Free Practice Quiz Question List

7 AI Limitations and Responsible Use Online Quiz Questions

Use this free practice quiz with 20 questions to review 7 AI Limitations and Responsible Use, test your knowledge, and prepare for your next test or exam.

20 questions

Choose one

1 point

An AI system performs well on familiar examples, but the data and conditions change in a new setting. What is the most responsible conclusion?

A
Assume it will perform just as well because the task is similar.
B
Treat its results as less certain and test it in the new setting.
C
Use its results without checking if they appear confident.
D
Stop evaluating it because it performed well before.

Written response

1 point

What term describes a fluent generative AI answer that contains false details or invented sources?

True or false

1 point

True or false: If an AI-generated answer sounds polished and confident, its important claims do not need to be checked against reliable evidence.

A
True
B
False

Fill in the blank

1 point

In the Govern, Map, Measure, and Manage cycle, is the stage for setting policies, roles, and oversight.

Choose one

1 point

A team wants to use an AI tool to help with a task involving confidential client information. Which approach best reflects responsible privacy practice?

A
Enter any available personal information so the system can use it if needed.
B
Use confidential information whenever a tool is convenient, without checking its data practices.
C
Collect only necessary information, protect access, and check data practices and authorization before entering confidential information.
D
Avoid considering privacy until after the system has been deployed.

Choose all

1 point

Select all factors through which bias can enter an AI system or its use.

A
Patterns in the training data
B
The labels used to prepare or describe data
C
Design choices
D
The system's ability to process information quickly
E
The way people use the system

True or false

1 point

True or false: Human oversight is meaningful when a person can assess an AI output and has the ability to challenge or correct it, rather than merely approving it.

A
True
B
False

Fill in the blank

1 point

In the Govern, Map, Measure, and Manage cycle, involves testing performance and risks such as uncertainty, fairness, privacy, security, and safety.

Written response

1 point

What practice should continue after an AI system is launched because its data, operation, and use can change?

Choose one

1 point

A public-service organization is considering an AI system whose errors could seriously affect people. Which plan best matches safeguards to the potential harm?

A
Use the same safeguards for every system, regardless of its possible impact.
B
Rely on the system's past success instead of testing it before use.
C
Allow the system to operate without a way to override or stop it.
D
Match safeguards to potential harm, including testing and a safe way to review, correct, or stop the system.

Choose all

1 point

An organization is documenting an AI system used in decisions that affect people. Select all practices that support accountability and transparency.

A
Tell people when AI meaningfully contributes to a decision that affects them.
B
Keep the system's purpose and limits undocumented to avoid confusion.
C
Document who is responsible for the system and how errors can be challenged or corrected.
D
Document the system's purpose and limitations.

Choose one

1 point

A hiring model has high overall accuracy, but its errors appear more frequent for one group of applicants. What is the best next step?

A
The overall score proves the system treats all groups fairly.
B
Evaluate errors for relevant groups; the overall score may conceal unequal error patterns.
C
Replace group-level evaluation with more training data.
D
Ignore the difference unless the system's average accuracy falls.

Open ended

1 point

An employer plans to use AI to help screen job applications. Explain what the employer should do before and after launch to use the tool responsibly.

True or false

1 point

True or false: A polished, confident AI answer is by itself proof that the answer is correct.

A
True
B
False

Choose one

1 point

An AI system performs well on familiar examples, but the organization plans to use it with different data and in a new setting. What is the most appropriate next step?

A
Assume the earlier results will hold because the system already performed well.
B
Test the system in the new setting and assess how reliable its results are there.
C
Use the system only if its answers sound confident.
D
Increase the amount of training data without checking its relevance.

Choose one

1 point

A team wants to use an AI tool that may process confidential customer information. Which approach best addresses privacy before use?

A
Enter all available personal information so the system has more context.
B
Share sensitive data with the tool first, then check its data practices later.
C
Limit the information to what is needed and verify that the tool's data practices and authorization make its use appropriate.
D
Avoid assessing privacy until the system has been deployed.

Written response

1 point

What term describes a generative AI answer that sounds fluent but contains false details or invented sources?

Choose one

1 point

A hiring model has high overall accuracy, but there is concern that it may make different kinds of errors for different applicant groups. What is the best response?

A
Compare performance and errors across relevant applicant groups, and involve affected people in addressing disparities.
B
Rely on the overall accuracy score if it is high.
C
Assume that using the same model for everyone guarantees fairness.
D
Remove the model's group-level results from evaluation to avoid seeing disparities.

Choose one

1 point

An organization uses AI to contribute to decisions that affect people. Which practice best supports accountability and transparency?

A
Keep the system's role confidential so people are not influenced by it.
B
Explain when AI meaningfully contributes and document responsibility and ways to challenge or correct errors.
C
Tell people only if the system makes every decision without human involvement.
D
Publish the system's purpose but leave responsibility for errors unspecified.

Written response

1 point

What ongoing practice is needed after an AI system is launched because its data, system, or use may change?