Snowflake SnowPro Specialty Gen AI GES-C01 Practice Test 41
Question 400
At an e-commerce marketplace, a Snowflake engineering team is undertaking a returns reason classification effort. An application role can query sales tables but receives a permission denial for an AI inference function. The change was trialed on a small dataset; billing and privacy queues need distinct ownership, while some tickets mention several products. Which acceptance condition BEST demonstrates that the intervention worked?
- Authorized calls succeed while nonapproved roles remain denied.
- No group has an unacceptable unreviewed error concentration.
- No claim is denied solely on unreviewed generated text.
- The supported replacement meets functional and privacy requirements.
Explanation
Correct Answer
A:This is the observable acceptance condition that would show the intended fix addressed ordinary select access on user tables does not automatically grant every managed ai capability. It is stronger than activity counts because it tests the user-facing effect. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
B:This could demonstrate success for ai evaluation bias, but it would not establish resolution of the current issue: an application role can query sales tables but receives a permission denial for an ai inference function. Validate authorized calls succeed while nonapproved roles remain denied. instead. These author-intended interpretations require independent technical review before operational use.
C:This could demonstrate success for human-in-loop decisions, but it would not establish resolution of the current issue: an application role can query sales tables but receives a permission denial for an ai inference function. Validate authorized calls succeed while nonapproved roles remain denied. instead. These author-intended interpretations require independent technical review before operational use.
D:This could demonstrate success for vendor feature deprecation, but it would not establish resolution of the current issue: an application role can query sales tables but receives a permission denial for an ai inference function. Validate authorized calls succeed while nonapproved roles remain denied. instead. These author-intended interpretations require independent technical review before operational use.
Question 401
At an insurance claims processor, a Snowflake engineering team is undertaking a form intake modernization. An application role can query sales tables but receives a permission denial for an AI inference function. An auditor will inspect the control design; the new system cannot automatically deny claims, and manual reviewers handle ambiguous scanned content. Which ongoing control MOST directly reduces the original risk?
- Document escalation rights and retain decision provenance.
- Make fairness testing part of model release gates.
- Separate informational output from executable action permissions.
- Review Cortex and data grants independently under least privilege.
Explanation
Correct Answer
D:This retains a specific safeguard against recurrence of ordinary select access on user tables does not automatically grant every managed ai capability. It complements the immediate intervention, check the appropriate snowflake cortex database-role grants and assign only the function privileges it needs., and allows the team to verify the intended result. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This is a sensible safeguard for human-in-loop decisions, but it does not target the concrete exposure in this scenario. The current failure is ordinary select access on user tables does not automatically grant every managed ai capability. These author-intended interpretations require independent technical review before operational use.
B:This is a sensible safeguard for ai evaluation bias, but it does not target the concrete exposure in this scenario. The current failure is ordinary select access on user tables does not automatically grant every managed ai capability. These author-intended interpretations require independent technical review before operational use.
C:This is a sensible safeguard for llm output safety, but it does not target the concrete exposure in this scenario. The current failure is ordinary select access on user tables does not automatically grant every managed ai capability. These author-intended interpretations require independent technical review before operational use.
Question 402
At a consumer travel platform, a Snowflake engineering team is undertaking a booking assistant redesign. An application role can query sales tables but receives a permission denial for an AI inference function. Budget and operational risk are both material; peak demand is highly seasonal; a bad response can trigger a financial adjustment. Which decision MOST appropriately balances the requested outcome with production constraints?
- Account for ongoing serving compute and explicitly suspend or retire unneeded services. Postpone the critical evidence checks until after a broad release.
- Restrict the system to decision support and require authorized human adjudication of consequential outcomes. Postpone the critical evidence checks until after a broad release.
- Apply a supported account-level model restriction and verify both allowed and denied model calls. Postpone the critical evidence checks until after a broad release.
- Check the appropriate Snowflake Cortex database-role grants and assign only the function privileges it needs. Then review Cortex and data grants independently under least privilege.
Explanation
Correct Answer
D:This combines a corrective action with a bounded production safeguard, addressing an application role can query sales tables but receives a permission denial for an ai inference function. Prove the result by checking that authorized calls succeed while nonapproved roles remain denied. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This proposal begins with an intervention for cortex search operating costs and postpones evidence that is essential for the current case. Before rollout, the team should check the appropriate snowflake cortex database-role grants and assign only the function privileges it needs. These author-intended interpretations require independent technical review before operational use.
B:This proposal begins with an intervention for human-in-loop decisions and postpones evidence that is essential for the current case. Before rollout, the team should check the appropriate snowflake cortex database-role grants and assign only the function privileges it needs. These author-intended interpretations require independent technical review before operational use.
C:This proposal begins with an intervention for model allowlist policy and postpones evidence that is essential for the current case. Before rollout, the team should check the appropriate snowflake cortex database-role grants and assign only the function privileges it needs. These author-intended interpretations require independent technical review before operational use.
Question 403
At a packaging manufacturer, a Snowflake engineering team is undertaking a product support search interface. Security approves two language models, but a developer's new app can still select other available models. Before implementation, knowledge articles update every workday; operators need reliable revision dates. Which first architectural or operating step MOST directly addresses this case?
- Record source revision, model identifier, prompt version, and time of each decision.
- Apply a supported account-level model restriction and verify both allowed and denied model calls.
- Treat retrieved content as untrusted evidence and test against instruction override attempts.
- Confirm feature regional availability and approved processing routes before enabling inference.
Explanation
Correct Answer
B:The decision addresses the central constraint: security approves two language models, but a developer's new app can still select other available models. The most direct next step is to apply a supported account-level model restriction and verify both allowed and denied model calls. Before changing production, review both: inspect allowlist parameter and enforcement behavior and test an unauthorized model request.. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This can be appropriate when a reviewer cannot reproduce why a classified record changed its label after a model refresh. It addresses audit trail and model changes, whereas the present limitation is better explained by application convention is not the same as enforced model governance. The target outcome here is that only the approved model identifiers are usable by the intended roles. These author-intended interpretations require independent technical review before operational use.
C:This can be appropriate when a retrieved product wiki page contains instructions to ignore an internal assistant's data-access policy. It addresses prompt injection threat modeling, whereas the present limitation is better explained by application convention is not the same as enforced model governance. The target outcome here is that only the approved model identifiers are usable by the intended roles. These author-intended interpretations require independent technical review before operational use.
D:This can be appropriate when a regulated tenant must keep processing inside a designated geography while a requested model is only available elsewhere. It addresses data residency and cross-region, whereas the present limitation is better explained by application convention is not the same as enforced model governance. The target outcome here is that only the approved model identifiers are usable by the intended roles. These author-intended interpretations require independent technical review before operational use.
Question 404
At a semiconductor equipment maker, a Snowflake engineering team is undertaking a component troubleshooting assistant. Security approves two language models, but a developer's new app can still select other available models. The proposed fix is still hypothetical; document titles are duplicated across product revisions, and engineers need verified part-specific guidance. Which TWO checks provide the most decisive evidence before approving a change?
- Compare each citation and metric to executed query output
- Inspect downstream workflow authority
- Inspect allowlist parameter and enforcement behavior
- Test an unauthorized model request.
- Review service usage history and active status
Explanation
Correct Answers
C:This observation directly tests the case mechanism: Application convention is not the same as enforced model governance. In the described setting, it helps determine whether apply a supported account-level model restriction and verify both allowed and denied model calls. is justified. These author-intended interpretations require independent technical review before operational use.
D:This complementary observation tests a different part of the same hypothesis and should be read together with the first check before authorizing change. Expected success would be: only the approved model identifiers are usable by the intended roles. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:Checking this might help diagnose model output hallucination, where a generated audit summary cites a warehouse table that was never queried. It is not one of the two most direct checks for the present case because the leading mechanism is application convention is not the same as enforced model governance. These author-intended interpretations require independent technical review before operational use.
B:Checking this might help diagnose human-in-loop decisions, where a generative model proposes denying insurance claims without individual review. It is not one of the two most direct checks for the present case because the leading mechanism is application convention is not the same as enforced model governance. These author-intended interpretations require independent technical review before operational use.
E:Checking this might help diagnose cortex search operating costs, where an indexed service remains available after a pilot stops querying it and costs continue. It is not one of the two most direct checks for the present case because the leading mechanism is application convention is not the same as enforced model governance. These author-intended interpretations require independent technical review before operational use.
Question 405
At a health-benefits administrator, a Snowflake engineering team is undertaking a contact-center quality pilot. Security approves two language models, but a developer's new app can still select other available models. Diagnosis is disputed because the helpdesk processes sensitive documents; responses must avoid unsupported policy assertions. Which underlying failure mechanism BEST explains the observed limitation?
- Ordinary SELECT access on user tables does not automatically grant every managed AI capability.
- Missing version lineage prevents reliable fault attribution.
- Application convention is not the same as enforced model governance.
- Unbounded retries can worsen overload and inflate spend.
Explanation
Correct Answer
C:This is the causal hypothesis consistent with the observed limitation: security approves two language models, but a developer's new app can still select other available models. Confirm it through both inspect allowlist parameter and enforcement behavior and test an unauthorized model request. before changing the service. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:That mechanism could explain a different case about cortex role grant, but does not most directly account for this specific symptom: security approves two language models, but a developer's new app can still select other available models. Here, the evidence to collect is inspect allowlist parameter and enforcement behavior These author-intended interpretations require independent technical review before operational use.
B:That mechanism could explain a different case about audit trail and model changes, but does not most directly account for this specific symptom: security approves two language models, but a developer's new app can still select other available models. Here, the evidence to collect is inspect allowlist parameter and enforcement behavior These author-intended interpretations require independent technical review before operational use.
D:That mechanism could explain a different case about resource quota throttling, but does not most directly account for this specific symptom: security approves two language models, but a developer's new app can still select other available models. Here, the evidence to collect is inspect allowlist parameter and enforcement behavior These author-intended interpretations require independent technical review before operational use.
Question 406
At a hospital-equipment supplier, a Snowflake engineering team is undertaking a document processing migration. Security approves two language models, but a developer's new app can still select other available models. The change was trialed on a small dataset; an incorrect number could delay critical replenishment, while thousands of older scanned PDFs have variable layouts. Which acceptance condition BEST demonstrates that the intervention worked?
- Clients cannot retrieve shared service secrets from the application.
- Only the approved model identifiers are usable by the intended roles.
- Disallowed routes fail closed and approved routes are logged.
- The assistant follows trusted policy while extracting only relevant facts.
Explanation
Correct Answer
B:This is the observable acceptance condition that would show the intended fix addressed application convention is not the same as enforced model governance. It is stronger than activity counts because it tests the user-facing effect. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This could demonstrate success for security credential handling, but it would not establish resolution of the current issue: security approves two language models, but a developer's new app can still select other available models. Validate only the approved model identifiers are usable by the intended roles. instead. These author-intended interpretations require independent technical review before operational use.
C:This could demonstrate success for data residency and cross-region, but it would not establish resolution of the current issue: security approves two language models, but a developer's new app can still select other available models. Validate only the approved model identifiers are usable by the intended roles. instead. These author-intended interpretations require independent technical review before operational use.
D:This could demonstrate success for prompt injection threat modeling, but it would not establish resolution of the current issue: security approves two language models, but a developer's new app can still select other available models. Validate only the approved model identifiers are usable by the intended roles. instead. These author-intended interpretations require independent technical review before operational use.
Question 407
At a regional airline services firm, a Snowflake engineering team is undertaking an incident summary assistant. Security approves two language models, but a developer's new app can still select other available models. An auditor will inspect the control design; network outages create intermittent retries, and audit leads demand source-grounded conclusions. Which ongoing control MOST directly reduces the original risk?
- Enforce secret scanning and lifecycle rotation.
- Require documented quality floors before cost optimization.
- Maintain model supply-chain approvals.
- Integrate model allowlist changes into change control.
Explanation
Correct Answer
D:This retains a specific safeguard against recurrence of application convention is not the same as enforced model governance. It complements the immediate intervention, apply a supported account-level model restriction and verify both allowed and denied model calls., and allows the team to verify the intended result. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This is a sensible safeguard for security credential handling, but it does not target the concrete exposure in this scenario. The current failure is application convention is not the same as enforced model governance. These author-intended interpretations require independent technical review before operational use.
B:This is a sensible safeguard for inference cost versus quality, but it does not target the concrete exposure in this scenario. The current failure is application convention is not the same as enforced model governance. These author-intended interpretations require independent technical review before operational use.
C:This is a sensible safeguard for third-party model provenance, but it does not target the concrete exposure in this scenario. The current failure is application convention is not the same as enforced model governance. These author-intended interpretations require independent technical review before operational use.
Question 408
At a public transit agency, a Snowflake engineering team is undertaking a fleet operations question-answering tool. Security approves two language models, but a developer's new app can still select other available models. Budget and operational risk are both material; the data includes safety incident narratives; changes require accountable signoff. Which decision MOST appropriately balances the requested outcome with production constraints?
- Record source revision, model identifier, prompt version, and time of each decision. Postpone the critical evidence checks until after a broad release.
- Confirm feature regional availability and approved processing routes before enabling inference. Postpone the critical evidence checks until after a broad release.
- Treat retrieved content as untrusted evidence and test against instruction override attempts. Postpone the critical evidence checks until after a broad release.
- Apply a supported account-level model restriction and verify both allowed and denied model calls. Then integrate model allowlist changes into change control.
Explanation
Correct Answer
D:This combines a corrective action with a bounded production safeguard, addressing security approves two language models, but a developer's new app can still select other available models. Prove the result by checking that only the approved model identifiers are usable by the intended roles. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This proposal begins with an intervention for audit trail and model changes and postpones evidence that is essential for the current case. Before rollout, the team should apply a supported account-level model restriction and verify both allowed and denied model calls. These author-intended interpretations require independent technical review before operational use.
B:This proposal begins with an intervention for data residency and cross-region and postpones evidence that is essential for the current case. Before rollout, the team should apply a supported account-level model restriction and verify both allowed and denied model calls. These author-intended interpretations require independent technical review before operational use.
C:This proposal begins with an intervention for prompt injection threat modeling and postpones evidence that is essential for the current case. Before rollout, the team should apply a supported account-level model restriction and verify both allowed and denied model calls. These author-intended interpretations require independent technical review before operational use.
Question 409
At a regional logistics provider, a Snowflake engineering team is undertaking a warehouse dispatch modernization. An analysis app includes unmasked customer identifiers in an LLM prompt unnecessarily. Before implementation, legacy barcode scanners cannot be replaced this quarter; customer-facing delays must remain measurable. Which first architectural or operating step MOST directly addresses this case?
- Treat retrieved content as untrusted evidence and test against instruction override attempts.
- Check the appropriate Snowflake Cortex database-role grants and assign only the function privileges it needs.
- Minimize and mask identifying fields before any generative AI invocation.
- Disable the affected release, restore a reviewed prompt/model configuration, and audit impacted requests.
Explanation
Correct Answer
C:The decision addresses the central constraint: an analysis app includes unmasked customer identifiers in an llm prompt unnecessarily. The most direct next step is to minimize and mask identifying fields before any generative ai invocation. Before changing production, review both: inspect actual request payloads in a safe test and verify masking policy effects under the application role.. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
A:This can be appropriate when a retrieved product wiki page contains instructions to ignore an internal assistant's data-access policy. It addresses prompt injection threat modeling, whereas the present limitation is better explained by warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. The target outcome here is that the request includes only fields justified by the approved task. These author-intended interpretations require independent technical review before operational use.
B:This can be appropriate when an application role can query sales tables but receives a permission denial for an ai inference function. It addresses cortex role grant, whereas the present limitation is better explained by warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. The target outcome here is that the request includes only fields justified by the approved task. These author-intended interpretations require independent technical review before operational use.
D:This can be appropriate when a prompt update causes increased leakage of private snippets in generated answers. It addresses ai incident rollback, whereas the present limitation is better explained by warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. The target outcome here is that the request includes only fields justified by the approved task. These author-intended interpretations require independent technical review before operational use.
Question 410
At a municipal utilities contractor, a Snowflake engineering team is undertaking a vendor compliance portal. An analysis app includes unmasked customer identifiers in an LLM prompt unnecessarily. The proposed fix is still hypothetical; document authors sometimes embed arbitrary instructions in descriptions, and audit requirements prohibit silent data leakage. Which TWO checks provide the most decisive evidence before approving a change?
- Inspect actual request payloads in a safe test
- Inspect trace fields and retention settings
- Compare token usage records to business job IDs
- Verify masking policy effects under the application role.
- Compare account region, cross-region parameters, and compliance policy
Explanation
Correct Answers
A:This observation directly tests the case mechanism: Warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. In the described setting, it helps determine whether minimize and mask identifying fields before any generative ai invocation. is justified. These author-intended interpretations require independent technical review before operational use.
D:This complementary observation tests a different part of the same hypothesis and should be read together with the first check before authorizing change. Expected success would be: the request includes only fields justified by the approved task. These author-intended interpretations require independent technical review before operational use.
Incorrect Answers
B:Checking this might help diagnose sensitive output logging, where a service traces entire user prompts and model outputs, accidentally retaining personal records for years. It is not one of the two most direct checks for the present case because the leading mechanism is warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. These author-intended interpretations require independent technical review before operational use.
C:Checking this might help diagnose token spend attribution, where a company sees rising ai credits but cannot identify which product line caused the spike. It is not one of the two most direct checks for the present case because the leading mechanism is warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. These author-intended interpretations require independent technical review before operational use.
E:Checking this might help diagnose data residency and cross-region, where a regulated tenant must keep processing inside a designated geography while a requested model is only available elsewhere. It is not one of the two most direct checks for the present case because the leading mechanism is warehouse masking on one output path does not guarantee all prompt-building logic excludes raw values. These author-intended interpretations require independent technical review before operational use.