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.

• ✗

  A technical error message returned when the model fails

  Why it's wrong here

  This option describes an output artifact rather than an input configuration. A technical error message—such as a 429 rate-limit, 400 invalid request, or a content-filter rejection—is a response returned by the hosting service after a failure occurs, not an instruction embedded in the request. A system prompt is proactive and defines the model's operating parameters ahead of generation, while error messages are reactive signals that bear no influence on the model's behavior, persona, or constraints.

• ✓

  An instruction that defines the model's behavior, persona, and constraints for a session

  Why this is correct

  This is the correct definition. A system prompt is an instruction included in the messages array with the 'system' role, which primes the model to adopt a particular persona, follow style guidelines, respect boundary constraints, and apply them consistently throughout the entire session. It differs from user prompts by being an application-level directive that governs the model's behavior across all subsequent turns, often used to enforce safety rules, set response formats, or define domain-specific knowledge handling.

• ✗

  A user's first message to start a conversation

  Why it's wrong here

  This option conflates the conversational turn order with the role-based prompting structure. The user's first message is delivered under the 'user' role and represents the human's initial input, whereas a system prompt is provided separately by the application developer before any user turn arrives. In APIs like Azure OpenAI Chat Completions, the system prompt occupies its own system-role message that persists across the conversation, giving high-level directives rather than starting a dialogue.

• ✗

  A command to restart the AI model instance

  Why it's wrong here

  This option confuses an infrastructure control-plane action with an inference-time configuration. Issuing a restart command (such as POST on a deployment endpoint or stopping/starting a compute instance in Azure AI) reinitializes the model's host process and memory, but it does not alter how the model interprets prompts. A system prompt is instead a text-based instruction sent inside the request payload via the 'system' role, which only affects generation behavior, not the underlying service lifecycle.

Go deeper

Related to this question