The generative AI-enabled era of learning design, where gen-AI tools are reshaping the way we teach and learn, has ushered in a new era of possibilities for course design. However, as with any new tool, there is a learning curve, and one aspect of gen-AI that requires careful consideration is how we write effective prompts. So, how can we maximise the efficacy of gen-AI outputs?
Unlocking the power of reusable prompts
When we first start using generative AI tools, we often find ourselves engaged in a back-and-forth conversation, fine-tuning prompts until we elicit the desired response. But using gen-AI through trial and error is a time-consuming process that may yield inconsistent outcomes. There exists a more efficient approach to prompt interaction – one that involves deliberate experimentation with various prompt combinations to create reusable prompts.
- Embracing generative AI as a co-design partner for designing and developing courses
- How can generative AI intersect with Bloom’s taxonomy?
- Collaborating with artificial intelligence? Use your metacognitive skills
Reusable prompts are single, longer prompts that are adaptable across multiple courses and contexts by merely swapping out specific course details while retaining the core prompt structure. In several instances, through developing reusable prompts, we have been able to distil a complex set of 10 or more prompts into a concise, focused command that achieves consistent results. This not only saves invaluable time and effort but streamlines our interactions. It also gives us the opportunity to share prompts and collaboratively refine them.
Creating your own reusable prompts
So, how can you create your own reusable prompts? Here is a process to get you started:
1. Select a use case: Consider the various scenarios in which you might use generative AI. These could range from content creation to assessment design or even administrative tasks. Defining a specific use case will help you craft an effective prompt.
2. Identify core elements: Begin by identifying the essential components that need to be included in your prompts. These elements could include the context, objectives, specific requirements or constraints of the task you want the AI to assist you with.
3. Experiment with variations: Start by crafting a few different prompt variations for a specific task. Experiment with wording, structure and format to see how the AI responds to different inputs. This experimentation phase is crucial for refining your prompts over time.
4. Test and refine: Interact with the AI using your initial prompts and observe the generated outputs. Analyse the results and iteratively refine your prompts based on the AI’s responses. Pay attention to what works well and what doesn’t.
5. Focus on reusability: Test your prompt by using different contextual information. Aim to create prompts that can be adapted for different tasks or courses by making only minor adjustments. Think about creating prompts with a flexible structure that allows you to swap out specific details while maintaining the overall format.
6. Record: Once you have a collection of successful prompts, create a repository or a document where you can easily access and modify these prompts as needed.
7. Share and collaborate: Consider sharing your reusable prompts with colleagues or the broader community. Collaboration can lead to further refinement and the discovery of new use cases for generative AI in learning design.
8. Iterate: Learning design and generative AI capabilities are constantly evolving. Regularly revisit and update your reusable prompts to ensure they remain effective and aligned with current needs and trends.
Once you have developed a few reusable prompts with a consistent structure, try sharing them with a gen-AI and asking it to develop others.
Example of a ready-to-use prompt
To set you on the path of harnessing the power of prompt engineering, here is an example of a ready-to-use prompt for you to experiment with. Simply copy and paste the italicised text below, then replace the content between the square brackets. You might even try this and notice areas where you can refine it.
Design an authentic assessment for [course name]. The authentic assessments should mirror real-world tasks and challenges relevant to the course.
The assessment should align with specific course learning outcomes and engage learners in practical application. The course learning outcomes are [course learning outcome 1], [course learning outcome 2] and [course learning outcome 3]. The level of the course is [paste in the relevant AQF level criteria].
Tailor the assessments to the course’s level and audience, considering the [course topics, content, aims, etc].
Write an explicit and clear set of student instructions for the assessment. These instructions should include the assessment: purpose (how it helps students achieve the course learning outcomes); description (answering the following questions: How does this task relate to other assessment tasks in the course? What product is the learner required to create? What course content is most significant to this assessment task?); instructions (numbered instructions that guide the learner to complete the task. Begin each instruction with an “action” verb. Include the following: Course content or learning activities/resources that the learner should participate in/revise/refer to before beginning the assessment task, specific questions or points which must be addressed in the task, suggestions and/or requirements regarding section headings, referencing and formatting).
Develop a rubric that outlines criteria for evaluating the assessment against [course learning outcome 1], [course learning outcome 2] and [course learning outcome 3].
Reusable prompts are a pivotal technique in using generative AI to support learning design. Through these carefully crafted prompts, we can remove some of the guesswork from gen-AI interactions to develop consistent, quality and replicable results. However, as with all uses of generative AI, reusable prompts are no substitute for human oversight of gen-AI outputs. All outputs of gen-AI should be vetted and refined by human actors.
Richard McInnes is learning designer and product lead, and Ajay Kulkarni is digital educational developer, both at the University of Adelaide, Australia.
If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.
comment