What is prompt engineering?
Generative AI models are powerful tools that can produce various kinds of outputs, such as text, images, and code. To use these models effectively, you need to know how to communicate with them and guide their output. This is where prompt engineering comes in. Prompt engineering is the skill of crafting prompts that produce the desired output from the model.
What is a prompt?
A prompt is the input that you give to a generative AI model to tell it what you want it to do and how you want it to do it. A prompt consists of two main components: instruction and context.
Instruction. The instruction is the written part of the prompt that states the task and the objective for the model. For example, do you want the model to write, revise, create, or something else? What is your purpose for using the model? What kind of information should the model use or avoid?
The instructions should be clear and specific, so that the model knows exactly what you expect from it. For example, if you want the model to write a summary of an article, you should include the article, the length of the summary needed, and the main points to include.
Context. The context is the part of the prompt that provides information for the model's output. For example, who is your intended audience and what tone do you want to use? Is your output for personal or professional use? Do you want the model to produce a simple or complex explanation of a topic? Are you trying to persuade using pathos, ethos, or logos?
The context should be relevant and appropriate, so that the model can tailor its output to your needs. For example, if you want the model to help create a training plan over utilizing AI in a study session, you should tell it the level of expertise of your readers, the style and format for the plan, and any information you what your learners to receive.
Prompts are the interface between the user and the generative AI model's capabilities. The quality and effectiveness of the model's output depend largely on the quality and effectiveness of the prompt. To create good prompts, the user needs to use critical thinking skills to define the desired output. A well-structured prompt can help the model generate a high quality, relevant, and useful output. A poorly written prompt can lead to a low quality, irrelevant, or inaccurate output.
For further understanding, refer to the example below.
Prompt: Summarize this email in three bullets in concise language. The email contains customer feedback from the past week.
- Instruction. The instruction is a section of a prompt that provides clear guidance on what steps an AI model should take to generate an output. For example, when you want the model to summarize an email using three bullet points in simple language, you can use the instruction, "Summarize this email in three bullets in concise language." Clear instructions communicate to the model what is expected based on the information presented in the context. Creating parameters and making your audience clear is an essential part of prompting efficiently.
- Context. The context constitutes part of the prompt and supplies all the relevant details the AI model needs to generate relevant and accurate responses. The context for the prompt given above is, "The email contains customer feedback from the past week." When you provide an AI model with context, the model is informed about the specific content, domain, or theme it should focus on. With good context, the model can generate meaningful and contextually coherent content.