
IF training_data is not diverse THEN AI_model will likely be biased.
Use this as a checklist when thinking about AI. The quality and diversity of the data used to train an AI is the most important factor in whether it will be fair. Always ask: 'Who is missing from this data?'
The Fairness Test Pattern
FUNCTION test_for_bias(model, group_A_data, group_B_data) {
result_A = model.predict(group_A_data);
result_B = model.predict(group_B_data);
RETURN are_results_similar(result_A, result_B);
}
This is a program design pattern. It shows that after building an AI, you must create a function to test it with different groups to see if it performs equally well for everyone. It's a way to actively check for bias.
The Human Review Rule...
4 more steps in this tutorial
Sign up free to access the complete tutorial with worked examples and practice.
Sign Up Free to ContinueSample Practice Questions
Challenging
A team wants to fix their biased hiring AI, which was trained on 900 resumes from men and 100 from women. A manager suggests 'fixing' it by adding 1,000 more resumes from men who have been successful at the company. Why is this a flawed approach?
A.It would cost too much money to get that many resumes.
B.The AI would become too slow if it had too much data.
C.It follows the 'More Data is Always the Solution' pitfall by adding more of the same biased data, which will likely increase the bias.
D.It is better to have a human read all the resumes instead of using an AI.
Challenging
You are testing a new voice assistant AI to see if it is biased against people with non-native accents. How would you best apply the 'Fairness Test Pattern' to investigate this?
A.Create two datasets: one with voice commands from native speakers (Group A) and one with commands from non-native speakers (Group B). Compare the AI's accuracy for both.
B.Give the AI thousands of voice commands from native speakers only and see if it gets better over time.
C.Ask the AI if it thinks it is biased and record its answer.
D.Translate all the voice commands into text first, then give them to the AI to remove any accent.
Challenging
A school uses an AI to predict which students are at risk of dropping out. The AI flags a student, and the school immediately puts them in a lower-level class without talking to them. What is the primary ethical failure here, based on the lesson's principles?
A.The AI's prediction might have been wrong, and there was no human review for an important decision.
B.The AI's algorithm was not published online for everyone to see.
C.The AI was probably trained on data from another school.
D.The school should have used two different AIs to check each other's work.
Want to practice and check your answers?
Sign up to access all questions with instant feedback, explanations, and progress tracking.
Start Practicing FreeMore from Chapter 6: Artificial Intelligence (AI) Basics: Making Machines Think
Lesson 1: What is Artificial Intelligence (AI)? Defining Intelligent Machines
Lesson 2: Types of AI: Narrow AI, General AI, and Super AI
Lesson 3: Machine Learning: Learning from Data
Lesson 4: Supervised Learning: Training AI with Labeled Data
Lesson 5: Unsupervised Learning: Finding Patterns in Unlabeled Data
Computer Science for other grades
Frequently asked questions
What grade level is "Lesson 8: Ethical Considerations in AI: Bias and Fairness"?
Lesson 8: Ethical Considerations in AI: Bias and Fairness is a Grade 7 Computer Science lesson on ExcelOS.
What will I learn in Lesson 8: Ethical Considerations in AI: Bias and Fairness?
You'll be able to: Identify at least three different types of biases that can occur in AI systems, providing a specific example for each; Explain in their own words how biased data can lead to unfair outcomes in AI applications, using an example….
Is "Lesson 8: Ethical Considerations in AI: Bias and Fairness" free to practice?
Yes. You can read the tutorial preview for free, and signing up for a free ExcelOS account unlocks the full tutorial and all practice questions with instant feedback.
How many practice questions are included with Lesson 8: Ethical Considerations in AI: Bias and Fairness?
This lesson includes 27 practice questions across multiple difficulty levels, each with instant feedback and explanations.