Page 69 - CT_AI_Class-8

B. Fill in the blanks using the given hints:
Hints
1. AI can make mistakes if the data it learns from is not or
Review
diverse.
Balanced
2. The data influences what AI detects, which means if the
Decisions
data is unbalanced, the AI's predictions can be biased or incorrect.
Training
3. AI can pick up human biases from data if the developers are Historical
not careful.
4. Human beings should AI decisions rather than trusting the system blindly.
5. It’s crucial to understand how an AI system works and how it makes its .

C. Write ‘T’ for true and ‘F’ for false.
1. AI looks for patterns it has learned from its training data.

2. The more we know about the AI’s processes, the better we can ensure it operates fairly.

3. AI systems do not require balanced and diverse training data to avoid errors.

4. AI should be designed to benefit only a particular group, not everyone.

5. Training data with more examples of one type of object makes AI better at recognising
that object.

D. Answer in one word.
1. The unfair representation of certain groups in AI datasets.
2. The type of AI training that involves learning from labelled images.

3. AI looks for based on its training data.

4. AI needs more of to improve its recognition and differentiation.
5. The AI's image recognition depends on in relation to the training data.

E. Answer the following questions:

1. What is the importance of ensuring diversity in training data for AI systems?

2. Why is it important to use balanced training data when building AI systems?

3. What was the main issue with the AI face recognition system tested by Joy Buolamwini?

4. Describe the meaning of bias.

Data and Fairness in AI 67