Topic 21 Practice Test 2 covers AI and Machine Learning in Network Operations for Cisco Certified Network Associate 200-301 CCNA and maps to objectives 6.4. For broader exam preparation, review the Cisco CCNA 200-301 Exam Dumps. Every option includes focused technical reasoning explaining both the networking concept and its fit to the scenario.
Question 1
A design review for a global NOC predicting future capacity pressure focuses on this use case: forecasting capacity pressure, failure risk, or performance degradation before the event occurs. Which mechanism should the engineer use? Choose ONE.
- Generative AI
- AI-driven baselining
- AI-assisted issue correlation and root-cause guidance
- Predictive AI
Correct Answer: D
Correct Answer
Answer D is correct because The evidence at a global NOC predicting future capacity pressure points to Predictive AI. It uses learned patterns in historical and current data to estimate likely future conditions or outcomes. That capability supports the requirement to estimate a likely future network condition from learned patterns.
Incorrect Answers
Answer A is incorrect because Generative AI creates new content such as explanations, summaries, suggested commands, or troubleshooting narratives from prompts and context. That can be valid elsewhere, but a global NOC predicting future capacity pressure needs to estimate a likely future network condition from learned patterns. Predictive AI matches that objective.
Answer B is incorrect because For a global NOC predicting future capacity pressure, AI-driven baselining solves the wrong problem. It uses observed network behavior to establish what normal performance looks like for a specific environment. The scenario needs to estimate a likely future network condition from learned patterns, which points to Predictive AI.
Answer C is incorrect because AI-assisted issue correlation and root-cause guidance correlates multiple signals and known relationships so operators can focus on the most likely underlying cause. At a global NOC predicting future capacity pressure, it does not provide the requirement to estimate a likely future network condition from learned patterns. Predictive AI does.
Question 2
A design review for a support assistant generating a troubleshooting narrative focuses on this use case: turning a set of alerts and telemetry into a draft incident summary or proposed remediation explanation. Which technology or concept should be selected? Choose ONE.
- Machine-learning trend analysis
- Model training on representative historical data
- Generative AI
- Machine-learning anomaly detection
Correct Answer: C
Correct Answer
Answer C is correct because For a support assistant generating a troubleshooting narrative, the requirement is to create new text or suggested content from operational context. Generative AI provides that function. It matches the evidence. Predictive AI estimates likely outcomes instead of producing a new narrative or configuration draft.
Incorrect Answers
Answer D is incorrect because For a support assistant generating a troubleshooting narrative, Machine-learning anomaly detection solves the wrong problem. It learns or models expected behavior and identifies observations that significantly depart from the normal pattern. The scenario needs to create new text or suggested content from operational context, which points to Generative AI.
Answer A is incorrect because Using Machine-learning trend analysis, the design finds recurring or directional patterns in telemetry over time to support proactive operations. For a support assistant generating a troubleshooting narrative, the missing function is to create new text or suggested content from operational context; Generative AI supplies it.
Answer B is incorrect because Using Model training on representative historical data, the design adjusts model parameters from examples so the system can learn useful patterns before operational use. For a support assistant generating a troubleshooting narrative, the missing function is to create new text or suggested content from operational context; Generative AI supplies it.
Question 3
A design review for a campus assurance engine detecting behavior outside normal patterns focuses on this use case: surfacing an unusual latency, client-failure, or traffic pattern without relying only on one fixed threshold. Which mechanism should the engineer use? Choose ONE.
- AI-assisted issue correlation and root-cause guidance
- AI-driven baselining
- Machine-learning anomaly detection
- Model inference
Correct Answer: C
Correct Answer
Answer C is correct because The evidence at a campus assurance engine detecting behavior outside normal patterns points to Machine-learning anomaly detection. It learns or models expected behavior and identifies observations that significantly depart from the normal pattern. That capability supports the requirement to detect behavior that departs materially from learned normal conditions.
Incorrect Answers
Answer B is incorrect because AI-driven baselining uses observed network behavior to establish what normal performance looks like for a specific environment. At a campus assurance engine detecting behavior outside normal patterns, it does not provide the requirement to detect behavior that departs materially from learned normal conditions. Machine-learning anomaly detection does.
Answer A is incorrect because AI-assisted issue correlation and root-cause guidance is intended for reducing alert noise when many symptoms trace back to the same network problem. The scenario at a campus assurance engine detecting behavior outside normal patterns instead requires a mechanism to detect behavior that departs materially from learned normal conditions. That is the role of Machine-learning anomaly detection.
Answer D is incorrect because This option uses Model inference for using live telemetry as input to a deployed model during day-to-day network operations. At a campus assurance engine detecting behavior outside normal patterns, the required function belongs to Machine-learning anomaly detection. The mechanism does not match.
Question 4
A design review for a wireless team learning normal roaming performance focuses on this use case: comparing current wireless or application KPIs with learned normal behavior for that site and time pattern. Which technology or concept should be selected? Choose ONE.
- AI-driven baselining
- Machine-learning trend analysis
- Model training on representative historical data
- Human validation of AI-generated recommendations
Correct Answer: A
Correct Answer
Answer A is correct because Choose AI-driven baselining for a wireless team learning normal roaming performance. The scenario needs to learn what normal performance looks like for this environment, and AI-driven baselining supplies that function. A single vendor default threshold cannot reflect the same deployment-specific normal range. That distinction is decisive.
Incorrect Answers
Answer B is incorrect because Machine-learning trend analysis finds recurring or directional patterns in telemetry over time to support proactive operations. That can be valid elsewhere, but a wireless team learning normal roaming performance needs to learn what normal performance looks like for this environment. AI-driven baselining matches that objective.
Answer C is incorrect because Model training on representative historical data adjusts model parameters from examples so the system can learn useful patterns before operational use. That can be valid elsewhere, but a wireless team learning normal roaming performance needs to learn what normal performance looks like for this environment. AI-driven baselining matches that objective.
Answer D is incorrect because Human validation of AI-generated recommendations checks generated content or recommended changes before they are trusted for production action. At a wireless team learning normal roaming performance, it does not provide the requirement to learn what normal performance looks like for this environment. AI-driven baselining does.
Question 5
A design review for a network planner finding a recurring growth trend focuses on this use case: identifying a steadily worsening utilization or error-rate pattern before users report a major outage. Which mechanism should the engineer use? Choose ONE.
- AI-assisted issue correlation and root-cause guidance
- Model inference
- High-quality and relevant telemetry for the ML workflow
- Machine-learning trend analysis
Correct Answer: D
Correct Answer
Answer D is correct because For a network planner finding a recurring growth trend, the requirement is to identify a directional or recurring pattern across time. Machine-learning trend analysis provides that function. It matches the evidence. A one-time packet capture explains a point in time but does not by itself model a longer trend.
Incorrect Answers
Answer A is incorrect because Using AI-assisted issue correlation and root-cause guidance, the design correlates multiple signals and known relationships so operators can focus on the most likely underlying cause. For a network planner finding a recurring growth trend, the missing function is to identify a directional or recurring pattern across time; Machine-learning trend analysis supplies it.
Answer B is incorrect because For a network planner finding a recurring growth trend, Model inference solves the wrong problem. It applies an already trained model to new operational data to produce a prediction, classification, anomaly score, or generated response. The scenario needs to identify a directional or recurring pattern across time, which points to Machine-learning trend analysis.
Answer C is incorrect because High-quality and relevant telemetry for the ML workflow is intended for improving operational AI results when missing, stale, or noisy data would otherwise distort the model output. The scenario at a network planner finding a recurring growth trend instead requires a mechanism to identify a directional or recurring pattern across time. That is the role of Machine-learning trend analysis.
Question 6
A design review for an operations tool correlating symptoms from several devices focuses on this use case: reducing alert noise when many symptoms trace back to the same network problem. Which technology or concept should be selected? Choose ONE.
- Model training on representative historical data
- AI-assisted issue correlation and root-cause guidance
- AI-driven comparative analytics
- Human validation of AI-generated recommendations
Correct Answer: B
Correct Answer
Answer B is correct because AI-assisted issue correlation and root-cause guidance fits an operations tool correlating symptoms from several devices: it correlates multiple signals and known relationships so operators can focus on the most likely underlying cause. That behavior matches the requirement to correlate symptoms so operators focus on a likely underlying cause. Other choices perform different roles.
Incorrect Answers
Answer A is incorrect because This option uses Model training on representative historical data for building a model from prior telemetry and labeled or unlabeled examples before applying it to live network observations. At an operations tool correlating symptoms from several devices, the required function belongs to AI-assisted issue correlation and root-cause guidance. The mechanism does not match.
Answer D is incorrect because Human validation of AI-generated recommendations is intended for preventing a plausible but incorrect generated command or explanation from being applied blindly to the network. The scenario at an operations tool correlating symptoms from several devices instead requires a mechanism to correlate symptoms so operators focus on a likely underlying cause. That is the role of AI-assisted issue correlation and root-cause guidance.
Answer C is incorrect because AI-driven comparative analytics compares network KPIs across time, sites, or peer baselines to highlight unusual relative behavior. That can be valid elsewhere, but an operations tool correlating symptoms from several devices needs to correlate symptoms so operators focus on a likely underlying cause. AI-assisted issue correlation and root-cause guidance matches that objective.
Question 7
A design review for an analytics team training a model with past incident data focuses on this use case: building a model from prior telemetry and labeled or unlabeled examples before applying it to live network observations. Which mechanism should the engineer use? Choose ONE.
- Predictive AI
- Model training on representative historical data
- Model inference
- High-quality and relevant telemetry for the ML workflow
Correct Answer: B
Correct Answer
Answer B is correct because Model training on representative historical data is appropriate for an analytics team training a model with past incident data. Its typical use is building a model from prior telemetry and labeled or unlabeled examples before applying it to live network observations. It adjusts model parameters from examples so the system can learn useful patterns before operational use. Both clues point to this option.
Incorrect Answers
Answer C is incorrect because Model inference applies an already trained model to new operational data to produce a prediction, classification, anomaly score, or generated response. At an analytics team training a model with past incident data, it does not provide the requirement to learn model parameters from representative historical examples. Model training on representative historical data does.
Answer D is incorrect because Using High-quality and relevant telemetry for the ML workflow, the design gives analytics models trustworthy features and observations that represent the network conditions they are expected to analyze. For an analytics team training a model with past incident data, the missing function is to learn model parameters from representative historical examples; Model training on representative historical data supplies it.
Answer A is incorrect because Predictive AI uses learned patterns in historical and current data to estimate likely future conditions or outcomes. At an analytics team training a model with past incident data, it does not provide the requirement to learn model parameters from representative historical examples. Model training on representative historical data does.
Question 8
A design review for a live service applying a trained model to fresh measurements focuses on this use case: using live telemetry as input to a deployed model during day-to-day network operations. Which technology or concept should be selected? Choose ONE.
- Model inference
- Human validation of AI-generated recommendations
- AI-driven comparative analytics
- Generative AI
Correct Answer: A
Correct Answer
Answer A is correct because Choose Model inference for a live service applying a trained model to fresh measurements. The scenario needs to apply a trained model to current operational input, and Model inference supplies that function. Training changes the model from examples, whereas inference uses the learned model without retraining for each observation. That distinction is decisive.
Incorrect Answers
Answer B is incorrect because Using Human validation of AI-generated recommendations, the design checks generated content or recommended changes before they are trusted for production action. For a live service applying a trained model to fresh measurements, the missing function is to apply a trained model to current operational input; Model inference supplies it.
Answer C is incorrect because This option uses AI-driven comparative analytics for finding that one branch performs abnormally compared with similar sites even though it has not crossed a universal threshold. At a live service applying a trained model to fresh measurements, the required function belongs to Model inference. The mechanism does not match.
Answer D is incorrect because This option uses Generative AI for turning a set of alerts and telemetry into a draft incident summary or proposed remediation explanation. At a live service applying a trained model to fresh measurements, the required function belongs to Model inference. The mechanism does not match.
Question 9
Engineers reviewing an assurance ML project that needs representative training examples and trustworthy input data need two complementary capabilities: one that adjusts model parameters from examples so the system can learn useful patterns before operational use, plus another that gives analytics models trustworthy features and observations that represent the network conditions they are expected to analyze. Select the TWO options that provide those capabilities. Choose TWO.
- Model inference
- High-quality and relevant telemetry for the ML workflow
- Model training on representative historical data
- Machine-learning anomaly detection
- Predictive AI
Correct Answers: B, C
Correct Answers
Answer C is correct because Model training on representative historical data is appropriate for an assurance ML project that needs representative training examples and trustworthy input data. Its typical use is building a model from prior telemetry and labeled or unlabeled examples before applying it to live network observations. It adjusts model parameters from examples so the system can learn useful patterns before operational use. Both clues point to this option.
Answer B is correct because The evidence at an assurance ML project that needs representative training examples and trustworthy input data points to High-quality and relevant telemetry for the ML workflow. It gives analytics models trustworthy features and observations that represent the network conditions they are expected to analyze. That capability supports the requirement to supply relevant, trustworthy telemetry to the analytics workflow.
Incorrect Answers
Answer A is incorrect because Using Model inference, the design applies an already trained model to new operational data to produce a prediction, classification, anomaly score, or generated response. An assurance ML project that needs representative training examples and trustworthy input data instead needs both Model training on representative historical data and High-quality and relevant telemetry for the ML workflow. This addresses a different mechanism.
Answer E is incorrect because Predictive AI does not satisfy the paired requirement at an assurance ML project that needs representative training examples and trustworthy input data. It uses learned patterns in historical and current data to estimate likely future conditions or outcomes. The needed choices are Model training on representative historical data and High-quality and relevant telemetry for the ML workflow.
Answer D is incorrect because Machine-learning anomaly detection learns or models expected behavior and identifies observations that significantly depart from the normal pattern. The pair required at an assurance ML project that needs representative training examples and trustworthy input data is Model training on representative historical data plus High-quality and relevant telemetry for the ML workflow. This option serves another role.
Question 10
A design review for a telemetry pipeline removing stale and missing observations focuses on this use case: improving operational AI results when missing, stale, or noisy data would otherwise distort the model output. Which technology or concept should be selected? Choose ONE.
- Generative AI
- AI-driven baselining
- High-quality and relevant telemetry for the ML workflow
- AI-driven comparative analytics
Correct Answer: C
Correct Answer
Answer C is correct because For a telemetry pipeline removing stale and missing observations, the requirement is to supply relevant, trustworthy telemetry to the analytics workflow. High-quality and relevant telemetry for the ML workflow provides that function. It matches the evidence. Adding model complexity cannot compensate reliably for telemetry that does not represent the environment.
Incorrect Answers
Answer D is incorrect because For a telemetry pipeline removing stale and missing observations, AI-driven comparative analytics solves the wrong problem. It compares network KPIs across time, sites, or peer baselines to highlight unusual relative behavior. The scenario needs to supply relevant, trustworthy telemetry to the analytics workflow, which points to High-quality and relevant telemetry for the ML workflow.
Answer A is incorrect because For a telemetry pipeline removing stale and missing observations, Generative AI solves the wrong problem. It creates new content such as explanations, summaries, suggested commands, or troubleshooting narratives from prompts and context. The scenario needs to supply relevant, trustworthy telemetry to the analytics workflow, which points to High-quality and relevant telemetry for the ML workflow.
Answer B is incorrect because AI-driven baselining is intended for comparing current wireless or application KPIs with learned normal behavior for that site and time pattern. The scenario at a telemetry pipeline removing stale and missing observations instead requires a mechanism to supply relevant, trustworthy telemetry to the analytics workflow. That is the role of High-quality and relevant telemetry for the ML workflow.
Question 11
A design review for an enterprise comparing a branch against peer locations focuses on this use case: finding that one branch performs abnormally compared with similar sites even though it has not crossed a universal threshold. Which mechanism should the engineer use? Choose ONE.
- Machine-learning trend analysis
- AI-driven comparative analytics
- Predictive AI
- Machine-learning anomaly detection
Correct Answer: B
Correct Answer
Answer B is correct because The evidence at an enterprise comparing a branch against peer locations points to AI-driven comparative analytics. It compares network KPIs across time, sites, or peer baselines to highlight unusual relative behavior. That capability supports the requirement to compare KPIs with peers or other sites to find relative outliers.
Incorrect Answers
Answer C is incorrect because Using Predictive AI, the design uses learned patterns in historical and current data to estimate likely future conditions or outcomes. For an enterprise comparing a branch against peer locations, the missing function is to compare KPIs with peers or other sites to find relative outliers; AI-driven comparative analytics supplies it.
Answer D is incorrect because Machine-learning anomaly detection is intended for surfacing an unusual latency, client-failure, or traffic pattern without relying only on one fixed threshold. The scenario at an enterprise comparing a branch against peer locations instead requires a mechanism to compare KPIs with peers or other sites to find relative outliers. That is the role of AI-driven comparative analytics.
Answer A is incorrect because This option uses Machine-learning trend analysis for identifying a steadily worsening utilization or error-rate pattern before users report a major outage. At an enterprise comparing a branch against peer locations, the required function belongs to AI-driven comparative analytics. The mechanism does not match.
Question 12
The case involving a reliability team estimating which link is likely to degrade turns on a specific distinction: generative AI primarily creates new content such as text, summaries, or suggested procedures rather than forecasting a numeric or categorical future outcome. Which option provides the required function? Choose ONE.
- AI-assisted issue correlation and root-cause guidance
- AI-driven baselining
- Generative AI
- Predictive AI
Correct Answer: D
Correct Answer
Answer D is correct because Predictive AI fits a reliability team estimating which link is likely to degrade: it uses learned patterns in historical and current data to estimate likely future conditions or outcomes. That behavior matches the requirement to estimate a likely future network condition from learned patterns. Other choices perform different roles.
Incorrect Answers
Answer C is incorrect because Generative AI creates new content such as explanations, summaries, suggested commands, or troubleshooting narratives from prompts and context. At a reliability team estimating which link is likely to degrade, it does not provide the requirement to estimate a likely future network condition from learned patterns. Predictive AI does.
Answer B is incorrect because Using AI-driven baselining, the design uses observed network behavior to establish what normal performance looks like for a specific environment. For a reliability team estimating which link is likely to degrade, the missing function is to estimate a likely future network condition from learned patterns; Predictive AI supplies it.
Answer A is incorrect because AI-assisted issue correlation and root-cause guidance correlates multiple signals and known relationships so operators can focus on the most likely underlying cause. That can be valid elsewhere, but a reliability team estimating which link is likely to degrade needs to estimate a likely future network condition from learned patterns. Predictive AI matches that objective.
Question 13
The case involving an AI assistant drafting a change explanation turns on a specific distinction: predictive AI estimates likely outcomes instead of producing a new narrative or configuration draft. Which option best addresses this requirement? Choose ONE.
- Machine-learning anomaly detection
- Machine-learning trend analysis
- Model training on representative historical data
- Generative AI
Correct Answer: D
Correct Answer
Answer D is correct because Choose Generative AI for an AI assistant drafting a change explanation. The scenario needs to create new text or suggested content from operational context, and Generative AI supplies that function. Predictive AI estimates likely outcomes instead of producing a new narrative or configuration draft. That distinction is decisive.
Incorrect Answers
Answer A is incorrect because Using Machine-learning anomaly detection, the design learns or models expected behavior and identifies observations that significantly depart from the normal pattern. For an AI assistant drafting a change explanation, the missing function is to create new text or suggested content from operational context; Generative AI supplies it.
Answer B is incorrect because For an AI assistant drafting a change explanation, Machine-learning trend analysis solves the wrong problem. It finds recurring or directional patterns in telemetry over time to support proactive operations. The scenario needs to create new text or suggested content from operational context, which points to Generative AI.
Answer C is incorrect because For an AI assistant drafting a change explanation, Model training on representative historical data solves the wrong problem. It adjusts model parameters from examples so the system can learn useful patterns before operational use. The scenario needs to create new text or suggested content from operational context, which points to Generative AI.
Question 14
The case involving an anomaly engine surfacing unusual packet-loss behavior turns on a specific distinction: static thresholding can alert on a predefined limit but does not inherently learn the environment baseline. Which option provides the required function? Choose ONE.
- AI-assisted issue correlation and root-cause guidance
- Model inference
- Machine-learning anomaly detection
- AI-driven baselining
Correct Answer: C
Correct Answer
Answer C is correct because Machine-learning anomaly detection fits an anomaly engine surfacing unusual packet-loss behavior: it learns or models expected behavior and identifies observations that significantly depart from the normal pattern. That behavior matches the requirement to detect behavior that departs materially from learned normal conditions. Other choices perform different roles.
Incorrect Answers
Answer D is incorrect because AI-driven baselining uses observed network behavior to establish what normal performance looks like for a specific environment. That can be valid elsewhere, but an anomaly engine surfacing unusual packet-loss behavior needs to detect behavior that departs materially from learned normal conditions. Machine-learning anomaly detection matches that objective.
Answer A is incorrect because This option uses AI-assisted issue correlation and root-cause guidance for reducing alert noise when many symptoms trace back to the same network problem. At an anomaly engine surfacing unusual packet-loss behavior, the required function belongs to Machine-learning anomaly detection. The mechanism does not match.
Answer B is incorrect because Model inference is intended for using live telemetry as input to a deployed model during day-to-day network operations. The scenario at an anomaly engine surfacing unusual packet-loss behavior instead requires a mechanism to detect behavior that departs materially from learned normal conditions. That is the role of Machine-learning anomaly detection.
Question 15
The case involving a controller learning normal latency by site and time turns on a specific distinction: a single vendor default threshold cannot reflect the same deployment-specific normal range. Which option best addresses this requirement? Choose ONE.
- Model training on representative historical data
- Machine-learning trend analysis
- AI-driven baselining
- Human validation of AI-generated recommendations
Correct Answer: C
Correct Answer
Answer C is correct because For a controller learning normal latency by site and time, the requirement is to learn what normal performance looks like for this environment. AI-driven baselining provides that function. It matches the evidence. A single vendor default threshold cannot reflect the same deployment-specific normal range.
Incorrect Answers
Answer B is incorrect because Machine-learning trend analysis finds recurring or directional patterns in telemetry over time to support proactive operations. At a controller learning normal latency by site and time, it does not provide the requirement to learn what normal performance looks like for this environment. AI-driven baselining does.
Answer A is incorrect because Model training on representative historical data adjusts model parameters from examples so the system can learn useful patterns before operational use. At a controller learning normal latency by site and time, it does not provide the requirement to learn what normal performance looks like for this environment. AI-driven baselining does.
Answer D is incorrect because Human validation of AI-generated recommendations checks generated content or recommended changes before they are trusted for production action. That can be valid elsewhere, but a controller learning normal latency by site and time needs to learn what normal performance looks like for this environment. AI-driven baselining matches that objective.
Question 16
The case involving an NOC spotting a gradual rise in retransmissions turns on a specific distinction: a one-time packet capture explains a point in time but does not by itself model a longer trend. Which option provides the required function? Choose ONE.
- Machine-learning trend analysis
- AI-assisted issue correlation and root-cause guidance
- Model inference
- High-quality and relevant telemetry for the ML workflow
Correct Answer: A
Correct Answer
Answer A is correct because Choose Machine-learning trend analysis for an NOC spotting a gradual rise in retransmissions. The scenario needs to identify a directional or recurring pattern across time, and Machine-learning trend analysis supplies that function. A one-time packet capture explains a point in time but does not by itself model a longer trend. That distinction is decisive.
Incorrect Answers
Answer B is incorrect because For an NOC spotting a gradual rise in retransmissions, AI-assisted issue correlation and root-cause guidance solves the wrong problem. It correlates multiple signals and known relationships so operators can focus on the most likely underlying cause. The scenario needs to identify a directional or recurring pattern across time, which points to Machine-learning trend analysis.
Answer C is incorrect because Using Model inference, the design applies an already trained model to new operational data to produce a prediction, classification, anomaly score, or generated response. For an NOC spotting a gradual rise in retransmissions, the missing function is to identify a directional or recurring pattern across time; Machine-learning trend analysis supplies it.
Answer D is incorrect because This option uses High-quality and relevant telemetry for the ML workflow for improving operational AI results when missing, stale, or noisy data would otherwise distort the model output. At an NOC spotting a gradual rise in retransmissions, the required function belongs to Machine-learning trend analysis. The mechanism does not match.
Question 17
The case involving an AI analytics service reducing duplicate alert noise turns on a specific distinction: simply collecting more raw alarms does not determine which symptom is causal. Which option best addresses this requirement? Choose ONE.
- Model training on representative historical data
- Human validation of AI-generated recommendations
- AI-driven comparative analytics
- AI-assisted issue correlation and root-cause guidance
Correct Answer: D
Correct Answer
Answer D is correct because The evidence at an AI analytics service reducing duplicate alert noise points to AI-assisted issue correlation and root-cause guidance. It correlates multiple signals and known relationships so operators can focus on the most likely underlying cause. That capability supports the requirement to correlate symptoms so operators focus on a likely underlying cause.
Incorrect Answers
Answer A is incorrect because Model training on representative historical data is intended for building a model from prior telemetry and labeled or unlabeled examples before applying it to live network observations. The scenario at an AI analytics service reducing duplicate alert noise instead requires a mechanism to correlate symptoms so operators focus on a likely underlying cause. That is the role of AI-assisted issue correlation and root-cause guidance.
Answer B is incorrect because This option uses Human validation of AI-generated recommendations for preventing a plausible but incorrect generated command or explanation from being applied blindly to the network. At an AI analytics service reducing duplicate alert noise, the required function belongs to AI-assisted issue correlation and root-cause guidance. The mechanism does not match.
Answer C is incorrect because AI-driven comparative analytics compares network KPIs across time, sites, or peer baselines to highlight unusual relative behavior. At an AI analytics service reducing duplicate alert noise, it does not provide the requirement to correlate symptoms so operators focus on a likely underlying cause. AI-assisted issue correlation and root-cause guidance does.
Question 18
Engineers reviewing a generative troubleshooting service whose recommendations require technical review before use need two complementary capabilities: one that creates new content such as explanations, summaries, suggested commands, or troubleshooting narratives from prompts and context, plus another that checks generated content or recommended changes before they are trusted for production action. Select the TWO options that provide those capabilities. Choose TWO.
- Human validation of AI-generated recommendations
- Generative AI
- Model training on representative historical data
- Machine-learning trend analysis
- Machine-learning anomaly detection
Correct Answers: A, B
Correct Answers
Answer B is correct because For a generative troubleshooting service whose recommendations require technical review before use, the requirement is to create new text or suggested content from operational context. Generative AI provides that function. It matches the evidence. Predictive AI estimates likely outcomes instead of producing a new narrative or configuration draft.
Answer A is correct because Human validation of AI-generated recommendations is appropriate for a generative troubleshooting service whose recommendations require technical review before use. Its typical use is preventing a plausible but incorrect generated command or explanation from being applied blindly to the network. It checks generated content or recommended changes before they are trusted for production action. Both clues point to this option.
Incorrect Answers
Answer E is incorrect because Machine-learning anomaly detection does not satisfy the paired requirement at a generative troubleshooting service whose recommendations require technical review before use. It learns or models expected behavior and identifies observations that significantly depart from the normal pattern. The needed choices are Generative AI and Human validation of AI-generated recommendations.
Answer D is incorrect because Machine-learning trend analysis finds recurring or directional patterns in telemetry over time to support proactive operations. The pair required at a generative troubleshooting service whose recommendations require technical review before use is Generative AI plus Human validation of AI-generated recommendations. This option serves another role.
Answer C is incorrect because Model training on representative historical data is used for building a model from prior telemetry and labeled or unlabeled examples before applying it to live network observations. In a generative troubleshooting service whose recommendations require technical review before use, the required functions come from Generative AI and Human validation of AI-generated recommendations. It is not one of them.
Question 19
The case involving a deployed model classifying current network conditions turns on a specific distinction: training changes the model from examples, whereas inference uses the learned model without retraining for each observation. Which option best addresses this requirement? Choose ONE.
- Generative AI
- Model inference
- Human validation of AI-generated recommendations
- AI-driven comparative analytics
Correct Answer: B
Correct Answer
Answer B is correct because For a deployed model classifying current network conditions, the requirement is to apply a trained model to current operational input. Model inference provides that function. It matches the evidence. Training changes the model from examples, whereas inference uses the learned model without retraining for each observation.
Incorrect Answers
Answer C is incorrect because For a deployed model classifying current network conditions, Human validation of AI-generated recommendations solves the wrong problem. It checks generated content or recommended changes before they are trusted for production action. The scenario needs to apply a trained model to current operational input, which points to Model inference.
Answer D is incorrect because AI-driven comparative analytics is intended for finding that one branch performs abnormally compared with similar sites even though it has not crossed a universal threshold. The scenario at a deployed model classifying current network conditions instead requires a mechanism to apply a trained model to current operational input. That is the role of Model inference.
Answer A is incorrect because Generative AI is intended for turning a set of alerts and telemetry into a draft incident summary or proposed remediation explanation. The scenario at a deployed model classifying current network conditions instead requires a mechanism to apply a trained model to current operational input. That is the role of Model inference.
Question 20
The case involving an operator verifying generated remediation against device state turns on a specific distinction: generative systems can produce fluent output that still needs technical verification against actual device state and policy. Which option provides the required function? Choose ONE.
- Human validation of AI-generated recommendations
- High-quality and relevant telemetry for the ML workflow
- Predictive AI
- Machine-learning anomaly detection
Correct Answer: A
Correct Answer
Answer A is correct because The evidence at an operator verifying generated remediation against device state points to Human validation of AI-generated recommendations. It checks generated content or recommended changes before they are trusted for production action. That capability supports the requirement to verify generated recommendations before production action.
Incorrect Answers
Answer B is incorrect because High-quality and relevant telemetry for the ML workflow gives analytics models trustworthy features and observations that represent the network conditions they are expected to analyze. At an operator verifying generated remediation against device state, it does not provide the requirement to verify generated recommendations before production action. Human validation of AI-generated recommendations does.
Answer C is incorrect because This option uses Predictive AI for forecasting capacity pressure, failure risk, or performance degradation before the event occurs. At an operator verifying generated remediation against device state, the required function belongs to Human validation of AI-generated recommendations. The mechanism does not match.
Answer D is incorrect because Machine-learning anomaly detection learns or models expected behavior and identifies observations that significantly depart from the normal pattern. That can be valid elsewhere, but an operator verifying generated remediation against device state needs to verify generated recommendations before production action. Human validation of AI-generated recommendations matches that objective.
