AI-900 Practice Question: Describe features of generative AI workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  GPT-4

  Why it's wrong here

  GPT-4 is a large language model (LLM) designed for natural language understanding and generation from textual input, but it does not have a native audio encoder. While GPT-4 can summarize, translate, or restructure text once it is provided, it cannot ingest raw audio waveforms or perform speech recognition. Meeting transcription requires converting sound into text, which is a speech-to-text task outside GPT-4's capabilities. Therefore, GPT-4 is incorrect for this use case.

  When this WOULD be correct

  If the question asked for a model to generate meeting summaries or answer questions based on transcribed text, GPT-4 would be the correct choice. For example: 'Which Azure OpenAI Service model can generate a concise summary of a meeting transcript?'

• ✗

  DALL-E 2

  Why it's wrong here

  DALL-E 2 is a generative image model that uses diffusion to synthesize high-quality visuals from text prompts, leveraging CLIP embeddings to align text and images. It processes image patches and text tokens, but it has no mechanism to parse audio input or detect phonemes, prosody, or language. As a result, DALL-E 2 cannot produce text transcripts from meeting recordings; its outputs are exclusively image-based. This makes it fundamentally wrong for the transcription requirement.

  When this WOULD be correct

  A question asks: 'Which Azure OpenAI Service model can generate images from textual descriptions?' In that context, DALL-E 2 would be the correct answer.

• ✓

  Whisper

  Why this is correct

  Whisper is an open-source automatic speech recognition (ASR) model developed by OpenAI, trained on a massive corpus of multilingual audio spanning 96 languages. Its encoder-decoder transformer architecture processes raw audio spectrograms and outputs text tokens, enabling both direct transcription and speech-to-English translation. This makes Whisper uniquely suited to convert multilingual meeting recordings into accurate written transcripts, even in noisy or accented conditions. Thus, Whisper is the correct choice for the stated requirement.

• ✗

  Codex

  Why it's wrong here

  Codex is a specialized variant of GPT-3 fine-tuned to generate source code—such as Python, JavaScript, or SQL—from natural language intents. Its training focused on public code repositories, and its inference pipeline accepts text tokens, not audio spectrograms or speech signals. Codex has no acoustic model or speech-to-text decoder, so it cannot transcribe multilingual meeting audio into text. Hence, Codex is entirely unsuitable for transcription workloads.

  When this WOULD be correct

  Codex would be correct if the question asked: 'Which Azure OpenAI Service model is designed to generate code from natural language prompts or to assist with programming tasks?'

Option-by-option analysis

Why each answer is right or wrong

Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The AI-900 exam frequently reuses these exact scenarios with slightly different constraints.

Analysis generated from the official AI-900blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”

Go deeper

Related to this question