100% Free Microsoft AB-731 Practice Test Questions

Question 1

Contoso Bank currently uses several AI systems: a model that predicts which customers are likely to churn, an anomaly detection model for fraud, and a new initiative to auto-draft personalised savings-plan emails for clients.In an executive steering committee, the CIO asks you to clarify what is distinct about the new initiative compared to the existing ML models.Which statement best captures the difference between generative AI and other common enterprise AI models in this context?

A : Generative AI is any AI that runs in the cloud rather than on-premises, while other AI models typically run in data centres you own.

B : Generative AI models learn a probability distribution over the data and can produce new content (such as text, images, or code), whereas most other enterprise ML models focus on predicting labels, scores, or anomalies based on existing features.

C : Generative AI always works only on unstructured text, while other AI models are restricted to structured numerical data and cannot be applied to images or audio.

D : Generative AI replaces rule-based engines entirely, while other AI types still rely on explicitly coded if/then rules for every decision.

Answer: B

Question 2

A telco is training a churn model. Midway through, you discover that:10% of labels are wrong due to a bug in how “churn” was recordedSome important features have inconsistent units across systemsMany recent records are missing key fieldsWhat is the most accurate description of how this data quality will affect the model?

A : The model will be largely unaffected because modern algorithms are explicitly designed to ignore label noise and unit inconsistencies during training.

B : Poor labels, inconsistent units, and missing values can distort the patterns the model learns, reducing accuracy and making outputs harder to trust, especially for borderline customers.

C : Only missing values matter; label errors and unit issues are irrelevant because the optimisation process doesn’t look at them directly.

D : Data quality problems affect only training speed, not model behaviour, so your main risk is longer training time rather than worse predictions.

Answer: B

Question 3

Fabrikam Health is building a “Ward Rounds Copilot” for clinicians.RequirementsR1: Extract text from photos of whiteboard notes, patient labels, and printed ward lists taken on a tabletR2: Allow doctors to ask questions over clinical guidelines, local policies, and anonymised case summaries using natural-language and semantically relevant resultsR3: Provide a unified environment where the digital team can connect models and tools, configure the copilot, and manage lifecycle, security, and monitoringProposed solutionUse Azure AI Vision to perform OCR and image analysis on the photos from the wardUse Azure AI Search to index guidelines, policies, and case summaries and to provide semantic retrieval for the copilotUse Azure AI Foundry as the workspace where the team wires together models, tools (including Vision and Search), and lifecycle policies for the solutionDoes the proposed solution meet the requirements?

A : Yes

B : No

Answer: A

Question 4

Contoso Bank currently uses several AI systems: a model that predicts which customers are likely to churn, an anomaly detection model for fraud, and a new initiative to auto-draft personalised savings-plan emails for clients.In an executive steering committee, the CIO asks you to clarify what is distinct about the new initiative compared to the existing ML models.Which statement best captures the difference between generative AI and other common enterprise AI models in this context?

A : Generative AI is any AI that runs in the cloud rather than on-premises, while other AI models typically run in data centres you own.

B : Generative AI models learn a probability distribution over the data and can produce new content (such as text, images, or code), whereas most other enterprise ML models focus on predicting labels, scores, or anomalies based on existing features.

C : Generative AI always works only on unstructured text, while other AI models are restricted to structured numerical data and cannot be applied to images or audio.

D : Generative AI replaces rule-based engines entirely, while other AI types still rely on explicitly coded if/then rules for every decision.

Answer: B

Question 5

In a case where AI was used to predict student success, it was found to disadvantage students from certain backgrounds. What corrective action should be taken?

A : Replace the AI system with another technology immediately.

B : Re-evaluate the training data and model to address bias.

C : Continue using the system to gather more data on its performance.

D : Disable all reporting about the model’s performance to avoid complaints.

Answer: B

Free Practice Microsoft AB-731 Exam Questions 2026