2 Prompt engineering with ChatGPT

 

This chapter covers

  • Structuring a good prompt to make the most of ChatGPT
  • Distinguishing between zero-shot, one-shot, and few-shot prompts
  • Examining role prompting and how can it help you become a ChatGPT power user
  • Determining how to use ChatGPT’s Custom Instructions feature for a tailored user experience

This chapter will teach you the art of prompt engineering, allowing you to harness the full potential of LLMs. We will delve into practical examples and real-world applications of various prompt engineering techniques, and I encourage you to follow alongside the presented content. This hands-on approach ensures that you not only learn the theoretical aspects of prompt design but also its practical application. By the end of this chapter, you will have a strong enough understanding of prompt engineering to tailor ChatGPT’s responses to your specific tasks and use cases.

2.1 Structuring a good prompt

In the previous chapter, we briefly looked at a simple example highlighting the difference between a good prompt and a bad one. These four essential features define an effective prompt: clarity, specificity, instruction, and assumptions.

2.1.1 Clarity

2.1.2 Specificity

2.1.3 Instruction

2.1.4 Assumptions

2.1.5 Recap

2.2 Zero-Shot, One-Shot, and Few-Shot prompts

2.2.1 Zero-Shot prompts

2.2.2 One-Shot prompts

2.2.3 Few-Shot prompts

2.2.4 Reasons to use Few-Shot prompts

2.3 Introduction to role prompting

2.4 Custom instructions for better responses

2.4.1 Information for personalized responses

2.4.2 Preferred response style

2.5 Summary