Microsoft AI-103 Practice Test 34 - Prepaway
Practice Exams:

Pass Microsoft AI-103 Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers!

30 Days Free Updates, Instant Download!

Verified By Experts
Premium File
$89.99 $98.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers. Last Update: Nov 15, 2026
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 34

Question 330

Production requirement: A university removes a former student’s restricted case file and needs its embeddings removed from retrieval results. Which TWO decisions would undermine the required system boundary? Select TWO answers.

  1. Keep deleted restricted documents in the vector index indefinitely without a removal plan
  2. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions
  3. Inspect records that explain the affected process, then check the result described in this case: A university removes a former student’s restricted case file and needs its embeddings removed from retrieval results
  4. Expose all indexed HR documents because the frontend chatbot itself requires login
  5. Use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval

Explanation

Correct Answer

A:

“Keep deleted restricted documents in the vector index indefinitely without a removal plan” would violate the requirement. The appropriate control instead is to use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

D:

“Expose all indexed HR documents because the frontend chatbot itself requires login” would violate the requirement.

Incorrect Answers

B:

“A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions” is the required outcome of a working control, not an action that weakens it.

C:

“Inspect records that explain the affected process, then check the result described in this case: A university removes a former student’s restricted case file” adds a relevant safeguard or verification step rather than weakening the solution.

E:

“Use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval” is the controlled implementation, not an unsafe workaround. Confirm that a changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

Question 331

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. Which design decision is justified by the failure and the required outcome?

  1. Run indexer once and never recheck policy updates
  2. Use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval
  3. Apply the proposed setting to all connected workloads before testing the case-specific boundary: a retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions
  4. Apply security trimming only to answer text after retrieval exposes raw restricted snippets

Explanation

Correct Answer

B:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. Confirm that a changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions. If the result fails, inspect indexer runs, source timestamps, deleted-document detection, per-record ACL fields, filters and query principal context.

Incorrect Answers

A:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. “Run indexer once and never recheck policy updates” does not implement the control that the facts call for. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. Instead, use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

C:

“Apply the proposed setting to all connected workloads before testing the case-specific boundary: a retailer grants regional teams different supplier documents” does not implement the control that the facts call for. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. Instead, use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

D:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. “Apply security trimming only to answer text after retrieval exposes raw restricted snippets” does not implement the control that the facts call for. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. Instead, use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

Question 332

Situation: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. Which underlying technical principle explains the recommended implementation?

  1. The fact that the team started work is enough to prove correctness for this exact case: a retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions
  2. Successful provisioning by itself proves all security, output, and operational constraints are satisfied
  3. The most powerful or least restricted model is always the safest choice regardless of task conditions
  4. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated

Explanation

Correct Answer

D:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. Use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

Incorrect Answers

A:

“The fact that the team started work is enough to prove correctness for this exact case: a retailer grants regional teams different supplier documents” does not fully explain the decisive technical boundary in this case. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. The implementation must use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

B:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. “Successful provisioning by itself proves all security, output, and operational constraints are satisfied” does not fully explain the decisive technical boundary in this case. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. The implementation must use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval.

C:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. “The most powerful or least restricted model is always the safest choice regardless of task conditions” does not fully explain the decisive technical boundary in this case. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

Question 333

Architecture case: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. What observable acceptance condition should be required before closing the task?

  1. A project dashboard remains green, although the affected request, output and permissions were not checked
  2. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions
  3. The team closes a change ticket without replaying a representative positive and negative test
  4. The operator reports success without reproducing the observed behavior: a retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions

Explanation

Correct Answer

B:

A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

Incorrect Answers

A:

“A project dashboard remains green, although the affected request, output and permissions were not checked” would not meet the requirement. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

C:

A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. “The team closes a change ticket without replaying a representative positive and negative test” alone does not establish all the required behavior. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

D:

“The operator reports success without reproducing the observed behavior: a retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions” would not meet the requirement. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

Question 334

Production requirement: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. Which first investigation is more useful than an unrelated health indicator?

  1. Assume the general service is at fault without checking the affected operation in this case: a retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions
  2. Inspect indexer runs, source timestamps, deleted-document detection, per-record ACL fields, filters and query principal context
  3. Restart every affected component before collecting request or tool traces
  4. Read only the broad service-health summary and skip actual workload evidence

Explanation

Correct Answer

B:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. Inspect indexer runs, source timestamps, deleted-document detection, per-record ACL fields, filters and query principal context. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

Incorrect Answers

A:

“Assume the general service is at fault without checking the affected operation in this case: a retailer grants regional teams different supplier documents” would not meet the requirement. Inspect indexer runs, source timestamps, deleted-document detection, per-record ACL fields, filters and query principal context. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

C:

“Restart every affected component before collecting request or tool traces” would not be the most direct first check for this failure. Inspect indexer runs, source timestamps, deleted-document detection, per-record ACL fields, filters and query principal context.

D:

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. “Read only the broad service-health summary and skip actual workload evidence” would not meet the requirement. Inspect indexer runs, source timestamps, deleted-document detection, per-record ACL fields, filters and query principal context. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

Question 335

A team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions. Which TWO actions could conceal a serious failure while appearing faster? Select TWO answers.

  1. A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions
  2. Keep deleted restricted documents in the vector index indefinitely without a removal plan
  3. Expose all indexed HR documents because the frontend chatbot itself requires login
  4. Use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval
  5. Inspect records that explain the affected process, then check the result described in this case: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions

Explanation

Correct Answer

B:

“Keep deleted restricted documents in the vector index indefinitely without a removal plan” would violate the requirement. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

C:

“Expose all indexed HR documents because the frontend chatbot itself requires login” would violate the requirement. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

Incorrect Answers

A:

Because a team is investigating this task: A retailer grants regional teams different supplier documents and finds query results ignore the source repository permissions, “A changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions” is the required outcome of a working control, not an action that weakens it. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

D:

“Use incremental indexing and query-time security filters that carry source-document access permissions and deletion events into retrieval” is the controlled implementation, not an unsafe workaround. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated.

E:

“Inspect records that explain the affected process, then check the result described in this case: A retailer grants regional teams different supplier documents” adds a relevant safeguard or verification step rather than weakening the solution. An AI Search index can hold stale data or incomplete access metadata unless ingestion schedules, tombstones and authorization filters are validated. Confirm that a changed or deleted record stops appearing after the expected sync window and callers never retrieve documents outside their permissions.

Question 336

Situation: A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. Which implementation best meets the requirement while honoring the stated constraints?

  1. Assume cosine similarity can compare arbitrary unequal-length arrays
  2. Modify vector dimensions only in a client-side constant
  3. Apply the proposed setting to all connected workloads before testing the case-specific boundary: a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests
  4. Align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization

Explanation

Correct Answer

D:

Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Confirm that new model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall. If the result fails, inspect embedding deployment versions, vector dimension declarations, sample vector lengths, indexing logs and retrieval regression cases.

Incorrect Answers

A:

A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. “Assume cosine similarity can compare arbitrary unequal-length arrays” would not meet the requirement. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Instead, align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization.

B:

Because a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests, “Modify vector dimensions only in a client-side constant” does not implement the control that the facts call for. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Instead, align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization.

C:

“Apply the proposed setting to all connected workloads before testing the case-specific boundary: a product search service switches embedding deployments” does not implement the control that the facts call for. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Instead, align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization.

Question 337

Architecture case: A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. Which design boundary is most relevant to this decision?

  1. The most powerful or least restricted model is always the safest choice regardless of task conditions
  2. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval
  3. The fact that the team started work is enough to prove correctness for this exact case: a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests
  4. Successful provisioning by itself proves all security, output, and operational constraints are satisfied

Explanation

Correct Answer

B:

Because a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests, query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization.

Incorrect Answers

A:

“The most powerful or least restricted model is always the safest choice regardless of task conditions” does not fully explain the decisive technical boundary in this case. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. The implementation must align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization.

C:

“The fact that the team started work is enough to prove correctness for this exact case: a product search service switches embedding deployments” does not fully explain the decisive technical boundary in this case. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

D:

A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. “Successful provisioning by itself proves all security, output, and operational constraints are satisfied” does not fully explain the decisive technical boundary in this case. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

Question 338

Production requirement: A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. Which evidence is strongest for proving this requirement was met?

  1. The team closes a change ticket without replaying a representative positive and negative test
  2. A project dashboard remains green, although the affected request, output and permissions were not checked
  3. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall
  4. The operator reports success without reproducing the observed behavior: a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests

Explanation

Correct Answer

C:

A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall.

Incorrect Answers

A:

“The team closes a change ticket without replaying a representative positive and negative test” alone does not establish all the required behavior. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

B:

Because a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests, “A project dashboard remains green, although the affected request, output and permissions were not checked” would not meet the requirement. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall.

D:

“The operator reports success without reproducing the observed behavior: a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests” alone does not establish all the required behavior. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

Question 339

A team is investigating this task: A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. What targeted records should the responder inspect to distinguish configuration from runtime behavior?

  1. Inspect embedding deployment versions, vector dimension declarations, sample vector lengths, indexing logs and retrieval regression cases
  2. Restart every affected component before collecting request or tool traces
  3. Assume the general service is at fault without checking the affected operation in this case: a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests
  4. Read only the broad service-health summary and skip actual workload evidence

Explanation

Correct Answer

A:

Because a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests, inspect embedding deployment versions, vector dimension declarations, sample vector lengths, indexing logs and retrieval regression cases. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall.

Incorrect Answers

B:

A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. “Restart every affected component before collecting request or tool traces” would not be the most direct first check for this failure. Inspect embedding deployment versions, vector dimension declarations, sample vector lengths, indexing logs and retrieval regression cases. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

C:

“Assume the general service is at fault without checking the affected operation in this case: a product search service switches embedding deployments” would not meet the requirement. Inspect embedding deployment versions, vector dimension declarations, sample vector lengths, indexing logs and retrieval regression cases. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

D:

“Read only the broad service-health summary and skip actual workload evidence” would not meet the requirement. Inspect embedding deployment versions, vector dimension declarations, sample vector lengths, indexing logs and retrieval regression cases.

Question 340

Situation: A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests. Which TWO proposed practices create material safety, correctness, or reliability risks? Select TWO answers.

  1. Switch the query embedding model but leave all stored vectors generated with an incompatible dimension
  2. Inspect records that explain the affected process, then check the result described in this case: A product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests
  3. New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall
  4. Align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization
  5. Silently truncate vectors to fit the old index schema without validating semantic retrieval

Explanation

Correct Answer

A:

Because a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests, “Switch the query embedding model but leave all stored vectors generated with an incompatible dimension” would violate the requirement. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

E:

“Silently truncate vectors to fit the old index schema without validating semantic retrieval” would violate the requirement. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

Incorrect Answers

B:

“Inspect records that explain the affected process, then check the result described in this case: A product search service switches embedding deployments” adds a relevant safeguard or verification step rather than weakening the solution. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval.

C:

Because a product search service switches embedding deployments and starts getting dimension-mismatch errors on query requests, “New model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall” is the required outcome of a working control, not an action that weakens it. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Confirm that new model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall.

D:

“Align the embedding model, vector-field dimensions and index migration strategy before changing document or query vectorization” is the controlled implementation, not an unsafe workaround. Query vectors and indexed vectors must have compatible dimensions and semantic representation; changing embedding models without reindexing can break similarity retrieval. Confirm that new model query vectors work with appropriately migrated indexed vectors and previously relevant test queries maintain acceptable recall.

Developing AI Apps and Agents on Azure (AI-103) Practice Tests