AB-731 Practice Questions

Answer Area

Larger datasets can increase the cost of a generative AI solution that uses an Azure Machine Learning workspace. Answer: Yes

The cost of consuming Azure OpenAI models is primarily identified by the number of input and output tokens processed. Answer: Yes

The cost of custom generative AI solutions always remains the same regardless of the model version or capability used. Answer: No

Yes — In Azure Machine Learning, cost is driven by the resources you consume to store, process, and train. Larger datasets typically require more storage, more data transfer, and more compute time for preprocessing, training, evaluation, and experimentation. Even if you are not training foundation models, handling larger corpora can increase pipeline duration and the number/size of compute instances used, which increases overall cost.

Yes — For Azure OpenAI usage under Standard (on-demand), pricing is primarily tied to token-based consumption (input tokens + output tokens). The more context you send and the longer the generated responses, the more tokens you consume, and the higher the cost. This is why prompt optimization, response length controls, and grounding strategies matter for cost management.

No — Costs vary with model choice and capability. Different model families and versions have different price points, and larger/more capable models generally cost more per token or per unit of throughput. Additionally, architecture choices (RAG, vector search, caching), usage patterns, and throughput requirements can significantly change total cost—so it is never “always the same.”

When a non-reasoning generative AI model produces inconsistent outputs, the most reliable improvement is to make the prompt more specific, constrained, and demonstrative of what “good” looks like.

A is correct because adding high-quality examples is a form of few-shot prompting. Examples act like “training wheels” at inference time: they show the model the desired structure, tone, level of detail, formatting rules, and boundaries. This reduces ambiguity and variance, especially for tasks like marketing copy, summaries, policy text, or customer replies. The more your examples resemble real target outputs (including edge cases), the more consistent the model’s completions become.

B is correct because adding context, relevant source material, and explicit expectations narrows the model’s degrees of freedom. Including the intended audience, purpose, constraints (length, voice, banned claims), and trusted reference content (approved facts, product specs, policy excerpts) helps the model stay aligned and reduces hallucinations and off-brand language. This is also where you specify acceptance criteria such as “must include 3 bullet points,” “use UK English,” or “cite only provided text.”

C is not best: technical jargon can confuse or bias output if it’s not aligned to the task; clarity beats jargon. D is not best: a single concise requirement is usually under-specified and often increases variability.

Provisioned (PTUs)

For continuous, large-scale workloads, the key needs are predictable throughput, consistent latency, and cost/performance stability . Azure OpenAI Provisioned Throughput Units (PTUs) are designed for this scenario because you reserve model capacity to meet sustained demand. This reduces the risk of variability that can occur with purely on-demand usage during peak periods and provides a more predictable operating model when the workload is always “on” and high volume.

Standard (On-Demand) is best when usage is variable or you’re starting small (PoCs, pilots, spiky workloads), because it is flexible pay-as-you-go but can be less predictable at sustained scale. Batch API is optimized for asynchronous, non-interactive processing where you can tolerate delayed results (for example, large offline summarization jobs), not for always-on, real-time interactions at scale.

Therefore, for continuous high-volume production workloads, Provisioned (PTUs) is the best pricing model choice.

Answer Area

Microsoft Copilot in Outlook can automatically send follow-up replies to emails. Answer: No

Microsoft Copilot in Outlook can identify and surface tasks mentioned in your emails. Answer: Yes

Microsoft Copilot in Outlook can summarize long email threads and highlight action items and deadlines. Answer: Yes

1) No — Copilot in Outlook can draft email responses and help you take actions, but sending messages is not positioned as an automatic “follow-up reply” behavior. Microsoft describes capabilities such as “Draft with Copilot” (generating email drafts from prompts) and Copilot chat actions that update the mailbox, which implies user-directed execution rather than autonomous follow-up sending.

2) Yes — Copilot supports contextual help over your inbox and can help you “catch up” and understand what’s due, which includes surfacing action-oriented items from recent communications. This aligns with using Copilot to pull together summaries of email messages and identify what requires attention.

3) Yes — Microsoft explicitly documents email-thread summarization in Outlook: Copilot scans a conversation and produces a summary of key points. Microsoft also describes Copilot’s summarization capability as extracting key points from threads, which is commonly used to surface what matters most (including actions to take) when the thread is long.

To align an AI customer support agent with Microsoft’s Responsible AI principles, two high-impact actions are fairness/inclusiveness validation and transparency to users . B is correct because testing for inclusive and culturally sensitive responses directly supports fairness and helps reduce harm. In practice, you evaluate responses across diverse user personas, languages/dialects, accessibility scenarios, and sensitive contexts. You look for biased assumptions, stereotyping, exclusionary language, and disparate quality of service. This also implies ongoing monitoring because model behavior can drift as prompts, knowledge sources, and user inputs evolve.

E is correct because a clear disclaimer supports transparency: customers should know they are interacting with an AI system, understand the type of assistance it can provide, and know what to do if the response is incorrect or they need a human. A disclosure is also a practical risk-control that reduces overreliance and sets expectations about limitations.

The other options are not best for Responsible AI alignment: A (retain all conversations) can conflict with privacy/data minimization; retention must be justified and governed, not automatic. C (operate independently) undermines accountability and human oversight. D (multiple purposes) increases scope and risk rather than improving responsible use.

Answer Area

A generative AI model guarantees factually accurate responses if the model is trained on a large dataset. Answer: No

Content filtering and responsible AI safeguards help a generative AI model generate safe and inoffensive content. Answer: Yes

A generative AI model always produces fair and unbiased results when the training data has been properly prepared and reviewed for fairness. Answer: No

No — A larger training dataset can improve coverage and fluency, but it does not guarantee factual accuracy. Generative models can still hallucinate, mix concepts, or produce plausible-but-incorrect statements because they generate likely text rather than verifying truth. This is why solution designs commonly add grounding/retrieval, validation, and human review for high-stakes outputs.

Yes — Content filtering and Responsible AI controls are specifically used to reduce harmful, unsafe, or policy-violating outputs . In practice, safeguards include input/output filters, safety classifiers, and governance controls that help enforce safety policies and minimize offensive content. These controls don’t make outputs “perfect,” but they materially reduce risk and are a standard part of production AI deployments.

No — Even with careful data preparation and fairness reviews, models can still produce biased outcomes due to residual bias in data, label/measurement issues, deployment context, and shifting real-world distributions. “Always fair and unbiased” is an absolute claim that is not achievable in real systems; fairness is managed through continuous evaluation, monitoring, and mitigations—not assumed as guaranteed.

Answer Area

Extracting structured data from forms and invoices: Answer: Azure Document Intelligence in Foundry Tools

Summarizing written content from business reports: Answer: Azure Language in Foundry Tools

Generating descriptive text for uploaded images: Answer: Azure Vision in Foundry Tools

These three tasks align to three different Azure AI capability families: document processing, language understanding/generation, and computer vision.

Forms and invoices are semi-structured documents where the business need is to extract specific fields (IDs, names, totals, dates) reliably into structured output. Azure Document Intelligence is designed for intelligent document processing and includes prebuilt models (such as invoices) as well as custom extraction options, making it the correct choice for structured data extraction from documents.

Summarizing written business reports is an NLP task focused on compressing long text into key points, themes, and action items. Azure Language provides language processing capabilities (including summarization features within language service capabilities), so it is the best fit for summarization scenarios.

Generating descriptive text for images (image captioning/description) is a computer vision task. Azure Vision can analyze uploaded images and return descriptions/captions and other visual insights, which directly matches the requirement to produce descriptive text from images.