Practice Exams:

Microsoft AI-103 Dumps

EXAM: Developing AI Apps and Agents on Azure
Verified By Experts
Premium Bundle
$49.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 22

Question 210

During a data protection inspection, a quality assurance lead must accept a change for private data telemetry. The issue to be addressed is: Model traces risk capturing customer health details in plaintext. Which concrete verification action demonstrates that the relevant behavior works?

  1. Enable full prompt-body logging in a broadly readable workspace by default
  2. Submit a known sensitive test string and inspect who can retrieve telemetry
  3. Redact or restrict sensitive content in trace instrumentation before storage
  4. Evaluate a fixed representative prompt set across both versions

Explanation

Correct Answer

B:

To verify this change, the industrial maintenance team should submit a known sensitive test string and inspect who can retrieve telemetry. Operators still need the implementation to redact or restrict sensitive content in trace instrumentation before storage and the preventive requirement to apply retention and role policies to all captured model inputs.

Incorrect Answers

A:

In the industrial maintenance team acceptance review, trying to enable full prompt-body logging in a broadly readable workspace by default does not test the exact private data telemetry condition where model traces risk capturing customer health details in plaintext; the relevant test is to submit a known sensitive test string and inspect who can retrieve telemetry. Acceptance should instead demonstrate that the service can submit a known sensitive test string and inspect who can retrieve telemetry, using the described failure as the test criterion.

C:

The action to redact or restrict sensitive content in trace instrumentation before storage can contribute to private data telemetry as a implementation step; here, model traces risk capturing customer health details in plaintext. In the industrial maintenance team acceptance review, trying to redact or restrict sensitive content in trace instrumentation before storage does not test the exact private data telemetry condition where model traces risk capturing customer health details in plaintext; the relevant test is to submit a known sensitive test string and inspect who can retrieve telemetry.

D:

For model output drift, it can be sensible to evaluate a fixed representative prompt set across both versions if an assistant becomes less accurate after a deployment or prompt change. Without the actual change to submit a known sensitive test string and inspect who can retrieve telemetry, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

Question 211

At a transport authority, the AI platform engineer is preparing a operational readiness review. The concrete requirement is: A regional service has intermittent errors but its health endpoint reports green. Which design action BEST addresses the stated need?

  1. Restrict retrieval connector data-plane permissions to authorized corpora
  2. Combine functional synthetic inference checks with service error and saturation metrics
  3. Depend solely on a single resource-state indicator to prove request success
  4. Run synthetic requests from each region and inspect response accuracy and delay

Explanation

Correct Answer

B:

The proposed implementation is to combine functional synthetic inference checks with service error and saturation metrics; this matches the operational requirement that a regional service has intermittent errors but its health endpoint reports green. The team can run synthetic requests from each region and inspect response accuracy and delay. The team should also alert on end-to-end success latency and failed token generation.

Incorrect Answers

A:

Choosing to restrict retrieval connector data-plane permissions to authorized corpora cannot directly replace the specified action to combine functional synthetic inference checks with service error and saturation metrics when a regional service has intermittent errors but its health endpoint reports green. The relevant corrective or preventive action here remains to combine functional synthetic inference checks with service error and saturation metrics, with its own validation criteria.

C:

Choosing to depend solely on a single resource-state indicator to prove request success cannot directly replace the specified action to combine functional synthetic inference checks with service error and saturation metrics when a regional service has intermittent errors but its health endpoint reports green. The actual implementation must combine functional synthetic inference checks with service error and saturation metrics, because the stated service condition is that a regional service has intermittent errors but its health endpoint reports green.

D:

The action to run synthetic requests from each region and inspect response accuracy and delay can contribute to deployment availability as a targeted acceptance check step; here, a regional service has intermittent errors but its health endpoint reports green. Choosing to run synthetic requests from each region and inspect response accuracy and delay cannot directly replace the specified action to combine functional synthetic inference checks with service error and saturation metrics when a regional service has intermittent errors but its health endpoint reports green.

Question 212

After a regional deployment at a large online retailer, the team observes this failure: A regional service has intermittent errors but its health endpoint reports green. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Run synthetic requests from each region and inspect response accuracy and delay
  2. Combine functional synthetic inference checks with service error and saturation metrics
  3. Enable full prompt-body logging in a broadly readable workspace by default
  4. Depend solely on a single resource-state indicator to prove request success

Explanation

Correct Answer

D:

This design mistake is to depend solely on a single resource-state indicator to prove request success; it conflicts with the actual deployment availability behavior needed when a regional service has intermittent errors but its health endpoint reports green. The large online retailer should instead combine functional synthetic inference checks with service error and saturation metrics, followed by a check to run synthetic requests from each region and inspect response accuracy and delay.

Incorrect Answers

A:

The alternative to run synthetic requests from each region and inspect response accuracy and delay is a targeted acceptance check step rather than the faulty assumption being diagnosed. This large online retailer failure-analysis alternative of run synthetic requests from each region and inspect response accuracy and delay does not diagnose the particular deployment availability defect where a regional service has intermittent errors but its health endpoint reports green; the keyed mistake is depend solely on a single resource-state indicator to prove request success.

B:

The alternative to combine functional synthetic inference checks with service error and saturation metrics is a implementation step rather than the faulty assumption being diagnosed. This large online retailer failure-analysis alternative of combine functional synthetic inference checks with service error and saturation metrics does not diagnose the particular deployment availability defect where a regional service has intermittent errors but its health endpoint reports green; the keyed mistake is depend solely on a single resource-state indicator to prove request success.

C:

This large online retailer failure-analysis alternative of enable full prompt-body logging in a broadly readable workspace by default does not diagnose the particular deployment availability defect where a regional service has intermittent errors but its health endpoint reports green; the keyed mistake is depend solely on a single resource-state indicator to prove request success. The alternative may be useful in its own context, but it does not directly supply depend solely on a single resource-state indicator to prove request success for the stated decision.

Question 213

The public library network is responding to a architecture migration. A product support lead discovers this issue: A regional service has intermittent errors but its health endpoint reports green. What corrective step should the team take to restore the intended functionality most directly?

  1. Depend solely on a single resource-state indicator to prove request success
  2. Route incidents to the failing dependency based on correlated traces
  3. Run synthetic requests from each region and inspect response accuracy and delay
  4. Mitigate capacity or dependency failures identified in the request trace

Explanation

Correct Answer

D:

The direct repair is to mitigate capacity or dependency failures identified in the request trace. Once the condition is corrected, the public library network should run synthetic requests from each region and inspect response accuracy and delay. To reduce recurrence, it should also alert on end-to-end success latency and failed token generation; those steps provide different recovery and prevention functions.

Incorrect Answers

A:

During public library network remediation, the proposed step to depend solely on a single resource-state indicator to prove request success is not the direct restoration action for deployment availability when a regional service has intermittent errors but its health endpoint reports green; the keyed repair is mitigate capacity or dependency failures identified in the request trace. After remediation the team must run synthetic requests from each region and inspect response accuracy and delay; the action in this option does not itself establish that target recovery outcome.

B:

This approach relates to provider health monitoring: an app needs to distinguish model failures from search and speech outages. If the team chose it instead, the targeted check to run synthetic requests from each region and inspect response accuracy and delay could still fail because the underlying required action is to mitigate capacity or dependency failures identified in the request trace.

C:

The action to run synthetic requests from each region and inspect response accuracy and delay can contribute to deployment availability as a targeted acceptance check step; here, a regional service has intermittent errors but its health endpoint reports green. During public library network remediation, the proposed step to run synthetic requests from each region and inspect response accuracy and delay is not the direct restoration action for deployment availability when a regional service has intermittent errors but its health endpoint reports green; the keyed repair is mitigate capacity or dependency failures identified in the request trace.

Question 214

A data engineer at a software consultancy wants the following issue to be prevented in later deployments: A regional service has intermittent errors but its health endpoint reports green. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Depend solely on a single resource-state indicator to prove request success
  2. Mitigate capacity or dependency failures identified in the request trace
  3. Tag every trace with deployment and prompt template identifiers
  4. Alert on end-to-end success latency and failed token generation

Explanation

Correct Answer

D:

Standardizing the boundary to alert on end-to-end success latency and failed token generation is preventive because the same deployment availability failure could otherwise return: a regional service has intermittent errors but its health endpoint reports green. For this software consultancy, the service should also combine functional synthetic inference checks with service error and saturation metrics, and tests should run synthetic requests from each region and inspect response accuracy and delay so that the ongoing requirement is measurable.

Incorrect Answers

A:

For prevention at the software consultancy, the measure to depend solely on a single resource-state indicator to prove request success is weaker or aimed at another issue than the deployment availability guardrail to alert on end-to-end success latency and failed token generation needed when a regional service has intermittent errors but its health endpoint reports green. The lasting guardrail for this specific threat is to alert on end-to-end success latency and failed token generation, not simply to complete an adjacent service task.

B:

The action to mitigate capacity or dependency failures identified in the request trace can contribute to deployment availability as a corrective remediation step; here, a regional service has intermittent errors but its health endpoint reports green. For prevention at the software consultancy, the measure to mitigate capacity or dependency failures identified in the request trace is weaker or aimed at another issue than the deployment availability guardrail to alert on end-to-end success latency and failed token generation needed when a regional service has intermittent errors but its health endpoint reports green.

C:

The action to tag every trace with deployment and prompt template identifiers addresses experiment reproducibility rather than the decisive deployment availability condition. The software consultancy should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to alert on end-to-end success latency and failed token generation.

Question 215

During a regional deployment, a service administrator must accept a change for deployment availability. The issue to be addressed is: A regional service has intermittent errors but its health endpoint reports green. Which concrete verification action demonstrates that the relevant behavior works?

  1. Combine functional synthetic inference checks with service error and saturation metrics
  2. Depend solely on a single resource-state indicator to prove request success
  3. Modify a test source and verify its latest passage becomes retrievable
  4. Run synthetic requests from each region and inspect response accuracy and delay

Explanation

Correct Answer

D:

The targeted acceptance test is to run synthetic requests from each region and inspect response accuracy and delay. The design should first combine functional synthetic inference checks with service error and saturation metrics. A robust regional hospital deployment also needs to alert on end-to-end success latency and failed token generation; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

A:

The action to combine functional synthetic inference checks with service error and saturation metrics can contribute to deployment availability as a implementation step; here, a regional service has intermittent errors but its health endpoint reports green. In the regional hospital acceptance review, trying to combine functional synthetic inference checks with service error and saturation metrics does not test the exact deployment availability condition where a regional service has intermittent errors but its health endpoint reports green; the relevant test is to run synthetic requests from each region and inspect response accuracy and delay.

B:

In the regional hospital acceptance review, trying to depend solely on a single resource-state indicator to prove request success does not test the exact deployment availability condition where a regional service has intermittent errors but its health endpoint reports green; the relevant test is to run synthetic requests from each region and inspect response accuracy and delay. Acceptance should instead demonstrate that the service can run synthetic requests from each region and inspect response accuracy and delay, using the described failure as the test criterion.

C:

For search index health, it can be sensible to modify a test source and verify its latest passage becomes retrievable if a knowledge assistant gives old policy answers after nightly ingestion. Without the actual change to run synthetic requests from each region and inspect response accuracy and delay, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

Question 216

At a industrial maintenance team, the cloud developer is preparing a reliability exercise. The concrete requirement is: An inference app succeeds locally using developer credentials but fails under deployment. Which design action BEST addresses the stated need?

  1. Grant the workload identity the required role and validate token audience
  2. Invoke the app under its actual service principal rather than a local user
  3. Assume personal developer sign-in roles transfer to deployed managed identities
  4. Migrate to managed identity where supported and rotate residual credentials

Explanation

Correct Answer

A:

Choosing to grant the workload identity the required role and validate token audience directly targets the described access token authorization gap: an inference app succeeds locally using developer credentials but fails under deployment. As a lasting boundary, they also need to use automated role permission tests for runtime identities, because an implementation by itself does not prove safe operation.

Incorrect Answers

B:

The action to invoke the app under its actual service principal rather than a local user can contribute to access token authorization as a targeted acceptance check step; here, an inference app succeeds locally using developer credentials but fails under deployment. Choosing to invoke the app under its actual service principal rather than a local user cannot directly replace the specified action to grant the workload identity the required role and validate token audience when an inference app succeeds locally using developer credentials but fails under deployment.

C:

Choosing to assume personal developer sign-in roles transfer to deployed managed identities cannot directly replace the specified action to grant the workload identity the required role and validate token audience when an inference app succeeds locally using developer credentials but fails under deployment. The actual implementation must grant the workload identity the required role and validate token audience, because the stated service condition is that an inference app succeeds locally using developer credentials but fails under deployment.

D:

Choosing to migrate to managed identity where supported and rotate residual credentials would direct effort toward key lifecycle for service apis, whose characteristic issue is that an older AI integration still uses a long-lived key shared by several developers. The relevant corrective or preventive action here remains to grant the workload identity the required role and validate token audience, with its own validation criteria.

Question 217

A site reliability engineer at a research university is conducting a source update concerning access token authorization. The observed problem is: An inference app succeeds locally using developer credentials but fails under deployment. Which TWO actions together represent the needed implementation AND a durable preventive safeguard? (Select TWO answers.)

  1. Grant the workload identity the required role and validate token audience
  2. Use automated role permission tests for runtime identities
  3. Restrict retrieval connector data-plane permissions to authorized corpora
  4. Fix deployed identity bindings and regenerate correctly scoped access tokens
  5. Invoke the app under its actual service principal rather than a local user

Explanation

Correct Answers

A:

The proposed implementation is to grant the workload identity the required role and validate token audience; this matches the operational requirement that an inference app succeeds locally using developer credentials but fails under deployment. The team can invoke the app under its actual service principal rather than a local user. The team should also use automated role permission tests for runtime identities.

B:

The sustained prevention measure is to use automated role permission tests for runtime identities. The implementation can still use the separate action to grant the workload identity the required role and validate token audience, but a guardrail constrains future changes. The research university can measure whether the rule works by arranging to invoke the app under its actual service principal rather than a local user.

Incorrect Answers

C:

This approach relates to data-connector permissions: an agent can read documents from projects it was never assigned. If the team chose it instead, the targeted check to invoke the app under its actual service principal rather than a local user could still fail because the underlying required action is to grant the workload identity the required role and validate token audience.

D:

The alternative to fix deployed identity bindings and regenerate correctly scoped access tokens is a corrective remediation step rather than the faulty assumption being diagnosed. This research university failure-analysis alternative of fix deployed identity bindings and regenerate correctly scoped access tokens does not diagnose the particular access token authorization defect where an inference app succeeds locally using developer credentials but fails under deployment; the keyed mistake is grant the workload identity the required role and validate token audience.

E:

The alternative to invoke the app under its actual service principal rather than a local user is a targeted acceptance check step rather than the faulty assumption being diagnosed. This research university failure-analysis alternative of invoke the app under its actual service principal rather than a local user does not diagnose the particular access token authorization defect where an inference app succeeds locally using developer credentials but fails under deployment; the keyed mistake is grant the workload identity the required role and validate token audience.

Question 218

The education technology company is responding to a cost review. A security engineer discovers this issue: An inference app succeeds locally using developer credentials but fails under deployment. What corrective step should the team take to restore the intended functionality most directly?

  1. Fix deployed identity bindings and regenerate correctly scoped access tokens
  2. Invoke the app under its actual service principal rather than a local user
  3. Optimize the slow tool or apply bounded timeouts and safe fallbacks
  4. Assume personal developer sign-in roles transfer to deployed managed identities

Explanation

Correct Answer

A:

The appropriate remediation is to fix deployed identity bindings and regenerate correctly scoped access tokens, because the failure concerns access token authorization: an inference app succeeds locally using developer credentials but fails under deployment. A meaningful verification should invoke the app under its actual service principal rather than a local user, and the durable operating rule should use automated role permission tests for runtime identities after the immediate fix.

Incorrect Answers

B:

The action to invoke the app under its actual service principal rather than a local user can contribute to access token authorization as a targeted acceptance check step; here, an inference app succeeds locally using developer credentials but fails under deployment. During education technology company remediation, the proposed step to invoke the app under its actual service principal rather than a local user is not the direct restoration action for access token authorization when an inference app succeeds locally using developer credentials but fails under deployment; the keyed repair is fix deployed identity bindings and regenerate correctly scoped access tokens.

C:

This approach relates to agent tool latency: an otherwise fast chatbot responds slowly after retrieving external data. If the team chose it instead, the targeted check to invoke the app under its actual service principal rather than a local user could still fail because the underlying required action is to fix deployed identity bindings and regenerate correctly scoped access tokens.

D:

During education technology company remediation, the proposed step to assume personal developer sign-in roles transfer to deployed managed identities is not the direct restoration action for access token authorization when an inference app succeeds locally using developer credentials but fails under deployment; the keyed repair is fix deployed identity bindings and regenerate correctly scoped access tokens. After remediation the team must invoke the app under its actual service principal rather than a local user; the action in this option does not itself establish that target recovery outcome.

Question 219

A agent designer at a pharmaceutical research group wants the following issue to be prevented in later deployments: An inference app succeeds locally using developer credentials but fails under deployment. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Alert on end-to-end success latency and failed token generation
  2. Assume personal developer sign-in roles transfer to deployed managed identities
  3. Use automated role permission tests for runtime identities
  4. Fix deployed identity bindings and regenerate correctly scoped access tokens

Explanation

Correct Answer

C:

The sustained prevention measure is to use automated role permission tests for runtime identities. The implementation can still use the separate action to grant the workload identity the required role and validate token audience, but a guardrail constrains future changes. The pharmaceutical research group can measure whether the rule works by arranging to invoke the app under its actual service principal rather than a local user.

Incorrect Answers

A:

The action to alert on end-to-end success latency and failed token generation addresses deployment availability rather than the decisive access token authorization condition. The pharmaceutical research group should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to use automated role permission tests for runtime identities.

B:

For prevention at the pharmaceutical research group, the measure to assume personal developer sign-in roles transfer to deployed managed identities is weaker or aimed at another issue than the access token authorization guardrail to use automated role permission tests for runtime identities needed when an inference app succeeds locally using developer credentials but fails under deployment. The lasting guardrail for this specific threat is to use automated role permission tests for runtime identities, not simply to complete an adjacent service task.

D:

The action to fix deployed identity bindings and regenerate correctly scoped access tokens can contribute to access token authorization as a corrective remediation step; here, an inference app succeeds locally using developer credentials but fails under deployment. For prevention at the pharmaceutical research group, the measure to fix deployed identity bindings and regenerate correctly scoped access tokens is weaker or aimed at another issue than the access token authorization guardrail to use automated role permission tests for runtime identities needed when an inference app succeeds locally using developer credentials but fails under deployment.

Question 220

During a source update, a responsible-AI reviewer must accept a change for access token authorization. The issue to be addressed is: An inference app succeeds locally using developer credentials but fails under deployment. Which concrete verification action demonstrates that the relevant behavior works?

  1. Use automated role permission tests for runtime identities
  2. Assume personal developer sign-in roles transfer to deployed managed identities
  3. Ask an agent to retrieve forbidden document IDs and confirm access is blocked
  4. Invoke the app under its actual service principal rather than a local user
  5. Grant the workload identity the required role and validate token audience

Explanation

Correct Answer

D:

To verify this change, the travel support provider should invoke the app under its actual service principal rather than a local user. Operators still need the implementation to grant the workload identity the required role and validate token audience and the preventive requirement to use automated role permission tests for runtime identities.

Incorrect Answers

A:

The action to use automated role permission tests for runtime identities can contribute to access token authorization as a ongoing preventive rule step; here, an inference app succeeds locally using developer credentials but fails under deployment. In the travel support provider acceptance review, trying to use automated role permission tests for runtime identities does not test the exact access token authorization condition where an inference app succeeds locally using developer credentials but fails under deployment; the relevant test is to invoke the app under its actual service principal rather than a local user.

B:

In the travel support provider acceptance review, trying to assume personal developer sign-in roles transfer to deployed managed identities does not test the exact access token authorization condition where an inference app succeeds locally using developer credentials but fails under deployment; the relevant test is to invoke the app under its actual service principal rather than a local user. Acceptance should instead demonstrate that the service can invoke the app under its actual service principal rather than a local user, using the described failure as the test criterion.

C:

Choosing to ask an agent to retrieve forbidden document IDs and confirm access is blocked would direct effort toward data-connector permissions, whose characteristic issue is that an agent can read documents from projects it was never assigned. The relevant corrective or preventive action here remains to invoke the app under its actual service principal rather than a local user, with its own validation criteria.

E:

The action to grant the workload identity the required role and validate token audience can contribute to access token authorization as a implementation step; here, an inference app succeeds locally using developer credentials but fails under deployment. In the travel support provider acceptance review, trying to grant the workload identity the required role and validate token audience does not test the exact access token authorization condition where an inference app succeeds locally using developer credentials but fails under deployment; the relevant test is to invoke the app under its actual service principal rather than a local user.

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