Prompt Engineering Techniques You Can Teach Students

Help your students become ChatGPT prompt masters, akin to skilled conductors leading a symphony. Teach them the art of crafting AI prompts that enable them to ask the right questions and generate valuable responses. The proficiency your students gain in prompt engineering will shape their AI interactions and outcomes. 

Integrating technology often increases productivity, but your students need to understand the tool on the other side of the conversation.

Unmasking the mystery of ChatGPT

ChatGPT stands for Chat Generative Pre-trained Transformer, a product of OpenAI, it’s a revolutionary AI language model, renowned for generating responses that mimic human conversation in real time. It became the fastest-growing platform in history after its launch, becoming an invaluable tool, with some limitations, in various sectors, including education. 

Benefits

Not only can ChatGPT answer questions and assist with research, but the chatbot can also:

• Assist with research
• Generate code
• Analyze data
• Brainstorm
• Compose draft emails
• Translate languages (similar to Google translate)
• Solve math problems
• Create text and images for presentation slides
  

And that’s only dusting the surface. It can do so much more, and it starts with ChatGPT prompts.

Showcase of AI prompt engineering techniques

Now that you have a basic feel for this helpful little companion chatbot, let’s explore the marvels of prompt engineering techniques. To get the most out of ChatGPT, you can guide the artificial intelligence behind it using prompt techniques.

Self-Consistency

This technique ensures the AI’s response is internally consistent with its generated content. Your students would reference previous responses in their conversation. Guiding the AI to check itself in this way helps it base new responses on material it’s already presented.

Example: “Using the list of words you provided, provide a definition for each term.”

Synonymous phrase variation

This engineering strategy entails using different words or phrases that have similar meanings. Using synonyms in your prompts is useful when seeking diverse perspectives on similar principles.

Example: A Biology student writes, “Explain the concepts of evolution and natural selection.”

Restrictive

You can define boundaries by requesting text in a certain style, voice, or point of view, and you can set parameters for specific time periods, subjects, or word counts.

Example: A U.S. History student writes, “Using a first-person point of view, describe a typical day on a 19th-century farm in the United States.”

* Note: The chatbot can’t pinpoint exact word count requests, so any results from that constraint will be estimated at ranges.

Knowledge generation

Your students can ask the chatbot to provide insights on topics that haven’t been widely discussed. The AI prompt writer will do its best to generate original perspectives on the topic.

Example: A sustainability club member asks, “Can you give me alternative solutions to climate change not covered in the news media?”

Leading questions

Using this technique, students can craft a prompt that nudges the AI model toward the desired output. You’d use this to support an existing viewpoint.

Example: A student in a debate club writes, “Discuss how the problem of deforestation is causing the overall decline of health.”

Multi-turn interaction or chain of thought

This strategy involves interviewing the AI model to build upon previous exchanges, using follow-up questions to resemble a natural conversation.

Example: Students working on a group project for STEM research write, “Describe a futuristic city where advanced technology has transformed the way people live.” After ChatGPT provides an answer, a follow-up prompt might be, “What are the key features and advancements that shape the city?” And then, “What kind of impact do the technological advances have on the city’s architecture?

Zero-shot

Using this technique, ask the model to generate a response despite not having seen similar data during any previous conversation. The answer will be based solely on its preexisting knowledge. 

Few-shot or context setting

In this situation, you guide the chatbot on the desired output.

Example: In a creative writing class, a student could ask this for inspiration, “What kind of story could result from a forest ranger in Yellowstone National Park and a chemical analyst who discovers alarming results while testing a river sample?” 

Priming and seed text

This involves providing a starting point for the generated content. The model will then continue the narrative to match. This is similar to improvisational theater, when an acting coach sets the scene and asks an actor to respond.

Example: In U.S. History, a student might write, “Using the speech style from an Irish settlement near New York City, discuss the impact of the Industrial Revolution on working conditions in the textile industry.”

Variations

Users can also experiment with different phrasings in the AI prompts to explore how the model responds and then refine the prompts accordingly.

Example: “Explain the causes of the American Civil War.”
Alternate prompt 1: “Discuss the primary reasons that led to the American Civil War.”
Alternate prompt 2: “What key figures and regions shaped the events leading up to the American Civil War?”

Persona/Role

Your students can instruct the AI to adopt a particular personality or role to shape the tone and perspective of the output. They can ask ChatGPT to respond as a famous figure, historian, child, or scientist.

Example: In World History class, a student asks, “Assuming you are Mohandas Gandhi, how might you present your wishes to the British authorities?”

Keyword emphasis

Bold or italicize crucial words or phrases within your prompt to guide the AI model to focus on them in the response.

Conditional

Introduce preexisting requirements within the prompt to guide the AI’s response under specific conditions, such as solutions to math problems or hypothetical scenarios.

Example: “If electric cars are better than gasoline-powered cars, how can production and maintenance costs be reduced?”

Least to Most

Generate a list or sequence starting from the least important and progressing to the most significant aspects of a topic. ChatGPT will provide an outline of a methodical solution or a list of essential steps.

Example: An Algebra student asks, “How do I solve a quadratic equation?”

Exclusion or Inclusion Parameters

Explicitly state what should be excluded or included in the generated content to refine the focus of the response and ensure the response stays within a specific area of study.

Example: A Biology student writes, “Explain the effects of pollution on marine life, excluding the impact on land animals.”

Structured Outlining

Providing a structured framework within the prompt, specify the sequence to be followed. This could produce a rough draft for an essay, a blog post, or a newsletter.

Example: “Prepare an outline for a blog post on the top 10 benefits of using AI in the workplace, including an introduction, conclusion and purpose for the post.”

Looking to the future

As prompt engineering is always being refined, utilizing certain ChatGPT prompts will continue to change and improve. Similar to humans, the model learns and grows from each interaction. By applying techniques such as these, your students can learn to expertly and effectively guide AI models to generate content tailored to their assignments or creative endeavors. The result? The output aligns with the specific objectives they wish to explore.

To read more helpful tips for ChatGPT, check out our ChatGPT for Teachers Guide.