4 Prompt engineering and problem formulation
This chapter covers
- Exploring the essentials of prompt engineering
- Techniques for effective AI prompting
- Comparing zero-shot, single-shot, few-shot, and many-shot prompting
- Advanced strategies for optimizing AI responses
- The role of problem formulation in AI interactions
This chapter focuses on two essential skills for working with AI: prompt engineering and problem formulation. Think of these as the building blocks for having productive conversations with AI systems.
Prompt engineering is like learning to speak a new language—it’s about finding the right words and approach to communicate effectively with AI. Just as you may speak differently to a child versus an adult, different AI tasks require different types of prompts. For example, asking AI to write a simple email may need only a brief instruction, whereas getting it to analyze a complex research paper requires more detailed guidance.
Problem formulation, which is often forgotten, actually comes before prompt engineering: it involves clearly defining what you want to achieve. Imagine that you’re developing an enterprise software system. Before writing any code (prompt engineering), you need to understand the business requirements and specifications (problem formulation). If you start coding without clear requirements, you’ll likely run into problems later.