Explanation:

Why C is correct:
Salesforce’s Trusted AI Principle of Transparency means that AI systems should provide clear, explainable, and understandable insights into how decisions are made.
This ensures accountability, builds trust with users, and helps identify potential biases or errors in AI-driven outcomes.
Examples include Einstein Prediction Builder’s reason codes, which explain why a prediction was made.

Why A is incorrect:
Customizing AI features relates to configuration, not transparency. Transparency focuses on explainability, not implementation.
Why B is incorrect:
Integrating AI models with workflows is about functionality, not the ethical principle of making AI decisions understandable.

Reference:
Salesforce’s Trusted AI Principles (Official Documentation) highlight Transparency as a core tenet, ensuring AI systems are explainable and auditable.

Explanation:

Einstein Prediction Builder relies heavily on high-quality data to generate accurate predictions. Poor data quality—such as missing values, inconsistent formats, or outdated records—can lead to unreliable models.

To assess and improve data quality, Salesforce recommends using third-party data quality apps available on the AppExchange. These apps can:
Audit and monitor data cleanliness
Identify duplicates and inconsistencies
Validate field completeness and accuracy
Provide dashboards and reports on data health
This approach is proactive and scalable, especially for organizations like Cloud Kicks that want to operationalize AI predictions across large datasets.

📘 Reference:
You can find this recommendation in Salesforce’s documentation and exam prep guides:
Salesforce Help: Einstein Prediction Builder
Salesforce AI Associate: How to Assess Data Quality

🧩 Why Not the Other Options?
A. Build a Data Management Strategy
While important for long-term governance, this is not a direct method for assessing data quality. It’s more about planning and policy.
B. Build reports to expire the data quality
This option is unclear and likely a distractor. Reports can help explore data, but they don’t “expire” data quality.

Explanation:

These three examples are some of the most common and powerful applications of AI within a CRM system. They are all centered on using data to predict future outcomes and provide actionable insights to improve sales, marketing, and customer service.

Predictive Scoring: AI analyzes a lead's or customer's data (e.g., website visits, email opens, demographic information) and assigns a score that indicates their likelihood of converting or making a purchase. This allows sales and marketing teams to prioritize their efforts on the most promising leads.

Forecasting: AI-powered forecasting uses historical sales data, market trends, and other external factors to predict future sales with greater accuracy than traditional methods. This helps businesses with resource planning, inventory management, and strategic decision-making.

Recommendations: This is where AI suggests the "next best action" for a sales representative or a product for a customer. For instance, a CRM might recommend a specific product to a customer based on their past purchases and browsing history (e.g., "Customers who bought this also liked...") or suggest a specific follow-up task for a sales rep to close a deal.

Bottom Line
The three most common examples of AI in CRM are predictive scoring, forecasting, and recommendations. These applications leverage AI to analyze data and provide predictive insights, helping businesses to automate processes, prioritize efforts, and make more informed decisions to enhance customer relationships and drive revenue.

Explanation:

To prevent bias from entering an AI system during its training, you should use diverse training data. Here's why:
Training Data is Key: AI systems learn from the data they're trained on. If the data is not diverse and representative of all the groups the AI will interact with, the system can learn and perpetuate biases present in that data. For example, if an AI for hiring is only trained on data from male candidates, it may learn to favor male applicants and unfairly penalize female applicants.
Diverse Data Creates Fairer Models: Importing diverse and representative data helps to ensure the AI system learns to make fair and equitable decisions across all groups. This means including data from various demographics, backgrounds, and situations to create a more robust and unbiased model.

The other options are incorrect:
A. Use alternative assumptions: This is too vague and doesn't specify a concrete action to address data bias. The core issue lies in the data itself, not just the assumptions.
C. Include Proxy variables: This is a method that can introduce or hide bias, not prevent it. A proxy variable is a substitute for a variable that is difficult to measure directly. For example, using a person's ZIP code as a proxy for their race or income can embed bias into the model, even if the sensitive variable itself is excluded. This is a common pitfall to avoid in ethical AI development.

Explanation:

Using AI to personalize the shopping experience of online customers—through tailored product recommendations, customized offers, or personalized content—enhances customer satisfaction. AI analyzes customer data (e.g., browsing history, preferences, and purchase behavior) to deliver relevant and engaging experiences, making shopping more convenient and enjoyable. Satisfied customers are more likely to return, engage further, and develop loyalty to the brand, ultimately benefiting the organization through increased retention and sales.

Why not A.
Customers are more likely to share personal information with a site that personalizes their experience? While personalization may build trust, there’s no direct evidence that it makes customers more likely to share personal information. In fact, privacy concerns may deter some customers from sharing data, especially if they’re unaware of how their information is used. This option is less directly tied to the primary benefit of personalization.

Why not C.
Customers are more likely to visit competitor sites that personalize their experience? This option describes a negative outcome and contradicts the benefit of personalization. If an organization uses AI effectively for personalization, it reduces the likelihood of customers turning to competitors by providing a superior, tailored experience.

Reference:
Salesforce’s Customer 360 platform emphasizes how AI-driven personalization (e.g., via Salesforce Einstein) enhances customer satisfaction by delivering relevant experiences.
Salesforce’s State of the Connected Customer report highlights that personalized experiences improve customer satisfaction and loyalty, a key benefit for organizations using AI in e-commerce.
General industry insights, such as those from Salesforce blogs, note that AI personalization drives engagement and customer retention in online shopping.