AI Associate Practice Exam 2025: Latest Questions

An AI model used for lead scoring is consistently giving lower scores to leads from a particular region, even when their conversion rates are similar to other regions. What is the BEST first step to address this issue?

A service team wants an AI feature that drafts case replies based on the case description and prior knowledge articles. What is the most appropriate expectation to set with users?

An admin is preparing Salesforce data for an AI use case and notices many duplicate Contact records. Why is deduplication important for AI outcomes?

A sales manager wants an AI-generated summary of a long Opportunity record including key activities, risks, and next steps. Which approach is the most appropriate use of AI in Salesforce?

A company wants to use generative AI to create email responses. Security requires that the AI only use approved knowledge articles and the current customer’s case details, not arbitrary internet content. Which design best supports this requirement?

A team says, "Our AI predictions look accurate overall," but stakeholders are concerned it performs poorly for a small subset of customers. What should the team do to evaluate this concern?

An admin is asked to prepare data for an AI use case that predicts case escalation risk. Which set of fields is MOST likely to be useful features while minimizing privacy risk?

A service center deploys an AI-generated response feature. After launch, agents report the drafts are often off-topic for a specific case type. What is the BEST troubleshooting action?

A company wants to use AI to generate personalized product recommendations in Salesforce. Their data is spread across Sales Cloud, a data warehouse, and a website analytics tool. What is the MOST important architectural consideration to ensure reliable recommendations?

A regulated financial services org wants to roll out generative AI assistance to advisors in Salesforce. They must prevent the AI from exposing confidential client data to unauthorized users and must be able to audit how responses were generated. Which approach best addresses these requirements?

A service team wants AI-generated reply suggestions in their support console, but leadership requires agents to approve any text before it is sent to customers. Which approach best meets this requirement?

An admin is asked to explain why an AI feature should not be trained on all available customer notes by default. Which risk is the best justification?

A sales manager wants to use AI insights to improve forecasting. Which data characteristic is MOST important to validate first to reduce misleading predictions?

A company wants an AI assistant to answer questions like, "What is our return policy?" using only an approved knowledge base and not the open internet. What is the best design choice?

Users report that an AI summary of cases frequently omits key troubleshooting steps found in the case feed. Which is the most likely root cause to investigate first?

A marketing team wants AI to help draft personalized outreach but must ensure the content does not include protected health information (PHI). What is the most appropriate control to apply?

A team wants to roll out an AI feature to thousands of users. They need to validate that outputs remain consistent with policy and that user feedback is captured for continuous improvement. Which rollout approach is best?

An AI feature is producing biased recommendations that consistently disadvantage a specific customer segment. What is the most appropriate first response?

A company wants an AI assistant to answer questions across Sales and Service data, but only return results the user is allowed to see. Which architectural principle best addresses this requirement?

A team is grounding a generative AI assistant on internal documents. Users complain that answers are accurate but not reproducible: the same question yields different results each time, complicating audits. What is the best corrective action?

A service team wants AI-generated replies for Email-to-Case, but leadership is concerned about agents sending incorrect information. What is the best way to reduce risk while still gaining productivity benefits?

A company is preparing CRM data to support AI features. They notice the same customer appears multiple times across Leads and Contacts with slightly different names and emails. What data practice most directly improves AI results in this situation?

A sales manager wants to use generative AI to summarize a customer’s recent interactions in Salesforce. Which approach best supports data governance and least-privilege access?

After enabling an AI feature that relies on CRM data, users report irrelevant suggestions. Investigation shows many Opportunity records have missing Close Date and inconsistent Stage values. What is the most likely root cause?

A company is piloting a generative AI assistant to help agents answer policy questions. During testing, the assistant occasionally invents policy details that are not in the approved knowledge base. What is the best mitigation strategy?