Artificial Intelligence Workloads

Question 1

What is the key idea behind the Fairness principle in AI?

Answer 1

AI systems should treat all people fairly

Question 2

What risk does the Fairness principle address?

Answer 2

Bias in training data

Question 3

Name three actions to ensure AI fairness

Answer 3

Ensure all people are treated fairly, avoid introducing bias from training data, conduct comprehensive testing to assess bias

Question 4

What Azure tool can help show feature influence for bias?

Answer 4

Azure ML

Question 5

What open-source tool can help improve fairness in AI?

Answer 5

Fairlearn

Question 6

What is the key idea behind Reliability and Safety in AI?

Answer 6

AI systems should perform reliably and safely

Question 7

What risk does Reliability and Safety address?

Answer 7

Errors in the model/system behavior

Question 8

Name three critical scenarios where AI reliability and safety are crucial

Answer 8

Self-driving cars, health diagnosis, autonomous weapons

Question 9

What should be done if AI mistakes happen?

Answer 9

Release a report quantifying risks and harms to end users

Question 10

What is the key idea behind Privacy and Security in AI?

Answer 10

AI systems should be secure and respect privacy

Question 11

What risk does Privacy and Security address?

Answer 11

Data being exposed or used inappropriately

Question 12

What does PII stand for in AI privacy contexts?

Answer 12

Personally identifiable information

Question 13

How does Edge Computing help with privacy?

Answer 13

Running models locally on a user’s device keeps PII on the device, avoiding vulnerability

Question 14

What is the key idea behind Inclusiveness in AI?

Answer 14

AI systems should empower everyone and engage people

Question 15

What risk does Inclusiveness address?

Answer 15

The AI not working for certain segments of the population

Question 16

What design principle can help with inclusiveness?

Answer 16

Designing solutions for minority users can help design for the majority

Question 17

What is the key idea behind Transparency in AI?

Answer 17

AI systems should be understandable

Question 18

What risk does Transparency address?

Answer 18

Lack of trust in what the AI is doing

Question 19

What are the two key concepts related to AI transparency?

Answer 19

Interpretability and intelligibility

Question 20

Name three benefits of AI transparency

Answer 20

Mitigates unfairness, aids developers in debugging, and gains user trust

Question 21

What is the key idea behind Accountability in AI?

Answer 21

People should be accountable for AI systems

Question 22

What risk does Accountability address?

Answer 22

Unclear responsibility when the AI does something wrong

Question 23

Who should be accountable for AI systems?

Answer 23

Developers, the company, and directors

Question 24

What frameworks should AI systems work within?

Answer 24

Frameworks of governance, organizational principles, ethical, and legal standards

Question 25

List all six Responsible AI Principles

Answer 25

Fairness, Reliability and Safety, Privacy and Security, Inclusiveness, Transparency, Accountability

Question 26

What are two ways to mitigate visual bias in AI interfaces?

Answer 26

Using gender-neutral icons or showing diverse representations

Question 27

What should be considered when using data for AI training regarding privacy?

Answer 27

Allowed source and whether data is sufficiently scrubbed

Question 28

Name five security considerations for AI data

Answer 28

Data origin, lineage, use (internal vs. external), corruption, and anomaly detection

Question 29

What is the current trend regarding AI regulation?

Answer 29

There is a push towards regulation based on AI principles

Question 30

How can open-source AI frameworks help with transparency?

Answer 30

They can provide technical transparency

Question 31

What improves transparency in AI-driven systems?

Answer 31

Providing explanations for AI-driven suggestions or actions