Prompt engineering plays a crucial role in maximising the potential of generative AI, particularly in language models. It is not limited to the mere creation and formulation of prompts, it encompasses a diverse array of skills and techniques.

What are prompts?

They are input elements such as words, phrases, questions, or keywords that users enter into AI tools to generate results. Now that you know what prompts mean, prompt generation is the skilful art of crafting highly effective prompts to achieve the desired outcomes.

In the context of ChatGPT, prompts refer to the instruction or discussion topic that users provide for the AI model to respond to. It can take the form of a question, statement, or any stimulus aimed at fostering creativity, reflection, or engagement. These prompts are designed to be open ended allowing customisation to align with the user's preferences and interests.

 

What makes a good prompt?

Firstly, when using AI models like ChatGPT and Midjourney, the better your prompts, the better the results you'll get. Simply put, an effective prompt not only sets the topic but also provides clear instructions on the task, effectively guiding the AI.

Is it challenging to write a good prompt? Creating effective prompts can be a challenging task, as the quality of the output relies on the specificity and clarity of the prompts or input instructions. This is why we've created this whole resource site, so that you can see some effective prompts for learning.

---

Principles for good prompting

1. Craft precise and unambiguous instructions.
2. Use the conversational interface on many generative AI tools to refine its outputs.

Specificity is key! The model will make an assumption on any point you don’t clearly explain.

 

Try to refine

There is no hard science to good prompting. Instead, approach it as a process of trial and error:

1. Formulate a relevant prompt. 
2. Consider when the completions fall short of expectations.
3. Enhance the clarity of the prompt. You could refine your prompts using a set of exemplars, such as the ones found on this site.
4. Repeat the process to iteratively improve your prompt's effectiveness.

Remember that as you chat with generative AI, the model will learn more about what you need. Use the conversational turns to help the AI improve its outputs.

Also remember that simple prompts lead to simple (and similar) completions.

---

The structure of a good prompt

A good prompt has 4 key elements: Role, Task, Requirements and Instructions.

Let's take a look at each one in depth.

Role

Prompts starting with "act as..." or "pretend to be..." will provide responses similar to that of the role which you provide. Setting a specific role for a given prompt increases the likelihood of more accurate information, when done appropriately. 

• E.g. act as an expert in the field of computer science
• Determines the type of information and the way the information is communicated to you.
• This also determines the interactivity of the conversation.

Task

The task is a summary of what you want the prompt to do. There is a lot of creativity that comes into writing a great task. It can range from generating birthday gift ideas to doing game show questions with the content from your last lecture.

• Outline what you want the AI to do.
• Be specific about the task’s objective, as unclear objectives lead to worse outputs.

Requirements

Writing clear requirements is all about giving as much information as possible to ensure your response doesn't use any incorrect assumptions. AI models make assumptions for any information they don't have in the prompt. 

• It is often important to define what an output should look like and conditions that affect the output.
• This includes the various conditions to limit the assumptions made by the model.

Instructions

Instructions will inform the AI how to complete the task. Instructions can include examples of how it is supposed to work, steps it can follow or any information. 

• How should the AI go about completing its task.
• Examples of how it could go about tasks.
• Feedback on steps it has taken will improve this even further.

 

Copyright © The University of Sydney. Unless otherwise indicated, 3rd party material has been reproduced and communicated to you by or on behalf of the University of Sydney in accordance with section 113P of the Copyright Act 1968 (Act). The material in this communication may be subject to copyright under the Act. Any further reproduction or communication of this material by you may be the subject of copyright protection under the Act.