Certification Practice: Ethics & Trusted Principles

Question 1

A Salesforce consultant is configuring Einstein to help the HR team screen resumes. It is discovered that the model is disproportionately recommending candidates from a specific demographic over another, despite having similar qualifications. What type of AI ethical risk does this represent?

Accepted Answer

Bias

Answer

Hallucination

Answer

Toxicity

Answer

Ungroundedness

Question 2

Which of the following are core principles of Trusted AI defined by Salesforce? (Select 3)

Accepted Answer

Answer

Speed

Answer

Accountability

Answer

Safety

Answer

Empowerment

Answer

Honesty

Question 3

A customer service agent uses Einstein Copilot to draft a response to an angry customer. Before sending the email, the agent reads the draft, corrects the tone, and verifies the data. What AI ethics concept is being applied here?

Accepted Answer

Human-in-the-Loop (HITL)

Answer

Zero Data Retention

Answer

Data Grounding

Answer

Prompt Engineering

Question 4

How does the Einstein Trust Layer help prevent the exposure of Personally Identifiable Information (PII) when a prompt is sent to an external LLM?

Accepted Answer

Using Data Masking to replace PII with generic text

Answer

Encrypting the data with a key only OpenAI has

Answer

Storing the data in Salesforce Data Cloud before sending

Answer

Asking user consent via a popup

Question 5

A financial company wants to use AI to recommend investment plans. Which Trusted AI principle dictates that AI must clearly identify itself as a machine and not pretend to be a human advisor?

Accepted Answer

Honesty

Answer

Transparency

Answer

Safety

Answer

Accuracy

Question 6

Which component of the Einstein Trust Layer ensures that the LLM provider (like OpenAI) does not retain customer data after processing a request?

Accepted Answer

Zero Data Retention Policy

Answer

Toxicity Detection

Answer

Secure Gateway

Answer

Data Masking

Question 7

A Generative AI model creates a very convincing and detailed response about a product the company has never sold. What is this phenomenon called?

Accepted Answer

Hallucination

Answer

Inaccurate Prediction

Answer

Algorithmic Bias

Answer

Overfitting

Question 8

What is the primary purpose of the 'Empowerment' principle in Salesforce AI?

Accepted Answer

Augment human capabilities and make employees more productive

Answer

Replace employees with autonomous agents to cut costs

Answer

Give AI the power to make legal decisions automatically

Answer

Allow AI to train itself without supervision

Question 9

Which of the following actions help reduce the carbon footprint of AI models, aligning with the Sustainability principle? (Select 2)

Accepted Answer

Answer

Always train the largest model possible to ensure accuracy

Answer

Use right-sized models for the specific task

Answer

Reuse pre-trained models instead of training from scratch

Answer

Run all models on on-premise servers instead of the cloud

Question 10

An administrator wants to ensure that Einstein does not generate responses containing hate speech or violence. Which feature of the Einstein Trust Layer handles this?

Accepted Answer

Toxicity Detection

Answer

Confidence Scoring

Answer

Data Masking

Answer

Log Auditing