In my book, ChatGPT for Teachers: Mastering the Skill of Crafting Effective Prompts, I argued that prompting is not an art but a skill—an essential competency for 21st-century educators. Like any other skill, prompting can be honed through iterative practice. Given the rapid spread of generative AI and our increasing reliance on this technology to carry out our work, learning how to write effective prompts has become an urgent necessity.
Conversational chatbots, also known as large language models, operate in a prompt-response manner, and their output is only as good as the input they receive. We can have the same query, but the one who knows how to communicate it in a well-structured prompt gets the best response from the chatbot.
Also, as I stated in my book, chatbots do not understand human language the way we do. For language is a system of tokens related to each other through a complex statistical system driven by probabilistic calculations. This means that while chatbots can predict and generate text based on patterns they’ve learned from vast amounts of data, they lack true comprehension and contextual awareness. They don’t grasp nuances, emotions, or cultural references in the way humans do. Their responses are guided by algorithms and probabilities, not by a deep understanding of meaning or intent.
For instance, when we engage in a conversation, we draw upon a rich tapestry of experiences, emotions, and social cues to interpret and respond appropriately. A chatbot, on the other hand, relies on statistical models to generate likely responses without truly ‘knowing’ what is being discussed. This is why, despite their impressive capabilities, chatbots can sometimes produce answers that are off the mark or lack the subtlety of human interaction.
Prompting Strategies for Teachers
I will talk more about AI limitations in a future, for now let me share with you some of the key prompting strategies I covered in my book. Learning these strategies will enable you to communicate more efficiently with any conversational chatbot be it ChatGPT, Claude, Copilot, Perplexity, you name it.
1. Prompting Strategy: Write Clear Instructions
Clear instructions are critical when interacting with AI, as concise and specific prompts lead to relevant responses. Effective prompts include necessary details without overwhelming the AI, balancing the information provided. Assigning a role to the AI, like “language teacher,” can tailor its responses more closely to the context. Structured, step-by-step instructions enhance clarity, while delimiters help define task boundaries. Using examples guides the AI in the desired response style, and specifying response length ensures the output meets requirements.
Example Prompt:
“Act as a language teacher. Provide a brief explanation of the present perfect tense in English. Include three examples of sentences using the present perfect tense. Ensure the explanation is clear and concise.”
2. Prompting Strategy: Provide Reference Text
Supplying ChatGPT with reference text improves accuracy and performance, especially for tasks requiring common sense and factual correctness. Uploading direct text instead of URLs ensures effective processing. Always verify the data’s accuracy, especially dates, statistics, and citations, as AI can struggle with these details. Provide manageable amounts of text to avoid overloading the system. Include instructions for non-found answers, prompting the AI to indicate when it cannot find the information. Reviewing the source material ensures its relevance and helps evaluate the AI’s response.
Example Prompt:
“Based on the following text, summarize the key points about the benefits of meditation:
‘Research has shown that meditation can reduce stress, improve concentration, and increase feelings of well-being. Regular meditation practice can lower blood pressure, improve sleep quality, and enhance emotional health. Furthermore, it can help manage symptoms of anxiety and depression, promoting overall mental health.’
Summarize the key benefits mentioned.”
3. Prompting Strategy: Split Complex Tasks into Simpler Subtasks
Dividing complex tasks into simpler subtasks reduces errors and increases clarity. Prompt chaining allows the AI to retain context across subtasks within the same session. Structure interactions by steps, creating specific prompts for each subtask and addressing them sequentially. Periodically summarize longer interactions to maintain continuity and retain critical details. Adapt to context limits by breaking down lengthy content into smaller segments and revisiting important points as needed. For comprehensive tasks, generate individual section summaries and compile them for an overview.
Example Prompt:
“First, explain the process of photosynthesis in plants. Then, describe the role of chlorophyll in this process. Finally, summarize how photosynthesis contributes to the oxygen supply on Earth.”
4. Prompting Strategy: Give the Model Time to “Think”
Encouraging detailed reasoning helps ChatGPT provide thorough analysis. Prompt the AI to explain its reasoning step-by-step and deconstruct problems into simpler components. Direct the model to outline relevant principles and walk through its calculations or logical steps. Instruct it to consider potential outcomes and analyze the given information, making logical deductions. Have the AI state any assumptions it makes and summarize key points before delivering a final answer. Comparing and contrasting reasoning with other approaches highlights the robustness of its analysis.
Example Prompt:
“Explain the steps involved in solving a quadratic equation using the quadratic formula. First, outline the formula and what each part represents. Then, walk through an example problem step-by-step, showing all calculations. Lastly, summarize why the quadratic formula is a reliable method for solving quadratic equations.”
Here is a short video I created explaining these prompting strategies:
Related: Free ChatGPT Plagiarism Checker
Final thoughts
Prompting is an important 21st-century skill. As teachers and educators, we need to learn how to effectively craft well-structured prompts so we can make the best of the powerful AI tools at our disposal. By mastering this skill, we enhance our ability to engage with AI in a way that maximizes its potential, improving our educational practices and outcomes. Besides generating accurate and relevant responses, well-crafted prompts also help us navigate the complexities of AI technology, ensuring that it serves as a valuable asset in our teaching toolkit.
