Prompt

A prompt is a command, question or instruction given to an artificial intelligence model that shapes the content and form of the generated response. It is the primary way of communicating with language models — the precision and completeness of the prompt largely determine the accuracy, tone and usefulness of the output you receive.

How a prompt works

A prompt is fed to the model, which uses it to predict the most likely continuation or answer, drawing on the knowledge gained during training. The model does not 'understand' the instruction the way a human does — it responds to patterns in the text. That is why the way a prompt is worded matters enormously. The key elements of an effective prompt are:

• Context — the background and information the model needs to answer accurately.
• Instruction — a clear statement of the task and the expected result.
• Format — specifying the response structure, such as a list, table or code.
• Examples — patterns showing the desired style and level of detail.

Prompts in practice

Prompts are at the heart of working with generative artificial intelligence:

1. Content generation — texts, summaries, translations and ideas.
2. Coding assistance — writing and explaining code, debugging and documentation.
3. Automation — driving large language models in applications and business processes.

The skill of designing prompts, known as prompt engineering, is becoming a discipline in its own right. Before a prompt ships into a production application, however, its effectiveness is worth validating as a proof of concept, confirming the model genuinely performs the intended task.