AI Update for Educators #6

Mastering Effective Prompting: Unlocking AI's Potential in Education

As educators, many of you are now familiar with large language model generative AI tools like ChatGPT, Claude, and Google Gemini. Despite this familiarity, crafting prompts that consistently produce high-quality outputs can be challenging. If you've found yourself frustrated by inconsistent or irrelevant results, you're not alone. The key to unlocking AI's full potential in education lies in mastering the art of effective prompting. This newsletter will break down the process of writing effective prompts to enable you to use generative AI tool to create quality articles, lesson plans or any other content.

Effective prompting is a dialogue, not a one-time command. Be prepared to invest time in guiding and refining your AI assistant's work.

Why "One-Shot" Prompts Often Miss the Mark

There's a common misconception that there's a universal set of prompts that work for every situation. However, AI can't read minds or infer the nuances of your specific educational context from a vague prompt. Large Language Models (LLMs) need your guidance to generate responses tailored to your unique teaching needs. This means you need to devote time to carefully unpacking your thinking throughout the prompting process.

Treating AI Like an Intern: A Practical Approach

Think of AI as a bright but inexperienced intern joining your teaching team. You wouldn't expect a new intern to understand the intricacies of your classroom, curriculum, or teaching style on their first day. Similarly, AI needs context, steering, direction, and feedback to perform effectively.

Let's break down a reliable method for guiding your AI "intern" to produce high-quality, relevant outputs. We call this the "3 C's" of effective prompting:

1. Contextualise

Just as you'd orient a new intern, you need to set the stage for your AI assistant. This involves providing initial context and guidelines.

Example of Effective Contextualisation:

Objective/Purpose:

• Clearly define the learning outcomes for the module.
• Example: "You are designing a learning module for 2nd year environmental science students. The aim of the module is for students to understand the principles of sustainable development and be able to apply them to real-world scenarios. Ensure the content is suitable for diverse student needs and learning styles. Present the material in a structured format with clear sections."

Key Message:

• Emphasise the importance and urgency of sustainability in today's world.
• Example: "The module will communicate the message that sustainability is essential for ensuring the well-being of future generations and the health of our planet."

Target Audience and prior understanding:

• Example: "Tailor content to 2nd year undergraduate environmental science students. The students have a foundational understanding of environmental science concepts but are looking to deepen their knowledge of sustainability."

Audience Needs and Concerns:

• Address the specific concerns and interests of the students.
• Example: "The task should focus on practical applications of sustainability in various industries and the potential career opportunities in this field."

Call to Action:

• Encourage active participation and further exploration of the topic.
• Example: "Students should engage in a group project to develop a sustainability plan for a local community or business, and present your findings to the group."

Check the LLM’s understanding

• Ask the LLM if it has further questions it would like answered about the task
• Example: "Please let me know if you need any additional information or have any questions before I provide the detailed instructions for the module."

This contextualisation is effective because it outlines:

• The task and task objectives/purpose
• The wider context for the task including target outcomes and audience
• Key messages to be delivered
• The student’s needs and existing knowledge
• A clear call to action

The LLM is also encouraged to ask questions to confirm its understanding of the context before it is given the detailed task instructions.

2. Construct

Once the LLM has understood the context, you can provide the detailed instructions for the task. This step involves clearly stating your requirements and defining what you want the AI to do.

Elements of Effective Construction:

1. Remind the LLM of the role it is performing in this task
2. Define the task: Describe the specific action you want the AI to perform.
3. Provide the detailed task requirements: language, tone, word length etc.
4. Specify the format: Indicate the desired structure or layout of the response.
5. Provide details: outline any additional considerations or constraints.

Example Construction:

"Act as an experienced curriculum developer for undergraduate courses. Your task is to create a comprehensive 60-minute sustainability module for 2nd year environmental science students. The module should include a case study analysis and be aligned with current higher education standards.

Objective/Purpose:

• The aim of the module is for students to understand the principles of sustainable development and be able to apply them to real-world scenarios. Ensure the content is suitable for diverse student needs and learning styles. Present the material in a structured format with clear sections.

Key Message:

• The module will communicate the message that sustainability is essential for ensuring the well-being of future generations and the health of our planet.

Target Audience and Prior Understanding:

• Tailor content to 2nd year undergraduate environmental science students. The students have a foundational understanding of environmental science concepts but are looking to deepen their knowledge of sustainability.

Audience Needs and Concerns:

• The module should include a section on practical applications of sustainability in various industries and the potential career opportunities in this field.

Call to Action:

• Students should engage in a group project to develop a sustainability plan for a local community or business, and present your findings to the group.

Task Requirements:

• Language: Formal and concise, using Australian spelling conventions.
• Tone: Professional and instructional.
• Word Length: Approximately 500 words.

Format:

1. Learning Objectives: Clearly define what students should achieve by the end of the module.
2. Materials Needed: List any resources or materials required for the module.
3. Introduction: Provide a brief overview of the topic and its relevance.
4. Main Activities: Detail the activities, including the case study analysis, that students will engage in.
5. Conclusion: Summarise the key takeaways and any follow-up actions or discussions.

Additional Considerations:

• Ensure the module is engaging and encourages active participation.
• Incorporate practical examples and real-world applications to enhance understanding.
• Consider any potential challenges students might face and suggest solutions."

3. Critique

The final stage involves reviewing the AI's work and providing feedback, just as you would with an intern's initial efforts.

Steps for Effective Critique:

1. Review: Carefully examine the initial output for accuracy, relevance, and alignment with your requirements.
2. Provide feedback: Offer specific suggestions for improvement or areas that need clarification.
3. Request revisions: Ask the AI to make targeted changes based on your feedback.
4. Iterate: Repeat this process until you're satisfied with the result.

Example Critique:

"Great start! Let's refine the module:

1. Include more interdisciplinary connections, linking sustainability to economic and social factors.
2. Focus the case study on a specific industry or region to make it more impactful.
3. Add a brief section on adapting the module for online or blended learning environments.
4. Please revise the module with these changes in mind, maintaining the overall 60-minute timeframe."

Advanced Tips for Power Users

As you become more comfortable with AI prompting, you could also add the following step:

1. Cross-platform verification: Use a different LLM to evaluate your initial output. For example, if you used Claude 3.5 for the original task, paste the prompt and output into ChatGPT4o and ask for a detailed evaluation.
2. Iterative refinement: After making revisions based on feedback, ask the original LLM whether the changes have improved the output and why. This helps ensure the final product meets your expectations.

Key Takeaways for Educators

1. Patience Pays Off: Effective prompting is a dialogue, not a one-time command. Be prepared to invest time in guiding and refining your AI assistant's work.
2. Be Specific: The more details you provide about your educational context, learning objectives, and desired outcomes, the better the results you'll achieve.
3. Iterate and Refine: Don't hesitate to provide feedback and request revisions. Each iteration is an opportunity to improve the output and tailor it to your specific needs.
4. Think Like a Mentor: You're guiding the AI, just as you would guide a new intern. Use your pedagogical expertise to steer the AI towards best practices in education.
5. Embrace the Learning Process: Mastering effective prompting takes practice. Don't get discouraged if your first attempts aren't perfect – each interaction is an opportunity to improve your skills.

The Path Forward

AI in education isn't about replacing educators – it's about augmenting your expertise and expanding your toolkit. By mastering effective prompting, you'll unlock powerful tools to:

• Personalise learning materials for diverse student needs
• Generate creative ideas for lessons, assessments, and projects
• Streamline administrative tasks and lesson planning
• Offer additional support and resources to learners
• Enhance professional development and lifelong learning opportunities.

Your Action Item: This week, try the "3 C's" approach (Contextualise, Construct, Critique) with one of your upcoming lessons, assessments, or professional development activities. If you're feeling ambitious, experiment with the advanced tips as well. Share your experience in the comments – what worked well, and where did you face challenges? How did the quality of the AI's output change as you refined your prompting technique?

Remember, every educator's journey with AI is unique. Keep experimenting, stay curious, and don't hesitate to reach out to the community for support. For a general overview of generative AI in education, see my interactive Perplexity blog on Generative AI for teachers . Subscribe to our AI in Education newsletter to stay informed and inspired as we continue to explore the potential of AI in enhancing teaching and learning across all educational settings.