Module Assessment

1.

Why might a generative AI model produce different responses to similar prompts?

It predicts likely content from learned patterns instead of retrieving a fixed response

It retrieves the same fixed response from a database for every request

It ignores the prompt and selects content at random

It updates its training data after every prompt

2.

An AI tool produces lower-quality recommendations for one group of customers than for similar customers in another group. Which responsible AI principle is most relevant?

Fairness

Reliability and safety

Privacy and security

Transparency

3.

You want Copilot to summarize a customer file marked confidential. What should you do first?

Check the sensitivity label and your organization's data-handling policies, then share only the information the task requires

Remove the label and submit the entire file

Submit the file because Copilot automatically approves every use of sensitive data

Paste the file into a personal AI service to avoid organizational restrictions

4.

An AI tool drafts a customer announcement. The approved pilot summary says: "In a four-week pilot with 12 employees, average weekly report-preparation time fell from 50 to 40 minutes. Results for other teams have not been evaluated." The draft says: "Our reporting tool reduces weekly report-preparation time by 20% for every team." What should you do?

Keep the draft because a drop from 50 to 40 minutes is a 20% reduction

Report the 20% reduction as the result of the 12-employee pilot, and remove the claim about every team

Remove the 20% figure and keep the claim that every team benefits

Ask the AI whether its claim is accurate, and publish the draft if it says yes

5.

A team needs a reusable tracker built from a plain-language description. Which Microsoft Copilot mode best fits this task?

Chat

Cowork

Code

Autopilot

Opinia