If other people can’t understand what you’re asking, the AI probably won’t either.
When modeling prompt-writing, I’ll frame the AI as a knowledgeable colleague or professor: if I wouldn’t ask a human this vaguely, I shouldn’t ask the AI that way either.
The three habits I want to emphasize:
- Clarity — say plainly what you want.
- Context — give the assignment, level, topic, and constraints.
- Specificity — indicate scope, type of answer, and format.
Audience: a few UF colleagues (reference librarians in humanities and social sciences), working in pairs at computer workstations.
1. Quick Examples for Intro (Show & Discuss)
Vague
“Tell me about labor unions.”
Clearer
“I’m writing a short paper about the rise of labor unions in the United States during the early 20th century. Give me a 200–300 word overview highlighting major events, important legislation, and key leaders from 1900–1940.”
Discussion prompt: What information did I add that a colleague would need to help me? (Place, time period, length, focus.)
Vague
“Help me find sources on Shakespeare.”
Clearer
“I’m a first-year English student writing about the political themes in Shakespeare’s Henry V. Suggest scholarly sources that discuss kingship, propaganda, or national identity. Prioritize academic journals or peer-reviewed books, 2010–present if possible.”
Discussion prompt: How does level (first-year), specific work, and time frame change the AI’s answer?
Vague
“Explain gender.”
Clearer
“I need a short explanatory paragraph (undergrad level) summarizing the key concepts in contemporary gender studies — especially gender as a social construct. Use accessible language and cite foundational scholars.”
Discussion prompt: Note how level, scope, and key concept (“social construct”) help the AI.
2. Vague Academic Questions to Use as Starters
These are useful for demonstrating the Golden Rule. I can ask: “What information is the AI missing here?”
- “What’s a good topic for my history paper?”
- “Help me with this statistics assignment.”
- “Explain Foucault.”
- “What are some sources on immigration?”
- “How do I start a literature review?”
- “Tell me about climate change.”
- “What databases should I use?”
Activity: Have participants identify what’s missing: course level, time period, geographic focus, type of assignment, page length, type of sources, etc.
3. Clearer Versions I Can Model Live
Vague
“Help me with this statistics assignment.”
Clearer
“I’m taking an introductory statistics class and I’m struggling with the concept of p-values. Explain what a p-value is and give a simple numerical example. Keep the explanation beginner-level and avoid advanced math.”
Vague
“What are some sources on immigration?”
Clearer
“I’m researching U.S. immigration policy for a 15-page graduate paper. I need 5–7 scholarly sources from 2015–present that discuss economic impacts. Please suggest databases and provide brief annotations for each recommended source.”
Vague
“Explain Foucault.”
Clearer
“I’m writing a 3-page essay on Michel Foucault’s concept of ‘power/knowledge.’ Give me a concise explanation (under 300 words) in plain academic English and include names of two major works where he develops the idea.”
4. Pair Exercise: Prompts for Participants
Give each pair 1–2 prompts and ask them to:
- Identify what’s missing.
- Rewrite with clarity + context + specificity.
- Paste into ChatGPT and compare responses.
Prompts
- “I need help finding sources about the Renaissance.”
- “What are some scholarly articles about homelessness?”
- “How can I research the Harlem Renaissance?”
- “Explain intersectionality.”
- “What is post-colonial theory?”
- “Explain the Enlightenment.”
- “Help me improve my thesis statement.”
- “How do I do a literature review?”
- “How do I analyze a novel?”
After rewriting, have pairs briefly describe how the output changed: more focused, more structured, closer to what they actually needed.
5. Teaching Trick: Human Colleague Test
When a pair shares their improved prompt, I can ask:
“If you handed this prompt to your colleague at the reference desk, would they know exactly what to help you find?”
If the answer is yes, they’ve followed the Golden Rule.
If the answer is no, they revise again.
This helps turn the Golden Rule from an abstract idea into a practical habit: write prompts that would make sense to another human.
6. Optional Framing Lines for My Verbal Intro
- “Think of the AI as a very smart colleague who knows nothing about your assignment until you tell it. If you would never phrase a question this vaguely to a professor or another librarian, don’t phrase it that way to the AI.”
- “Clear prompt = clear answer. Confusing prompt = confusing answer.”
- “If people wouldn’t understand your question, the AI won’t either.”
I can choose one line as my mantra for the session and loop back to it during discussion.