McAfee-Secured Website

Exam Code: AI-103

Exam Name: Developing AI Apps and Agents on Azure

Certification Provider: Microsoft

Corresponding Certification: Microsoft Certified: Azure AI Apps and Agents Developer Associate

100% Microsoft AI-103 Guaranteed Success With Testking Training Materials

Get Real AI-103 Exam Questions & Expert Verified Answers as Experienced on the Real Exam!

187 Questions & Answers with Testing Engine

"Developing AI Apps and Agents on Azure Exam", also known as AI-103 exam, is a Microsoft certification exam.

Pass your tests with the always up-to-date AI-103 Exam Engine. Your AI-103 training materials keep you at the head of the pack!

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Was: $137.49
Now: $124.99

Microsoft AI-103 Practice Test 40

Question 390

A retailer supplies a photo of a specific chair and asks for a new background while preserving its upholstery and legs. What input design helps?

  1. Choose a multimodal prompt with a reference image for visual consistency.
  2. Use an audio-only input instead of the photograph.
  3. Describe the chair only as 'furniture' with no reference asset.
  4. Generate multiple approximate product variants and choose by aesthetics without comparing the approved reference.

Explanation

Correct Answer

A:

Multimodal reference grounding constrains generation to the actual source appearance.

Incorrect Answers

B:

Speech evidence does not encode the requested visual details.

C:

A generic prompt loses the unique upholstery and silhouette.

D:

Visual resemblance alone cannot guarantee preservation of the particular supplied product.

Question 391

A generative designer wants to use a photographer's reference image, but licensing is unclear. What source control should be performed?

  1. Assume all publicly viewable images are licensed for unrestricted derivatives.
  2. Preserve source attribution and usage permissions for supplied reference media.
  3. Generate a derivative first and address licensing only after publication.
  4. Preserve only generic asset tags while omitting the licensed photographer and permitted usage terms.

Explanation

Correct Answer

B:

Source permissions determine whether the creative workflow may lawfully transform or distribute the reference.

Incorrect Answers

A:

Public availability is not equivalent to permission for reuse.

C:

Post-release checks may be too late to avoid unauthorized use.

D:

Asset tags alone do not retain the required rights/provenance trail.

Question 392

A publisher needs an audit trail for campaign imagery created with generative tools. What provenance should asset management record?

  1. Store model, source, prompt version and approval provenance.
  2. Retain only the final PNG without any linked metadata.
  3. Discard source-media licenses once the image is edited.
  4. Record an approval timestamp but no model version, source assets or identified reviewer.

Explanation

Correct Answer

A:

Generation provenance supports dispute handling, traceability and controlled reuse.

Incorrect Answers

B:

An image alone may not reveal how or from which sources it was made.

C:

Derivative use may remain subject to the original permissions.

D:

A timestamp alone cannot reconstruct creation provenance or authorization.

Question 393

A user supplies a wiring diagram and asks which circuit breaker protects the branch labeled B-7. The team considers OCR but the correct answer requires following diagram connections. What capability is necessary?

  1. Use multimodal visual reasoning to trace B-7 to its upstream breaker in the original diagram.
  2. Run label OCR only and choose the nearest breaker label without tracing circuit wires.
  3. Use OCR on the diagram labels without tracing the circuit connections among them.
  4. Generate a plausible replacement wiring diagram and answer using its new connections.

Explanation

Correct Answer

A:

The model must reason about connected graphical elements, not merely read isolated label text.

Incorrect Answers

B:

Text recognition locates names but does not establish which upstream circuit connects to the queried branch.

C:

Reading text from a schematic does not identify the upstream protective breaker.

D:

Newly generated diagram geometry is not evidence of the actual breaker arrangement in the supplied schematic.

Question 394

An accessibility service needs a brief alt-text description for a single product image. The description must state visible content without inventing a manufacturing claim. What output design fits?

  1. Generate a marketing paragraph claiming the product meets every safety standard.
  2. Return only the internal asset UUID and no description of the picture.
  3. Generate an evidence-grounded caption, not unverifiable product claims.
  4. Return the product category from filename metadata as the complete accessibility description.

Explanation

Correct Answer

C:

Alt text can summarize visible objects and relationships while avoiding ungrounded product specifications.

Incorrect Answers

A:

Such claims cannot be inferred from pixels and exceed the accessibility caption requirement.

B:

An identifier is not an accessible description of visually relevant content.

D:

A category may provide general context but cannot describe the particular visible objects and relationships in the image.

Question 395

A museum prepares extended image descriptions for visitors with low vision. A short label such as “painting” omits the setting, composition, and major visual objects. What should the app request?

  1. A document byte count in place of visual description.
  2. Only an object-class label without relative placement.
  3. An automatically generated artist biography with unsupported life events.
  4. Request a grounded extended description of objects and spatial composition.

Explanation

Correct Answer

D:

Extended descriptions should convey relevant visible composition at greater depth without inventing unseen facts.

Incorrect Answers

A:

File size provides no useful description of subjects or spatial relationships.

B:

A single class is insufficient for the requested extended composition description.

C:

Biographical details are not necessarily visible and can become hallucinated content.

Question 396

A quality engineer has two photos of the same mechanical panel taken before and after maintenance and needs a list of visibly changed labels and switch positions. Which analysis strategy fits?

  1. Classify each photo into the same equipment category without comparing component positions.
  2. Compare multiple product photos to identify visual differences.
  3. Describe both photos independently without comparing corresponding component positions.
  4. Run text-to-image generation to produce an ideal panel and compare the ideal to nothing.

Explanation

Correct Answer

B:

Joint comparison can enumerate visual changes while explicitly flagging unclear or occluded parts.

Incorrect Answers

A:

Independent classification does not identify the changed switch states and labels between before-and-after photographs.

C:

Independent captions can miss the specific before-versus-after changes requested.

D:

A generated ideal state is not the requested comparison of two actual photographs.

Question 397

A financial chart image has a y-axis with irregular increments. An assistant estimates values from line height without reading tick labels. Which analysis practice is required?

  1. Redraw the plot from estimated data without validating the original y-axis intervals or labels.
  2. Assume the y-axis always begins at zero and increases uniformly.
  3. Ground measurements in visible axis labels, scale and plotted marks, noting uncertainty when unreadable.
  4. Use the chart title as the only source for every plotted value.

Explanation

Correct Answer

C:

The chart's actual scale determines numeric interpretation; visual height alone can mislead.

Incorrect Answers

A:

A guessed redraw can materially change the numeric relationship being measured.

B:

Charts may use different baselines or nonuniform scales.

D:

Titles do not encode exact point coordinates.

Question 398

A technician uploads a panel photo and asks which switch sits directly below label P-8. The image includes many similar switches. What model input is needed?

  1. Rely on a text-only model given the device model number alone.
  2. Locate equipment labels in the actual photograph using region-aware vision.
  3. Choose the most common switch in the model family without inspecting the panel.
  4. Generate a new ideal panel and use its position to answer.

Explanation

Correct Answer

B:

Spatial relationships and labels must be read together to identify the correct switch.

Incorrect Answers

A:

A product identifier does not reveal the photographed switch positions.

C:

Generic frequency is not evidence of the pictured arrangement.

D:

A generated representation can differ from the actual installation.

Question 399

An inspection app must return the bounding box of a cracked pipe, not a general description of a factory. What output should it request?

  1. Use broad image captioning without returning object-region localization for the crack.
  2. Return bounding-box coordinates for the cracked pipe using a supported object detector.
  3. Return a whole-image caption without coordinates.
  4. Return only an image-wide classification that a cracked pipe is present.

Explanation

Correct Answer

B:

Localization identifies where the defect appears, enabling a precise overlay.

Incorrect Answers

A:

A whole-scene caption cannot reliably identify the crack’s coordinates.

C:

A caption cannot provide the requested bounding box.

D:

A global classification cannot supply the requested bounding-box coordinates for localizing the defect.

Question 400

A training video shows a sequence of tightening, testing and releasing equipment. Users must identify when the test occurs. Which analysis is appropriate?

  1. Use only the upload date to identify the testing step.
  2. Choose video segment analysis for a changing assembly sequence.
  3. Ask a text model to invent the maintenance sequence without viewing frames.
  4. Caption only the video's opening frame.

Explanation

Correct Answer

B:

Temporal segmentation permits retrieval of changing procedural steps in order.

Incorrect Answers

A:

Metadata cannot indicate the visible timeline.

C:

Guessed actions are not grounded in footage.

D:

The required action may occur much later.

Developing AI Apps and Agents on Azure (AI-103) Practice Tests