AI Associate Ethical Considerations of AI Practice Test 17 — 15 Questions

Question 1

A Salesforce developer is building an AI-powered chatbot for a healthcare company. The chatbot will provide general health information and triage symptoms. The developer must ensure the AI adheres to ethical principles. Which two actions should the developer take to promote transparency and accountability? (Choose two.)

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Ensure the chatbot clearly discloses that it is an AI and not a human.. Logging AI interactions and disclosing the chatbot's AI nature both enhance transparency and accountability. Logging provides a record for audits and improvement, while disclosure ensures users are informed. Together, they build trust and support responsible AI deployment in healthcare, where ethical standards are paramount.

Answer

Use a proprietary algorithm that cannot be explained to users to protect trade secrets.

Answer

Train the chatbot on a dataset that includes only positive patient outcomes to avoid negative bias.

Answer

Allow the chatbot to provide definitive medical diagnoses without human oversight.

Question 2

A Salesforce developer is implementing an AI-powered chatbot for customer service that uses Einstein Bots and an LLM integration. The company wants to ensure the solution adheres to ethical AI principles, particularly around transparency and user consent. Which two actions should the developer take? (Choose two.)

Accepted Answer

Allow users to opt out of AI interactions and connect with a human agent.. Transparency and user consent are fundamental ethical principles for AI chatbots. Informing users they are interacting with AI and allowing them to opt out to a human agent respects their autonomy and ensures informed interaction. Other actions like indefinite data storage or lack of content moderation violate ethical guidelines.

Answer

Collect user feedback only after the conversation ends and only if the user initiates it.

Answer

Store all conversation logs indefinitely for future model training.

Answer

Use the LLM to generate responses without any content moderation.

Question 3

A Salesforce developer is integrating an AI chatbot into a customer service portal. The chatbot uses natural language processing to handle customer inquiries. To ensure ethical AI, which practice should be prioritized during the design phase?

Accepted Answer

Implement clear disclosure that the user is interacting with an AI and provide an option to escalate to a human agent.. Disclosing AI interaction and offering human escalation are fundamental ethical practices. They ensure transparency, respect user choice, and prevent deception. Salesforce's ethical AI guidelines emphasize transparency and user control. The other options either violate privacy, deceive users, or collect data without consent, all of which are unethical. Thus, clear disclosure and escalation options are the most ethical design choices.

Answer

Train the chatbot to mimic human emotions to create a more engaging user experience.

Answer

Use the chatbot to collect additional personal data, such as phone numbers, without explicit consent to improve future interactions.

Answer

Ensure the chatbot can access all customer data to provide personalized responses.

Question 4

A Salesforce administrator is implementing Einstein AI for customer service chatbots. To ensure ethical AI practices, which two actions should be taken? (Choose two.)

Accepted Answer

Regularly audit the chatbot's responses for bias and accuracy.. Regularly auditing the chatbot for bias and accuracy ensures fairness and accountability, while disclosing AI interaction upholds transparency. These actions align with Salesforce's responsible AI principles, which emphasize fairness, transparency, and accountability. Together, they help build trust and mitigate ethical risks in AI deployment.

Answer

Use only synthetic data for training to avoid any privacy concerns.

Answer

Deploy the chatbot without testing to quickly gather user feedback.

Answer

Allow the chatbot to access all customer data without restrictions to improve personalization.

Question 5

A Salesforce administrator is configuring Einstein Discovery to analyze customer churn. The dataset includes sensitive attributes such as gender and ethnicity. The administrator wants to ensure the model does not use these attributes to make predictions. Which feature should they use?

Accepted Answer

Field Exclusion. Field Exclusion is the correct feature because it directly prevents specified fields from being used as predictors in Einstein Discovery. By excluding gender and ethnicity, the administrator ensures the model does not learn from these sensitive attributes, thus mitigating bias. Other options either focus on post-hoc detection or data privacy, not on proactive feature exclusion during model training.

Answer

Einstein Trust Layer

Answer

Data Masking

Answer

Bias Detection

Question 6

A company uses an Einstein Discovery model to recommend which customers should receive a retention offer. The model was trained on historical data where offers were disproportionately given to male customers. To mitigate bias, the data science team plans to use a fairness metric. Which metric would best detect disparate impact against female customers?

Accepted Answer

Demographic parity difference. Demographic parity difference directly compares the proportion of positive outcomes between groups. If female customers receive retention offers at a lower rate than male customers, this metric will show a disparity, indicating potential bias. It is a standard fairness metric for detecting disparate impact and aligns with ethical AI practices of monitoring group outcomes.

Answer

Mean squared error

Answer

Precision-recall AUC

Answer

Overall model accuracy

Question 7

A Salesforce admin is setting up Einstein Trust Layer for a new generative AI feature. The company's ethical AI policy requires that all AI-generated content be reviewed for toxic or harmful language before being shown to customers. Which Einstein Trust Layer capability should the admin enable?

Accepted Answer

Toxicity Detection. Toxicity Detection is the Einstein Trust Layer feature that identifies and mitigates harmful or offensive language in AI-generated responses. By enabling it, the admin ensures that customers are not exposed to toxic content, aligning with the company's ethical AI policy. Other features address different concerns such as data privacy or auditability.

Answer

Zero Data Retention

Answer

Audit Trail

Answer

Data Masking

Question 8

A data scientist at a financial services company is using Salesforce Einstein Discovery to build a model that predicts customer churn. The model uses historical customer data including age, income, and postal code. After deployment, the compliance team notices that the model's predictions disproportionately favor customers from affluent neighborhoods. Which ethical principle is most directly violated by this outcome?

Accepted Answer

Fairness, because the model produces biased outcomes against certain socioeconomic groups.. The model's predictions disproportionately favor affluent neighborhoods, which is a clear case of algorithmic bias. Fairness in AI demands that models do not create or reinforce discrimination against protected groups. While transparency, accountability, and reliability are also important ethical principles, the specific violation here is fairness, as the model produces inequitable outcomes based on socioeconomic status.

Answer

Transparency, because the model's decision-making process is not explainable to customers.

Answer

Accountability, because no one is responsible for the model's predictions.

Answer

Reliability, because the model's predictions are not consistently accurate.

Question 9

A Salesforce admin is configuring Einstein Prediction Builder to predict which customers are likely to default on payments. The model will use historical data including payment history, credit score, and demographic information. Which TWO actions should the admin take to ensure the model aligns with ethical AI principles? (Choose two.)

Accepted Answer

Regularly audit the model's predictions for disparate impact across demographic groups.. Regular audits for disparate impact and thorough documentation are essential ethical AI practices. Audits ensure fairness by detecting bias, while documentation provides transparency and accountability. Together, they align with Salesforce's responsible AI principles. The other options either introduce bias, skip necessary checks, or remove human oversight, all of which are unethical. Thus, auditing and documenting are the correct actions.

Answer

Use the model to make final decisions without human review to increase efficiency.

Answer

Deploy the model immediately to test its real-world performance before conducting any bias checks.

Answer

Include all available demographic data to maximize predictive accuracy.

Question 10

A company is implementing an AI-powered chatbot using Salesforce Einstein Bots to handle customer inquiries. To ensure ethical AI practices, which two actions should the company take? (Choose two.)

Accepted Answer

Clearly disclose to users that they are interacting with an AI and not a human.. Ethical AI chatbots must prioritize transparency and user control. Disclosing AI interaction ensures users are informed, and providing an escalation path to humans respects their needs for complex issues. These actions build trust and align with principles of honesty, accountability, and user-centric design. The other options involve deception, overcollection of data, or aggressive tactics that undermine ethical standards.

Answer

Collect as much personal data as possible to improve the chatbot's responses.

Answer

Provide a seamless experience by mimicking human responses without indicating AI involvement.

Answer

Use the chatbot to upsell products aggressively to maximize revenue.

Question 11

A company uses Salesforce Einstein to predict employee attrition. The model's recommendations lead to managers taking preemptive actions that disproportionately affect older employees. Which ethical principle is most directly violated?

Accepted Answer

Fairness. The scenario describes an AI system whose recommendations lead to actions that disproportionately affect older employees, which is a clear case of age discrimination. This violates the ethical principle of fairness, which demands that AI systems treat all individuals equitably and do not disadvantage protected groups. Fairness is a core component of Salesforce's responsible AI principles, and addressing this would require bias testing and mitigation.

Answer

Accountability

Answer

Privacy

Answer

Transparency

Question 12

A Salesforce developer is building an AI-powered chatbot to handle customer inquiries. The chatbot will have access to customer account details. Which action best aligns with the ethical principle of transparency?

Accepted Answer

Clearly informing users that they are interacting with an AI and providing an option to escalate to a human agent.. Transparency in AI means users should be aware when they are interacting with an AI system and understand its capabilities and limitations. Informing users that they are chatting with a bot and providing a human escalation option ensures they are not misled and can choose an alternative if desired. This practice builds trust and is a fundamental ethical requirement for customer-facing AI.

Answer

Limiting the chatbot's responses to a predefined set of FAQs to ensure accuracy.

Answer

Using a complex neural network to generate responses that mimic human writing style.

Answer

Storing all customer conversations in a secure database for future model training.

Question 13

A Salesforce administrator at a retail bank configures Einstein Trust Layer for a new customer service agent that summarizes case notes. The administrator enables data masking for social security numbers and credit card numbers. Which ethical principle does this configuration primarily uphold?

Accepted Answer

Privacy. Data masking in Einstein Trust Layer protects sensitive personal information from being exposed to or processed by the AI model. This directly supports privacy by preventing unauthorized access to social security and credit card numbers. While transparency, accountability, and fairness are also ethical principles, the specific action of masking sensitive data is most closely tied to privacy. The configuration ensures that the AI system does not handle raw sensitive data, reducing risk of data breaches.

Answer

Fairness

Answer

Accountability

Answer

Transparency

Question 14

A Salesforce consultant is advising a client on the ethical use of AI for lead scoring. The client wants to ensure their AI system respects user privacy and complies with regulations. Which action should the consultant recommend as a best practice?

Accepted Answer

Implement data minimization and obtain explicit consent for data usage.. Implementing data minimization and obtaining explicit consent are core ethical AI practices. They protect user privacy, ensure compliance with regulations, and build trust. Salesforce's ethical guidelines strongly advocate for these principles. The other options involve overcollection, non-disclosure, or ignoring laws, all of which are unethical and risky. Thus, data minimization and explicit consent are the best recommendations.

Answer

Collect as much personal data as possible to improve lead scoring accuracy.

Answer

Ignore privacy regulations as they are not enforceable for AI systems.

Answer

Use third-party data without informing users to gain a competitive edge.

Question 15

A data scientist at a healthcare company is using Salesforce Einstein Discovery to predict patient readmission risks. The model uses historical patient data, including socioeconomic status. The data scientist notices that the model's predictions are less accurate for patients from lower socioeconomic backgrounds. Which ethical principle is being violated?

Accepted Answer

Fairness. The model's reduced accuracy for patients from lower socioeconomic backgrounds indicates unequal performance, which is a fairness violation. Fairness in AI requires equitable treatment and outcomes across groups, and this model fails to provide reliable predictions for all patients. Addressing this requires bias detection and mitigation to ensure the model serves all demographics equitably, aligning with ethical AI principles.

Answer

Reliability

Answer

Accountability

Answer

Transparency