Back to glossary

Prompt

İstem

D3

A prompt is input text or other input supplied at inference time to guide a generative model’s response for a task.

Review status: 2026-11-26

Technical explanation

A prompt can include instructions, examples, and supporting content for the current task, so its supplied input can shape the model’s output.

Conceptual boundaries

A prompt is request-time input rather than model training; in the cited Azure guidance, examples condition only the current inference rather than permanently changing the model.

Provider-neutral example

A support assistant can receive the prompt “Summarize this ticket in three bullets and flag missing account details.”

Limitations

Prompting can influence behavior, but prompt-engineering tests do not by themselves establish validity or reliability across heterogeneous use contexts.

Related concepts

Atomic claims and evidence

  1. 1.1NIST AI 100-2e2025, Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations
    Source
    NIST AI 100-2e2025, Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations
    Source role
    Authoritative source
    Exact locator
    Section 3.1.1 printed p. 38: In-context instructions and system prompts
    Supported claim
    In-context instructions supplied at inference time can shape LLM behavior, and prompts are used to align a model to an application-specific use case.
    Last verification
    Review due
    Scope limitation
    This describes inference-time prompting in LLM applications and does not claim that an instruction is always followed.
  2. 2.1NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile — D3 evidence slice
    Source
    NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile — D3 evidence slice
    Source role
    Authoritative source
    Exact locator
    Appendix A.1.4 printed p. 50: prompt sensitivity and broad heterogeneity of contexts of use
    Supported claim
    Prompt-sensitive behavior varies across contexts of use, so prompting is not a guarantee of validity or reliability.
    Last verification
    Review due
    Scope limitation
    This is a risk statement about generative-AI use and does not measure any particular prompt or model.
  3. 3.1Microsoft Learn, Prompt engineering techniques for Azure OpenAI
    Source
    Microsoft Learn, Prompt engineering techniques for Azure OpenAI
    Source role
    Authoritative source
    Exact locator
    Examples; Supporting content; Few-shot learning
    Supported claim
    Microsoft’s Azure OpenAI prompt guidance presents instructions, few-shot examples, and supporting content as prompt components for a task and states that prompt examples condition only the current inference rather than permanently changing the model.
    Last verification
    Review due
    Scope limitation
    This is provider guidance for Azure OpenAI and does not prescribe a universal prompt format.