Overview of Prompting Strategies

In our trainings with schools and districts, we do a lot of hands-on work with prompting of GenAI chatbots. We’ve also talked at length regarding why prompting still matters quite a bit as both a practical and educational tool.

From a practical standpoint, learning effective prompting technique leads to more accurate and relevant results, saves time, and minimizes environmental impact. The clearer and more specific the prompt, the faster and more precise the AI’s response.

The four main strategies we outline are:

• ZERO-SHOT: Best for quick, general-purpose responses

• FEW-SHOT: Best for generating specific responses that need to confirm to an established standard

• CHAIN OF THOUGHT: Best for encouraging a more thoughtful and accurate response

• EXPLAIN-THEN-RESPOND: Best for ensuring that foundational understanding is accurate

These prompting strategies - which work with non-reasoning models like Claude Sonnet 3.5 and ChatGPT 4o - are more detailed than most casual users will likely require, but can be helpful to understand, especially for those of you tasked with training others or for adding to your toolbox of strategies when supporting educators with specific queries.

The majority of our prompts in our own Prompt Library are zero- or few-shot prompts, and this table above is by no means exhaustive but designed as a cheat-sheet to understanding some key prompting strategies and what they’re best at.

Want to partner with AI for Education at your school or district? LEARN HOW