For Individuals
- Learning & Certification
- Specialist Tracks & Tools
Consulting & Enterprise
- Programmes & Solutions
- Diagnostics & Evidence
Insights
- Analysis & Research
  - Executive Reports In-depth whitepapers for decision-makers
  - Academic Research Peer-reviewed publications and frameworks
- Editorial & Dispatches
  - The C-Suite AI Memo Weekly executive strategy newsletter
  - News & Commentary Press, updates, and market notes
Events
About
Contact
LOGIN

Skip to content

AI Bias Testing: Measuring Fairness With Numbers You Can Defend

Intersectional Analysis Beyond Single Attributes

Quiz: Intersectional Analysis Beyond Single Attributes

Bias in Generative Output: Representation and Stereotyping

Quiz: Bias in Generative Output: Representation and Stereotyping

Mitigation: Pre Processing, In Processing and Post Processing

Quiz: Mitigation: Pre Processing, In Processing and Post Processing

Trade Offs Between Accuracy, Fairness and Business Constraints

Quiz: Trade Offs Between Accuracy, Fairness and Business Constraints

Documenting a Bias Audit for Publication or Regulator Review

Quiz: Documenting a Bias Audit for Publication or Regulator Review

Ongoing Fairness Monitoring After Deployment

Quiz: Ongoing Fairness Monitoring After Deployment

Final exam: AI Bias Testing: Measuring Fairness With Numbers You Can Defend

2 of 2

Final exam: AI Bias Testing: Measuring Fairness With Numbers You Can Defend

AI Bias Testing: Measuring Fairness With Numbers You Can Defend
Final exam: AI Bias Testing: Measuring Fairness With Numbers You Can Defend

Time limit: 0

Quiz Summary

0 of 10 Questions completed

Questions:

Information

You have already completed the quiz before. Hence you can not start it again.

Quiz is loading…

You must sign in or sign up to start the quiz.

You must first complete the following:

Results

Quiz complete. Results are being recorded.

Results

0 of 10 Questions answered correctly

Your time:

Time has elapsed

You have reached 0 of 0 point(s), (0)

Earned Point(s): 0 of 0, (0)
0 Essay(s) Pending (Possible Point(s): 0)

Categories

Not categorized 0%

Review
Answered
Correct
Incorrect

Question 1 of 10

1. Question

Statistical bias in machine learning models is most likely to occur when there are systematic differences in outcomes across demographic groups.

When models are trained on large datasets
Correct Incorrect Correct answer

When there is equal representation of all groups in training data
Correct Incorrect Correct answer

When models systematically favour certain outcomes over others
Correct Incorrect Correct answer

When algorithms are designed to be neutral
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 2 of 10

2. Question

Bias in AI systems often originates from the data used to train them, which may reflect historical patterns of discrimination.

Data bias is rare and usually unintentional
Correct Incorrect Correct answer

Bias typically enters through the data used for training
Correct Incorrect Correct answer

Bias only appears during the deployment phase
Correct Incorrect Correct answer

Bias is caused by faulty hardware or software
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 3 of 10

3. Question

Which of the following is NOT listed as a protected characteristic under UK law?

Age
Correct Incorrect Correct answer

Sexual orientation
Correct Incorrect Correct answer

Political opinion
Correct Incorrect Correct answer

Disability
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 4 of 10

4. Question

What must organisations demonstrate when collecting demographic data for AI bias testing under UK law?

That the data is collected from at least five different sources
Correct Incorrect Correct answer

That the data collection serves a legitimate testing purpose and is proportionate
Correct Incorrect Correct answer

That the data is stored in cloud-based systems for accessibility
Correct Incorrect Correct answer

That the data is used exclusively for marketing research purposes
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 5 of 10

5. Question

Statistical parity examines whether the probability of positive outcomes remains consistent across protected attributes such as gender or ethnicity.

True positive rates are compared across demographic groups
Correct Incorrect Correct answer

The probability of positive outcomes is checked for consistency across groups
Correct Incorrect Correct answer

False positive rates must be equal for all demographic segments
Correct Incorrect Correct answer

Equal representation in favorable outcomes is required
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 6 of 10

6. Question

What is the primary purpose of impact ratios when assessing AI system performance?

To measure the total number of applications processed by an AI system
Correct Incorrect Correct answer

To compare the rate of positive outcomes between different demographic groups
Correct Incorrect Correct answer

To calculate the cost efficiency of AI implementation across departments
Correct Incorrect Correct answer

To determine the technical accuracy of machine learning models
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 7 of 10

7. Question

Individual fairness requires that similar individuals receive similar treatment from AI systems.

True
Correct Incorrect Correct answer

False
Correct Incorrect Correct answer

Only in healthcare applications
Correct Incorrect Correct answer

It depends on the industry
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 8 of 10

8. Question

According to the lesson, what is a fundamental tension in fairness definitions within AI systems?

They are too complex to implement in practice
Correct Incorrect Correct answer

Different definitions make different assumptions about equitable treatment
Correct Incorrect Correct answer

They are not aligned with existing legal frameworks
Correct Incorrect Correct answer

They require too much computational power to test
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 9 of 10

9. Question

According to ISO 25012:2011 clause 7.2.3, what is a key requirement for bias testing in AI systems?

Testing must include at least 100 observations per group
Correct Incorrect Correct answer

Testing must provide sufficient statistical power to identify relevant differences
Correct Incorrect Correct answer

Testing must use exact statistical tests for all subpopulations
Correct Incorrect Correct answer

Testing must always involve clinical decision support systems
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points
Question 10 of 10

10. Question

What is a key limitation of traditional bias testing according to the lesson?

It only considers gender and ethnicity
Correct Incorrect Correct answer

It fails to account for multiple overlapping identities
Correct Incorrect Correct answer

It is too expensive to implement
Correct Incorrect Correct answer

It cannot be applied to UK organisations
Correct Incorrect Correct answer

Correct / 1 Points

Incorrect / 1 Points

Login

Accessing this course requires a login. Please enter your credentials below!

Username or Email Address

Password

Remember Me

Continue with Facebook

Continue with Google

Lost Your Password?

Register

Don't have an account? Register one!

Register an Account

Keep building practical AI skills

Short, practical courses from Tesseract Academy, founded by Dr Stelios Kampakis, CStat.

Explore Individual Learning

© 2026 Tesseract Academy. All Rights Reserved.