Microsoft AI-103 (Developing AI Apps and Agents on Azure) Practice Test 16

Question 150

The support ticket describes this behavior: A quote assistant produces inconsistent totals after discounts and tax rules are updated. Which engineering change should come first?

  1. Retrieve current authorized passages and cite them when forming the response
  2. Define a typed calculation tool and have the application execute it with validated arguments
  3. Provide image content to a vision-capable model with a bounded question and verification rubric
  4. Implement a controlled request-tool-result loop with a maximum number of steps

Explanation

Correct Answer

B:

The fault signal is that a quote assistant produces inconsistent totals after discounts and tax rules are updated; the targeted fix works because a tool call delegates deterministic computation to trusted code instead of relying on a model's arithmetic approximation, restoring the ability to obtain exact tax or pricing totals from an authoritative business calculation endpoint.

Incorrect Answers

A:

If the failing function were to answer questions about changing internal procedures with source-backed responses, this change could be relevant because RAG retrieves relevant evidence at runtime and gives it to a generative model; model pretraining alone cannot guarantee current policy text; the actual symptom is that a quote assistant produces inconsistent totals after discounts and tax rules are updated, which calls for a different repair.

C:

If the failing function were to reason about a photo and related user instructions together in one supported inference flow, this change could be relevant because a multimodal model can interpret visual input alongside textual context; relying on OCR alone loses non-textual relationships; the actual symptom is that a quote assistant produces inconsistent totals after discounts and tax rules are updated, which calls for a different repair.

D:

If the failing function were to allow a model to request multiple independent data lookups while limiting cost and duration, this change could be relevant because an orchestration loop can inspect tool requests, execute authorized tools, return results, and terminate under explicit limits; the actual symptom is that a quote assistant produces inconsistent totals after discounts and tax rules are updated, which calls for a different repair.

Question 151

The planned deployment includes a mechanism to orchestrate a deterministic sequence around model steps. Which outcome should the QA team measure?

  1. Compare branch decisions against a complete table of allowed status transitions
  2. Trace required step ordering and block completions that omit a mandatory stage
  3. Measure cancellation propagation and discarded tokens after client disconnects
  4. Compare critical-path latency and result consistency with serial execution

Explanation

Correct Answer

B:

The accepted change is to orchestrate a deterministic sequence around model steps; this evaluation checks whether it can execute a known approval workflow where inputs must be validated before a model drafts a response, since explicit workflow stages make ordering and failure handling reproducible; free-form agent planning is unnecessary for a fixed process.

Incorrect Answers

A:

This test concerns the ability to choose among distinct business flows based on a validated account status, for which rules engines offer predictable decisions for simple structured conditions, while language models handle unstructured interpretation; it does not establish whether the change to orchestrate a deterministic sequence around model steps works for the current release.

C:

This test concerns the ability to show partial assistant output to a user while stopping work after the request is canceled, for which streaming delivers incremental output; cancellation and backpressure prevent orphaned inference work and stale interface content; it does not establish whether the change to orchestrate a deterministic sequence around model steps works for the current release.

D:

This test concerns the ability to reduce latency when separate read-only lookups have no ordering dependency, for which parallelizing independent operations shortens critical-path latency but is unsafe when one action requires another's output; it does not establish whether the change to orchestrate a deterministic sequence around model steps works for the current release.

Question 152

The support ticket describes this behavior: A single round of function calling cannot answer a question that requires a second lookup based on the first result. Which engineering change should come first?

  1. Request structured outputs conforming to the expected schema and validate before ingestion
  2. Validate eligibility first, run generation second, and require approval before sending
  3. Use deterministic status rules for routing and reserve the LLM for language-specific subtasks
  4. Implement a controlled request-tool-result loop with a maximum number of steps

Explanation

Correct Answer

D:

The fault signal is that a single round of function calling cannot answer a question that requires a second lookup based on the first result; the targeted fix works because an orchestration loop can inspect tool requests, execute authorized tools, return results, and terminate under explicit limits, restoring the ability to allow a model to request multiple independent data lookups while limiting cost and duration.

Incorrect Answers

A:

If the failing function were to send model-produced classifications into a strict downstream API contract, this change could be relevant because schema-constrained generation helps enforce machine-readable fields and types; application-level validation is still needed for business rules; the actual symptom is that a single round of function calling cannot answer a question that requires a second lookup based on the first result, which calls for a different repair.

B:

If the failing function were to execute a known approval workflow where inputs must be validated before a model drafts a response, this change could be relevant because explicit workflow stages make ordering and failure handling reproducible; free-form agent planning is unnecessary for a fixed process; the actual symptom is that a single round of function calling cannot answer a question that requires a second lookup based on the first result, which calls for a different repair.

C:

If the failing function were to choose among distinct business flows based on a validated account status, this change could be relevant because rules engines offer predictable decisions for simple structured conditions, while language models handle unstructured interpretation; the actual symptom is that a single round of function calling cannot answer a question that requires a second lookup based on the first result, which calls for a different repair.

Question 153

The planned deployment includes a mechanism to use comparative evaluations for prompt changes. Which outcome should the QA team measure?

  1. Count handled failures and confirm consistent error contracts on negative test suites
  2. Compare win rate, task accuracy, and safety failures with statistical uncertainty
  3. Measure answer support and document freshness on questions with recently revised policies
  4. Compute unsupported-claim rate on source-grounded summaries

Explanation

Correct Answer

B:

The accepted change is to use comparative evaluations for prompt changes; this evaluation checks whether it can choose between candidate prompt versions by measured task quality rather than intuition, since paired evaluation on identical cases isolates prompt impact; production traffic shifts make uncontrolled comparisons misleading.

Incorrect Answers

A:

This test concerns the ability to verify that an app recovers safely from empty messages, long context, and invalid tool responses, for which boundary-case testing surfaces parsing and orchestration failures that standard benchmark prompts miss; it does not establish whether the change to use comparative evaluations for prompt changes works for the current release.

C:

This test concerns the ability to answer questions about changing internal procedures with source-backed responses, for which RAG retrieves relevant evidence at runtime and gives it to a generative model; model pretraining alone cannot guarantee current policy text; it does not establish whether the change to use comparative evaluations for prompt changes works for the current release.

D:

This test concerns the ability to catch plausible statements that are not substantiated by the retrieved evidence, for which faithfulness checks whether generated claims follow provided source material, not whether wording is eloquent; it does not establish whether the change to use comparative evaluations for prompt changes works for the current release.

Question 154

The team plans to use token-based credentials rather than embedded keys. Which test provides the clearest evidence that this change works?

  1. Measure answer support and document freshness on questions with recently revised policies
  2. Prove the app can generate responses after removing the static key from configuration
  3. Measure cancellation propagation and discarded tokens after client disconnects
  4. Measure unsupported-answer rate and correct abstentions on out-of-scope questions
  5. Test boundary values and confirm no rejected argument causes a side effect

Explanation

Correct Answer

B:

The accepted change is to use token-based credentials rather than embedded keys; this evaluation checks whether it can authenticate a server-side generative app through Entra where supported, since a workload identity and appropriate RBAC reduce static credential exposure while preserving auditable authorization decisions.

Incorrect Answers

A:

This test concerns the ability to answer questions about changing internal procedures with source-backed responses, for which RAG retrieves relevant evidence at runtime and gives it to a generative model; model pretraining alone cannot guarantee current policy text; it does not establish whether the change to use token-based credentials rather than embedded keys works for the current release.

C:

This test concerns the ability to show partial assistant output to a user while stopping work after the request is canceled, for which streaming delivers incremental output; cancellation and backpressure prevent orphaned inference work and stale interface content; it does not establish whether the change to use token-based credentials rather than embedded keys works for the current release.

D:

This test concerns the ability to avoid fabricating an answer to a question not covered by the organization's knowledge collection, for which an evidence threshold and abstention path distinguish unknown facts from model completion skill; confident text is not proof of source support; it does not establish whether the change to use token-based credentials rather than embedded keys works for the current release.

E:

This test concerns the ability to prevent malformed model-generated parameters from reaching a write-capable API, for which tool schemas guide argument shape, but the application must check types, constraints, identity and authorization before execution; it does not establish whether the change to use token-based credentials rather than embedded keys works for the current release.

Question 155

A cloud engineering team needs its AI workflow to choose among distinct business flows based on a validated account status. What technical approach most directly enables that?

  1. Use token-based credentials rather than embedded keys
  2. Route conditional branches using rules
  3. Run independent tool requests concurrently where safe
  4. Add bounded tool-use loops for multi-step tasks

Explanation

Correct Answer

B:

For the stated requirement to choose among distinct business flows based on a validated account status, rules engines offer predictable decisions for simple structured conditions, while language models handle unstructured interpretation; this is the mechanism that the system must implement.

Incorrect Answers

A:

For a system whose objective is to authenticate a server-side generative app through Entra where supported, a workload identity and appropriate RBAC reduce static credential exposure while preserving auditable authorization decisions; this does not directly meet the required outcome to choose among distinct business flows based on a validated account status.

C:

For a system whose objective is to reduce latency when separate read-only lookups have no ordering dependency, parallelizing independent operations shortens critical-path latency but is unsafe when one action requires another's output; this does not directly meet the required outcome to choose among distinct business flows based on a validated account status.

D:

For a system whose objective is to allow a model to request multiple independent data lookups while limiting cost and duration, an orchestration loop can inspect tool requests, execute authorized tools, return results, and terminate under explicit limits; this does not directly meet the required outcome to choose among distinct business flows based on a validated account status.

Question 156

The engineering change request is to use a multimodal request for document-plus-image reasoning. What is the most direct acceptance criterion?

  1. Validate required fields, types, parse success and rejected malformed payloads
  2. Compare critical-path latency and result consistency with serial execution
  3. Measure answer support and document freshness on questions with recently revised policies
  4. Score relationship questions on manually labeled image-and-text examples

Explanation

Correct Answer

D:

The accepted change is to use a multimodal request for document-plus-image reasoning; this evaluation checks whether it can reason about a photo and related user instructions together in one supported inference flow, since a multimodal model can interpret visual input alongside textual context; relying on OCR alone loses non-textual relationships.

Incorrect Answers

A:

This test concerns the ability to send model-produced classifications into a strict downstream API contract, for which schema-constrained generation helps enforce machine-readable fields and types; application-level validation is still needed for business rules; it does not establish whether the change to use a multimodal request for document-plus-image reasoning works for the current release.

B:

This test concerns the ability to reduce latency when separate read-only lookups have no ordering dependency, for which parallelizing independent operations shortens critical-path latency but is unsafe when one action requires another's output; it does not establish whether the change to use a multimodal request for document-plus-image reasoning works for the current release.

C:

This test concerns the ability to answer questions about changing internal procedures with source-backed responses, for which RAG retrieves relevant evidence at runtime and gives it to a generative model; model pretraining alone cannot guarantee current policy text; it does not establish whether the change to use a multimodal request for document-plus-image reasoning works for the current release.

Question 157

Stakeholders want a Foundry-based application able to obtain exact tax or pricing totals from an authoritative business calculation endpoint. Which choice implements that requirement?

  1. Use a multimodal request for document-plus-image reasoning
  2. Use function calling for external calculations
  3. Use chunk overlap for boundary-spanning facts
  4. Use grounded retrieval-augmented generation

Explanation

Correct Answer

B:

For the stated requirement to obtain exact tax or pricing totals from an authoritative business calculation endpoint, a tool call delegates deterministic computation to trusted code instead of relying on a model's arithmetic approximation; this is the mechanism that the system must implement.

Incorrect Answers

A:

For a system whose objective is to reason about a photo and related user instructions together in one supported inference flow, a multimodal model can interpret visual input alongside textual context; relying on OCR alone loses non-textual relationships; this does not directly meet the required outcome to obtain exact tax or pricing totals from an authoritative business calculation endpoint.

C:

For a system whose objective is to keep context that crosses page or section boundaries from being lost during indexing, chunk overlap can retain nearby context across splits, although excessive overlap inflates index size and duplicates retrieval; this does not directly meet the required outcome to obtain exact tax or pricing totals from an authoritative business calculation endpoint.

D:

For a system whose objective is to answer questions about changing internal procedures with source-backed responses, RAG retrieves relevant evidence at runtime and gives it to a generative model; model pretraining alone cannot guarantee current policy text; this does not directly meet the required outcome to obtain exact tax or pricing totals from an authoritative business calculation endpoint.

Question 158

The engineering change request is to use grounded retrieval-augmented generation. What is the most direct acceptance criterion?

  1. Measure answer support and document freshness on questions with recently revised policies
  2. Score relationship questions on manually labeled image-and-text examples
  3. Validate required fields, types, parse success and rejected malformed payloads
  4. Inspect request-target metadata and run an environment-isolation smoke test

Explanation

Correct Answer

A:

The accepted change is to use grounded retrieval-augmented generation; this evaluation checks whether it can answer questions about changing internal procedures with source-backed responses, since RAG retrieves relevant evidence at runtime and gives it to a generative model; model pretraining alone cannot guarantee current policy text.

Incorrect Answers

B:

This test concerns the ability to reason about a photo and related user instructions together in one supported inference flow, for which a multimodal model can interpret visual input alongside textual context; relying on OCR alone loses non-textual relationships; it does not establish whether the change to use grounded retrieval-augmented generation works for the current release.

C:

This test concerns the ability to send model-produced classifications into a strict downstream API contract, for which schema-constrained generation helps enforce machine-readable fields and types; application-level validation is still needed for business rules; it does not establish whether the change to use grounded retrieval-augmented generation works for the current release.

D:

This test concerns the ability to avoid routing production questions to a test model through hardcoded SDK settings, for which endpoint and deployment identifiers are environment-specific; runtime configuration keeps inference calls aligned to the project; it does not establish whether the change to use grounded retrieval-augmented generation works for the current release.

Question 159

A customer-facing AI service must consistently send model-produced classifications into a strict downstream API contract. Which architecture decision is most defensible?

  1. Use token-based credentials rather than embedded keys
  2. Use a multimodal request for document-plus-image reasoning
  3. Use function calling for external calculations
  4. Use constrained structured JSON output

Explanation

Correct Answer

D:

For the stated requirement to send model-produced classifications into a strict downstream API contract, schema-constrained generation helps enforce machine-readable fields and types; application-level validation is still needed for business rules; this is the mechanism that the system must implement.

Incorrect Answers

A:

For a system whose objective is to authenticate a server-side generative app through Entra where supported, a workload identity and appropriate RBAC reduce static credential exposure while preserving auditable authorization decisions; this does not directly meet the required outcome to send model-produced classifications into a strict downstream API contract.

B:

For a system whose objective is to reason about a photo and related user instructions together in one supported inference flow, a multimodal model can interpret visual input alongside textual context; relying on OCR alone loses non-textual relationships; this does not directly meet the required outcome to send model-produced classifications into a strict downstream API contract.

C:

For a system whose objective is to obtain exact tax or pricing totals from an authoritative business calculation endpoint, a tool call delegates deterministic computation to trusted code instead of relying on a model's arithmetic approximation; this does not directly meet the required outcome to send model-produced classifications into a strict downstream API contract.

Question 160

The support ticket describes this behavior: A model supplies a negative quantity to an order placement tool even though the schema expects an allowed positive amount. Which engineering change should come first?

  1. Reject invalid parameters with application-side validation before the business API call
  2. Propagate cancellation to the model request and close the stream on disconnect
  3. Configure a credential chain using the workload identity and least-privilege role grants
  4. Adjust chunk boundaries and limited overlap, then rebuild affected embeddings

Explanation

Correct Answer

A:

The fault signal is that a model supplies a negative quantity to an order placement tool even though the schema expects an allowed positive amount; the targeted fix works because tool schemas guide argument shape, but the application must check types, constraints, identity and authorization before execution, restoring the ability to prevent malformed model-generated parameters from reaching a write-capable API.

Incorrect Answers

B:

If the failing function were to show partial assistant output to a user while stopping work after the request is canceled, this change could be relevant because streaming delivers incremental output; cancellation and backpressure prevent orphaned inference work and stale interface content; the actual symptom is that a model supplies a negative quantity to an order placement tool even though the schema expects an allowed positive amount, which calls for a different repair.

C:

If the failing function were to authenticate a server-side generative app through Entra where supported, this change could be relevant because a workload identity and appropriate RBAC reduce static credential exposure while preserving auditable authorization decisions; the actual symptom is that a model supplies a negative quantity to an order placement tool even though the schema expects an allowed positive amount, which calls for a different repair.

D:

If the failing function were to keep context that crosses page or section boundaries from being lost during indexing, this change could be relevant because chunk overlap can retain nearby context across splits, although excessive overlap inflates index size and duplicates retrieval; the actual symptom is that a model supplies a negative quantity to an order placement tool even though the schema expects an allowed positive amount, which calls for a different repair.

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