Many Institutions now have AI access. That part wasn't the hard part.
The hard part is what happens after. An instructor opens a tool for the first time, types a vague question, gets a generic answer, and walks away thinking: this isn't that useful. They're not wrong. They just haven't learned the skill yet.
The University of Michigan figured this out early. They treat prompt literacy as a real academic competency, not a soft bonus skill. Their framework covers what AI models can and can't do, how to write specific and detailed instructions, how to refine through iteration, and how to weigh the ethical implications of each use. The difference between a typed-casually prompt and a well-crafted one isn't subtle. One hands you an average answer. The other gives you a genuine thinking partner.
That gap is a problem of practice, not access. And it's hard to close one educator at a time.
Where prompting actually shows up in your work
In practice, most institution prompt work falls into four areas.
Teaching and lesson planning. Build lesson scaffolds, explainers, and discussion prompts grounded in your actual course material, not generic examples pulled from somewhere else.
Assessment and feedback. Draft quiz items, generate rubrics, and explore how traditional exams might shift toward project-based work. The model doesn't just write questions; it helps you think through what you're actually assessing.
Administrative and operations. Summarize dense policy texts, draft slide content, and standardize routine communications. These are the tasks that eat hours and reward no one.
Student support. Design interactive tutor blueprints and practice simulators so students can get personalized study help outside of office hours.
Each of these areas benefits from the same thing: a prompt that's specific, contextual, and built to be reused.
The habits behind prompts that actually work
Wharton's Generative AI Labs and the University of Michigan point to the same core habits. Start with a clear goal. Give the model context and a specific persona. Provide step-by-step instructions. Add an example of what good looks like.
Then test it. Run the prompt several times. Try it across different models if you can, because the same prompt can behave differently from one to the next. Iteration isn't a workaround. It's the method.
That process sounds like extra work until you realize you only have to do it once well. After that, you save the prompt.
Why prompt libraries matter more than individual prompts
This is where reusable prompt libraries come in. Wharton describes them as collections of evidence-based prompt templates that staff can save, share, and adapt to their own context. The idea is simple: carry good practice from a few early adopters to the whole campus.
Because the prompts use plain language, anyone can pick one up. Because they're standardized, your outputs stay consistent and high quality. Ethan and Lilach Mollick's widely shared library organizes prompts exactly this way, across instructor aids, student exercises, and broader uses.
A prompt library turns individual expertise into institutional capacity. That's the shift.
Where to start
If you're ready to pilot a shared prompt library, IgniteAI by Instructure is a natural place to begin. The IgniteAI Agent Prompt Library in the Instructure Community gives you ready-made prompts built for the platform so you don't have to start from scratch.
Tools like Prompt Cowboy, help educators turn a rough idea into a clear, well-structured prompt. It's a practical on-ramp, especially for people who know what they want to accomplish but aren't sure how to phrase it yet.
The institutions already building this way aren't doing anything exotic. They picked a few good prompts, shared them, and made iteration the norm. That's it. The barrier is lower than it looks.
Your Customer Success Team is glad to help you get started. Reach out whenever you're ready.
For more inspiration and real-world examples, check out the Prompt Party blog post.
Sources: University of Michigan prompt literacy framework | Wharton Generative AI Labs prompt library | More Useful Things prompt library, Ethan and Lilach Mollick | IgniteAI agent prompt library, Instructure Community
