Artificial intelligence is like a highly capable but very literal assistant. A prompt is simply the instruction, question or brief you give that assistant. It is how you tell the AI what to do. Think of a prompt as the most precise brief you can give an artificial intelligence. It is much like telling a very capable assistant what to write, draw, summarise or translate for you. The better and more thoroughly you explain what you have in mind, the closer the result will be to what you expect. It can be a simple question such as “Translate ‘good morning’ into French”, but also a more complex task such as “Create an image of a cat sitting on the moon in the style of van Gogh”. The quality of your brief directly influences the quality of the result. For optimal output you can use a framework for writing prompts.
A prompt is the input instruction, query or set of data provided to a generative model in order to elicit a specific response. This input may take the form of natural language, code, image data or a combination of these, and serves as the primary specification of the desired output. A good prompt often contains not only the task itself but also context, examples (so-called few-shot prompting) or explicit formatting rules the model is to follow. The structure and clarity of the prompt directly influence the relevance, accuracy and style of the generated content. The process of systematically designing and optimising such inputs to achieve the best possible results is known as prompt engineering.