Whatever you give an AI to work with: your question, instructions and examples.
A prompt is the whole input you give an AI. It can be a one-line question or a long message with examples and rules. Because the AI picks its next tokens based on that text, a different prompt means a different answer.
Vague prompts get vague answers. Saying who it's for, what format you want and how long it should be works much better. The 2020 GPT-3 paper showed that a few examples inside a prompt could teach a model a new task on the spot.
Since ChatGPT arrived in 2022, almost anyone can write prompts. Good prompting isn't tricking the AI; it's saying clearly what you want.
Think of it like this
Ordering food works the same way. Say "whatever's fine" and you get whatever. Say "a medium pizza, half pepperoni, extra cheese" and you get what you actually wanted.
For example
"Help with science" is weak; "Explain photosynthesis in three sentences for an eighth grader" is strong.
Records about this
Source-checked moments from AI history.
Fixed the scattered lineages of techniques into a single taxonomy. The v1 title was Survey of Prompting Techniques
The conversational UI opened prompting to non-experts, turning a skill of a few practitioners into a mass phenomenon
Introduced in-context learning, which performs a task from examples in the prompt alone, without updating parameters