Microsoft Azure AI Fundamentals

→ Why the correct answer is right:

Sentiment analysis is needed for the customer-review requirement because it's the technique that assigns positive, neutral, or negative sentiment labels. Summarization is needed for the transcript requirement because it produces a condensed version of longer content. The question combines two different text-analysis goals: emotional polarity and shortening content, requiring two distinct techniques.

→ Why the other options are wrong:

Option A: Keyword extraction identifies main concepts/talking points (helps highlight themes), but doesn't directly tell if review is positive/negative OR generate condensed readable transcript version

Option C: Entity detection categorizes named items (people, locations, organizations) for structured info extraction, but neither requirement asks for named entities (doesn't replace polarity detection or create summary)

Option E: Detect language is preprocessing for multilingual systems (tells what language text is written in), not sentiment classification or summarization (out of scope for requirements)

Quick Memory Tip 🧠

"Sentiment = positive/negative, Summarization = condensed transcript"

→ Why the correct answer is right:

Microsoft's multimodal chat completions documentation states that the model can read audio content from an accessible cloud location by passing the payload as audio_url. The example shows a user content item with "type": "audio_url" and a URL pointing to an MP3 file. audio_url is an input method, the URL-based alternative to sending encoded audio inline with input_audio.

→ Why the other options are wrong:

Option A: Requesting spoken output is handled through response settings (modalities=["text", "audio"] + audio block), not audio_url (audio_url is about input transport, not output configuration)

Option B: audio_url doesn't deploy anything (deployment done earlier under Models + endpoints; audio_url is just request payload after deployment)

Option D: Recording microphone input is Chat playground capability, not what audio_url field does in API request (audio_url points to hosted audio file)

Quick Memory Tip 🧠

"audio_url = cloud location audio input"

→ Why the correct answer is right:

Keyword extraction is used to identify main concepts in text, entity detection identifies and categorizes named items, and summarization condenses longer content. Those three pairings align directly with Microsoft's descriptions of the corresponding Azure Language capabilities. These are the key distinctions for understanding text analysis techniques.

→ Why the other options are wrong:

Option B: Sentiment analysis determines emotional tone (positive/neutral/negative), not categorizing people/places/organizations (entity detection does that)

Option E: Keyword extraction identifies main concepts/talking points, not polarity labels (sentiment analysis produces sentiment labels and confidence scores)

Option F: Summarization generates shorter representation of content, not extracting/categorizing named entities (entity detection identifies organizations/locations)

Quick Memory Tip 🧠

"Keywords = talking points, Entity = named items, Summarization = condensed text"

→ Why the correct answer is right:

Entity detection is the correct answer because this requirement is about finding and categorizing named items in text. Microsoft documents this capability as named entity recognition, which identifies entities such as people, places, organizations, and quantities in unstructured text. This directly matches the need to systematically extract and categorize people, organizations, and locations.

→ Why the other options are wrong:

Option A: Summarization shortens content while preserving central meaning (doesn't label specific phrases as person/location/organization entities)

Option C: Sentiment analysis measures positive/neutral/negative attitude (useful for satisfaction/reviews, not for categorizing names and places)

Option D: Keyword extraction finds main concepts/talking points (not meant to classify into entity categories like Organization or Location)

Quick Memory Tip 🧠

"Entity detection = people, places, organizations, quantities"

→ Why the correct answer is right:

Sentiment analysis is the best choice because it is designed to determine whether text expresses positive, neutral, or negative sentiment. Microsoft documents this as a core Azure Language capability, and it can return sentiment at both document and sentence level with confidence scores. This directly matches the requirement to classify product reviews by emotional tone.

→ Why the other options are wrong:

Option A: Keyword extraction identifies main concepts/talking points (topics like "battery life"), not emotional tone or polarity classification

Option B: Entity detection categorizes items (people, places, organizations, dates), not positive/neutral/negative labels (might detect brand but not if customer liked it)

Option D: Summarization creates condensed version of longer content (overview), not sentiment labels (classification of mood, not compression)

Quick Memory Tip 🧠

"Sentiment analysis = positive/neutral/negative classification"

→ Why the correct answer is right:

The organization remains accountable for how the AI solution is applied, governed, monitored, and reviewed, even when parts of the technology stack are managed by a provider. Microsoft's responsible AI guidance emphasizes that people and organizations must remain answerable for the behavior and use of systems they deploy. Governance boards, sign-off processes, documented practices, and human oversight still matter with managed services.

→ Why the other options are wrong:

Option A: Anonymous contradicts accountability (accountability requires named owners, approvers, documented responsibilities; anonymous governance makes review harder)

Option B: Unbiased relates to fairness, not accountability (can attempt fairness improvements but still lack clear accountability processes)

Option C: Encrypted is privacy/security (protects data), not accountability (doesn't establish who is responsible for approvals, monitoring, incident response)

Quick Memory Tip 🧠

"Managed services ≠ organization escapes accountability"

→ Why the correct answer is right:

Requiring a human reviewer for high-impact cases best reflects accountability because it preserves meaningful human control over decisions that can materially affect people. Microsoft states AI systems should not be the final authority on decisions affecting people's lives and humans should remain in control of highly autonomous systems. Accountability is about defining who reviews outcomes, who can override them, and who is responsible.

→ Why the other options are wrong:

Option A: Hiding escalation rules works against good governance and weakens oversight (reviewers need context for meaningful review; makes accountable oversight harder)

Option B: Replacing all reviewers with automation is opposite of accountability (removes human control, makes organizations defer responsibility to model)

Option C: Short data retention relates to privacy practices, not accountability (doesn't establish owners, sign-off process, or monitoring)

Quick Memory Tip 🧠

"Accountability = human reviewer for high-impact decisions"

→ Why the correct answer is right:

Representative is the correct answer because Microsoft guidance on responsible AI fairness emphasizes using diverse and representative data to reduce bias. If data doesn't reflect the people who will be affected by the AI system, some groups can be underrepresented and receive worse outcomes. Microsoft Learn also highlights using real-world data to evaluate fairness.

→ Why the other options are wrong:

Option A: Encrypted data protects confidentiality (privacy/security), but encryption doesn't make dataset balanced or representative (can be well-protected but still unfair)

Option C: Compressed data is storage/transmission concern, not fairness control (reducing file size doesn't address demographics or real-world usage)

Option D: Anonymized data supports privacy (reduces personal info exposure), but anonymization doesn't ensure all groups are fairly represented (can be anonymized but still biased)

Quick Memory Tip 🧠

"Reduce bias = representative data of affected people"

→ Why the correct answer is right:

Accountability is the principle that best fits this sentence. Microsoft describes accountability as requiring the people who design and deploy AI systems to remain responsible for how those systems operate, including how decisions are reviewed and corrected when needed. This shows up through governance, documented responsibilities, monitoring, review, and mechanisms that preserve human control over important decisions.

→ Why the other options are wrong:

Option A: Fairness is about avoiding unjust differences in treatment across similar people/groups (asks whether system behaves equitably, not about ownership/answerability)

Option B: Transparency is about helping people understand how AI decisions are made (explainability, disclosure), but doesn't establish who is responsible for system actions

Option D: Inclusiveness focuses on designing AI systems usable by people with diverse needs/abilities (accessibility/participation), not ownership of outcomes and governance

Quick Memory Tip 🧠

"Accountability = people remain answerable for AI system"

→ Why the correct answer is right:

Microsoft states that prebuilt analyzer definitions can change across API versions. To ensure consistent behavior, the documentation recommends making a copy of the prebuilt analyzer instead of relying on the prebuilt version directly in production. This is the production-safe customization path for Content Understanding analyzers in Foundry Tools.

→ Why the other options are wrong:

Option A: Calling prebuilt analyzer directly ignores Microsoft's warning that definitions can change across API versions (introduces unexpected behavior after service/API changes)

Option C: prebuilt-documentFields is a utility analyzer for key-value pairs, not a general replacement for version stability of customized analyzer behavior

Option D: tableFormat controls output format for tables (HTML vs markdown), not production stability across API versions (configuration distraction, not core answer)

Quick Memory Tip 🧠

"Copy prebuilt analyzer = stable production behavior across API versions"

→ Why the correct answer is right:

Evaluating results using real-world data from affected groups is the best practical step because Microsoft Learn explicitly highlights the importance of real-world data when assessing fairness. This helps teams see whether the system behaves equitably for the people who will actually be affected by it. AI-901 fairness questions usually focus on representative data, subgroup analysis, and unequal outcomes.

→ Why the other options are wrong:

Option A: Longer password policy supports privacy and security (access control), not fairness review (doesn't determine if model treats groups fairly)

Option C: Increasing batch size is ML tuning for training behavior/efficiency, not fairness evaluation (doesn't show if real users/groups treated equitably)

Option D: Disabling user feedback could reduce opportunities to identify harms or uneven impacts (removes feedback, doesn't improve fairness review)

Quick Memory Tip 🧠

"Fairness review = real-world data from affected groups"

→ Why the correct answer is right:

Statement 1 is correct because underrepresentation in training data can contribute to biased or uneven outcomes. Microsoft's responsible AI guidance emphasizes representative data and real-world evaluation as important fairness considerations. Statement 2 is incorrect because fairness and inclusiveness are separate responsible AI principles, even though they are related. Statement 3 is correct because comparing model behavior across demographic or sensitive groups is a practical fairness assessment step.

→ Why the other options are wrong:

Option A: Wrongly marks Statement 2 as true (fairness ≠ inclusiveness) and Statement 3 as false (group comparison is relevant for fairness evaluation)

Option C: Marks Statement 1 as false (wrong; underrepresentation is a common fairness concern) and Statement 2 as true (wrong; they are separate principles)

Option D: Marks Statement 1 as false (wrong) and repeats mistake on Statement 2; gets Statement 3 right but that's not enough

Quick Memory Tip 🧠

"Fairness ≠ Inclusiveness (separate principles), underrepresentation = fairness risk"

→ Why the correct answer is right:

model="ops-assistant" is correct because Microsoft documents that the deployment name is used in the model parameter during inference. The Code tab exists to show programmatic access details for the deployment you created in the portal, so this is the correct argument to target that deployment. This tests the small implementation distinction between portal deployment and client call mapping.

→ Why the other options are wrong:

Option A: endpoint is useful (viewable from deployment details page), but deployment guide specifically says deployment name is used in the model parameter, not endpoint

Option B: deployment_type is a configuration choice, not the field that identifies which deployed model handles inference requests

Option C: project is a valid Foundry concept but Microsoft doesn't describe project name as the inference argument that routes requests to deployed model

Quick Memory Tip 🧠

"model parameter = deployment name for inference routing"

→ Why the correct answer is right:

Evaluating model results across age-based groups is the best fairness-focused response because fairness issues are identified by checking whether similar people or groups are being affected differently. Microsoft's responsible AI guidance points to fairness assessment across sensitive groups as an important way to understand model behavior. This is the most relevant next step for potential bias in outcomes.

→ Why the other options are wrong:

Option A: Increasing model endpoints improves scale/availability but doesn't reveal whether system disadvantages one age group (operational, not fairness evaluation)

Option B: Removing audit trail reduces oversight rather than improving it (auditability supports accountability and bias investigation)

Option C: Publishing source code increases openness but doesn't directly test whether model treats age groups fairly (transparency ≠ fairness)

Quick Memory Tip 🧠

"Fairness issue = evaluate results across affected groups"

→ Why the correct answer is right:

Those three pairings match Microsoft's documented Foundry workflow. You get the OpenAI-compatible client from the project client, use the deployment name in model, and create a conversation for multi-turn chat when needed.

→ Why the other options are wrong:

Option A: AIProjectClient is the project client, not the final chat client for direct responses.

Option C: previous_response_id is not required for the first request.

Option E: az login authenticates you; it does not create deployments.

Quick Memory Tip 🧠

Project client → get OpenAI client → use deployment name.

→ Why the correct answer is right:

Speech recognition turns audio into text, and text to speech turns text into audio. Speaker diarization does not control voice style; it separates speakers in audio.

→ Why the other options are wrong:

Option B: It wrongly says text to speech is not correct.

Option C: It wrongly denies speech recognition.

Option D: It incorrectly denies both the recognition and synthesis statements.

Quick Memory Tip 🧠

Diarization = who spoke, not how the voice sounds.

→ Why the correct answer is right:

Speech recognition in Azure Speech includes converting spoken audio into text, and Microsoft lists both real-time transcription and batch transcription as core speech-to-text features. Real-time transcription is suited to live inputs, while batch transcription is designed for larger volumes of prerecorded audio.

→ Why the other options are wrong:

Option C: Neural voice output is part of text to speech.

Option D: SSML customization is used for synthesized speech.

Option E: Audio synthesis is also speech generation, not recognition.

Quick Memory Tip 🧠

Recognition = transcription.

→ Why the correct answer is right:

Microsoft's Azure AI Inference Python documentation shows that EmbeddingsClient uses the embed method to generate embeddings. The returned data includes embedding vectors, which are numeric representations of semantic meaning.

→ Why the other options are wrong:

Option A: complete is for chat or text generation, not for creating embeddings with EmbeddingsClient.

Option B: get_model_info retrieves endpoint metadata, not vectors for semantic comparison.

Option C: send_request is a lower-level transport method and not the intended high-level embeddings call shown in the SDK examples.

Quick Memory Tip 🧠

Embeddings client = embed().

→ Why the correct answer is right:

A base model is the original model before task-specific customization. It has not yet been fine-tuned for a particular business use case. That is the term Microsoft uses for the starting point of many generative AI solutions.

→ Why the other options are wrong:

Option A: A grounded response is an output supported by data, not a model type.

Option B: An agent session is an interaction context, not a model.

Option D: A vector store holds embeddings, not a model definition.

Quick Memory Tip 🧠

Untuned model = base model.

→ Why the correct answer is right:

Prompts shape the model's response, and fine-tuning can change the model's weights for a task. Those are both true descriptions of how generative AI systems are guided and adapted. The other statements are too absolute or describe different technologies.

→ Why the other options are wrong:

Option B: Outputs are not guaranteed to be identical in every context.

Option D: Base models do not automatically know private company data.

Option E: Embeddings are semantic vectors, not relational table rows.

Quick Memory Tip 🧠

Prompts guide output; fine-tuning changes weights.

→ Why the correct answer is right:

Microsoft's Responses API examples use input for the user prompt text. So the prompt goes in the input parameter when you call responses.create(...). This is the documented pattern for simple requests.

→ Why the other options are wrong:

Option B: prompt is a common word, but not the documented field here.

Option C: message belongs to a different request style.

Option D: content is used in some message structures, not this top-level field.

Quick Memory Tip 🧠

Responses API prompt field = input.

→ Why the correct answer is right:

Reliability and safety includes resisting harmful manipulation from adversarial or unexpected prompts. In other words, the system should remain safe and dependable even when someone tries to push it into bad behavior. That is exactly what this sentence is describing.

→ Why the other options are wrong:

Option A: Model cards are documentation, not something the system resists.

Option B: Stakeholder mapping is part of risk analysis, not a threat the model resists.

Option D: Explainability relates more to transparency than to resisting manipulation.

Quick Memory Tip 🧠

Adversarial prompts = harmful manipulation.

→ Why the correct answer is right:

These three terms are foundational. Embeddings represent meaning numerically, prompts are the instructions you give the model, and fine-tuning updates model weights using task-specific examples. Together they describe how modern AI systems are guided, adapted, and represented.

→ Why the other options are wrong:

Option B: A base model is not yet customized.

Option D: Hallucination is unsupported output, not guaranteed grounding.

Option F: A completion token is generated text, not a form field.

Quick Memory Tip 🧠

Embedding, prompt, and fine-tuning are separate concepts.

→ Why the correct answer is right:

Medical summarization is a sensitive scenario. The most important reliability-and-safety step is to test how the system behaves in high-risk, unusual, or potentially harmful situations before it goes live. That is how teams catch unsafe behavior early instead of discovering it after deployment.

→ Why the other options are wrong:

Option B: A brighter theme affects appearance, not safety.

Option C: Marketing copy does not reduce clinical risk.

Option D: Renaming the workspace changes administration, not model safety.

Quick Memory Tip 🧠

High-risk use case = test edge cases first.

→ Why the correct answer is right:

If the safety score drops below the threshold, the system should notify operators or trigger an alert. That is a practical reliability-and-safety response because it helps people react quickly when the model output is no longer within acceptable limits. Monitoring is useful only when it leads to action.

→ Why the other options are wrong:

Option A: A placeholder prompt does not respond to the safety issue.

Option B: Printing a model name does not alert anyone.

Option D: Setting the endpoint to None does not solve the problem.

Quick Memory Tip 🧠

Low safety score = alert operators.

→ Why the correct answer is right:

Microsoft's Python quickstart for speech recognition shows the recognizer calling recognize_once_async().get() to capture one utterance and return the recognition result. That matches the scenario here because the app is listening to the microphone and needs a speech-to-text result.

→ Why the other options are wrong:

Option A: speak_text_async is a text-to-speech method, so it belongs to synthesis rather than recognition.

Option C: SpeechSynthesizer is the class used for text-to-speech output, not for recognizing microphone input.

Option D: speak_ssml_async is also a synthesis method, specifically for synthesizing SSML input into speech.

Quick Memory Tip 🧠

One utterance from mic = recognize_once_async().

→ Why the correct answer is right:

When a moderation system misses harmful content or blocks safe content, you need tuning and control. Adjusting severity thresholds and using blocklists are direct ways to improve moderation behavior and align the system with your safety policy. They help reduce both false negatives and false positives.

→ Why the other options are wrong:

Option C: A color change has nothing to do with moderation quality.

Option D: Shortening the FAQ does not change safety behavior.

Option E: Changing an icon does not affect harmful-content detection.

Quick Memory Tip 🧠

Moderation tuning = thresholds + blocklists.

→ Why the correct answer is right:

Speech to text is the Azure Speech feature that listens to audio and converts it into text in real time. That makes it the right choice for live captions, dictation, and call transcription. The app needs recognition, not voice generation or search.

→ Why the other options are wrong:

Option A: Text to speech does the reverse job.

Option C: Agent Service is not the core speech-recognition capability.

Option D: Search is for retrieval, not transcription.

Quick Memory Tip 🧠

Live captions = speech to text.

→ Why the correct answer is right:

Microsoft's MAI image model documentation shows that image generation requests are sent to the mai/v1/images/generations endpoint path. This is specifically designed for producing new images from text prompts after a supported image generation model has been deployed.

→ Why the other options are wrong:

Option A: vision/v1/ocr is associated with reading and extracting text from existing images — the opposite of creating new ones.

Option B: speech/v1/synthesize is for generating spoken audio from text — wrong output modality entirely.

Option D: language/v1/entities is for entity recognition in text — a language analysis service, not image creation.

Quick Memory Tip 🧠

"MAI image generation endpoint = mai/v1/images/generations | OCR = vision/ocr | Speech = speech/synthesize | NLP = language/entities"

→ Why the correct answer is right:

All three correct pairs align with Microsoft's documented capability definitions: OCR reads text from images, object detection identifies and localizes objects with bounding box coordinates, and image captioning generates a human-readable sentence describing an image.

→ Why the other options are wrong:

Option B: Sentiment scores come from text analysis (NLP) — not from image generation models. Image generation creates visuals, it doesn't classify sentiment.

Option D: Speech synthesis converts text to audio — it doesn't analyze photos or create image captions. Captioning is a vision task.

Option F: OCR reads existing text from images — it cannot create a new poster from a prompt. Creating content from a prompt is image generation.

Quick Memory Tip 🧠

"OCR = Read text | Object Detection = Locate + Coordinates | Image Captioning = Describe in a sentence | None of these CREATE images"

→ Why the correct answer is right:

Microsoft's image generation documentation states these models create images from user-provided text prompts, and in some workflows they also accept optional images as input (for tasks like variations or inpainting). Both are documented image generation capabilities.

→ Why the other options are wrong:

Option C: Returning people-detection bounding boxes is a computer vision analysis capability — it identifies and locates people in existing images, not a generation task.

Option D: Reading printed text from posters is OCR — a text extraction capability from existing images, not image creation.

Option E: Producing sentiment labels is a text analysis (NLP) task — completely different workload family from image generation.

Quick Memory Tip 🧠

"Image Generation = Create from prompt + Optional image input | NOT analysis, NOT OCR, NOT sentiment"

→ Why the correct answer is right:

Microsoft's image generation documentation states that image generation models create images from user-provided text prompts. The prompt field is the core input in the request payload — it contains the natural language description that the model uses to generate the new image.

→ Why the other options are wrong:

Option B: "ocr" is not a field for image generation. OCR is a reading capability for existing images — not a creation request field.

Option C: "objects": ["bicycle", "park"] resembles image analysis output, not the text prompt that drives creation. The API uses a natural language prompt, not an object list.

Option D: "caption": True is associated with image analysis (generating a description of an existing image) — the opposite of creating a new one.

Quick Memory Tip 🧠

"Image generation request = prompt field | OCR = reads existing image text | Caption = describes existing image"

→ Why the correct answer is right:

Microsoft describes image captioning as generating a caption of an image in human-readable language using complete sentences. This is exactly the capability described — producing a descriptive sentence about what an existing image shows.

→ Why the other options are wrong:

Option A: OCR extracts text that is already printed or written in the image — it doesn't interpret the visual scene and write a new descriptive sentence.

Option C: Speech synthesis converts existing text into spoken audio — it doesn't analyze images or create descriptions.

Option D: Entity recognition identifies named entities in text — it requires text input and belongs to language analysis, not computer vision.

Quick Memory Tip 🧠

"Describe image in a sentence = Image Captioning | Read text from image = OCR | Text → Audio = Speech Synthesis"

→ Why the correct answer is right:

Microsoft's quickstart shows that PromptAgentDefinition requires an instructions field to define the agent's behavior. Instructions tell the agent how to act — without them, the resource is just a model reference without any agent-specific behavior.

→ Why the other options are wrong:

Option B: temperature=AGENT_NAME is nonsensical — temperature expects a numeric value, and the quickstart shows instructions as the core definition field, not generation temperature.

Option C: conversation=conversation.id is used when calling an agent in a multi-turn flow — not when defining the agent's version.

Option D: agent_reference={"name": AGENT_NAME} is used when referencing an existing agent in a call — not when creating one with PromptAgentDefinition.

Quick Memory Tip 🧠

"Create agent = PromptAgentDefinition(model=..., instructions=...) | Call agent = agent_reference in extra_body | They are different steps!"

→ Why the correct answer is right:

Embedding models convert text into semantic vectors for retrieval and similarity. Multimodal chat models accept text plus image input. Image generation models create new images from text prompts. All three are directly documented in Microsoft Learn.

→ Why the other options are wrong:

Option B: Returning similarity vectors is the job of an embedding model, not an image generation model.

Option C: Azure AI Search is a retrieval service — it does not create images. That's an image generation model's job.

Option E: Speech synthesis converts text to audio — it does not extract invoice fields. That's a document intelligence task.

Quick Memory Tip 🧠

"Embedding = Vectors | Multimodal Chat = Text + Images | Image Generation = Create images | Azure AI Search = Retrieval only"

→ Why the correct answer is right:

Microsoft Learn describes image generation models as creating images from user-provided text prompts and optional images. The same documentation notes editing capabilities like inpainting (filling or replacing parts of an image). Both capabilities — original image creation and inpainting — are signature image generation features.

→ Why the other options are wrong:

Option B: Returning floating-point vectors is the role of an embedding model — not an image generation model.

Option C: Scoring semantic similarity depends on embedding vectors and cosine similarity — an embedding/retrieval concept, not image generation.

Option D: Retrieval optimization points to embedding models and Azure AI Search — not visual content creation.

Quick Memory Tip 🧠

"Image Generation = Create images + Inpainting | Embedding = Vectors + Similarity scores | Search = Retrieval optimized"

→ Why the correct answer is right:

Avoiding hard-coded secrets reduces the risk of credential leakage and unauthorized access to AI resources and connected data. This is a core privacy and security practice — one of Microsoft's six Responsible AI principles. It protects both the system and any data it processes.

→ Why the other options are wrong:

Option A: Transparency is about communicating system behavior and limitations to users — not about how credentials are stored or managed.

Option C: Fairness addresses equitable treatment across groups and avoiding bias — it has nothing to do with secret management or authentication design.

Option D: Summarization is a text-analysis AI capability — not a Responsible AI principle, and unrelated to credential security.

Quick Memory Tip 🧠

"No hard-coded secrets → Privacy & Security | Explain system behavior → Transparency | Equal treatment → Fairness"

→ Why the correct answer is right:

Microsoft's Azure Direct Models privacy documentation explicitly states that (A) prompts and completions are not available to other customers, and (B) customer data is not used to improve Microsoft or third-party products or services without explicit permission. These are strong, documented privacy commitments.

→ Why the other options are wrong:

Option C: This directly contradicts Microsoft's documentation — prompts and completions are NOT shared with model providers like OpenAI.

Option D: Fine-tuned models are exclusively available to the customer who created them — they are not shared across tenants.

Option E: Uploaded training data is not public — it is processed and stored under controlled conditions for features like fine-tuning.

Quick Memory Tip 🧠

"Azure Direct Models = Your prompts are PRIVATE | Your data is NOT used to train others | Fine-tuned models = YOURS only"

→ Why the correct answer is right:

Microsoft Learn shows two supported image input approaches: (1) pass a public image URL for cloud-hosted images, and (2) use ImageUrl.load() to convert a local image file into a data URL. Both give the model actual image content for visual interpretation.

→ Why the other options are wrong:

Option A: OCR text output is extracted text from an image — not the image itself. The model needs image data, not a post-processed text substitute.

Option D: An embedding vector encodes semantic meaning for retrieval — it cannot represent image pixels for visual reasoning in a chat prompt.

Option E: A speech transcript is text derived from audio — not visual data. It belongs to a different modality entirely.

Quick Memory Tip 🧠

"Image input = Public URL OR Data URL (local file) | NOT vectors, NOT transcripts, NOT OCR text"

→ Why the correct answer is right:

Student conversations can contain sensitive personal information. Microsoft's guidance for operating generative AI systems recommends a privacy review when the system processes personal data. This is the review most directly tied to protecting student information before rollout.

→ Why the other options are wrong:

Option A: Accessibility review aligns with the inclusiveness principle — focusing on usability for people with different needs, not data protection.

Option B: Fairness review checks for equitable treatment and bias across groups — important for educational AI, but the stem specifically highlights personal conversation data.

Option D: Prompt creativity review is not a formal responsible AI review category in Microsoft's guidance — good prompt design is useful but not a governance review type.

Quick Memory Tip 🧠

"Student/personal data → Privacy Review | Group bias → Fairness Review | Usability for all → Accessibility Review"