MCRPA · A 60-minute thinking workshop
AI That Works for You.
A 60-minute thinking workshop. Use AI to improve the process around your thinking. Not to replace it.
Facilitated by Daniel Umana, MSEd·Rockville Campus, Montgomery College
The Main Rule
Do not use AI to replace your thinking. Use AI to improve the process around your thinking.
Use AI for
- Clarifying ideas
- Organizing
- Learning
- Comparing options
- Building outlines and checklists
- Generating questions
- Checking assumptions
- Practicing communication
Do not use AI as
- The final decision-maker
- A replacement for professional judgment
- A place for sensitive data
- An accuracy guarantee
- Your voice without review
This is not official Montgomery College training.
The MC Generative AI Guidelines and Use Cases are published by MC's Office of Information Technology. This session translates that guidance into plain language so it is easier to use.
Always review the official policy. Direct policy questions to the IT Service Desk.
Official source: info.montgomerycollege.edu GenAI Guidelines and Use Cases
Applicable MC policies: 76001 Acceptable Use of Information Technology · 76002 Confidential Data Management and Security · 70001 Responsible Use of Emerging Technologies.
Sources: Digital Promise AI Literacy Framework · UNESCO AI Competency Framework · NIST AI RMF Generative AI Profile (AI 600-1)
MC Guardrails
The Guardrails.
This is not official Montgomery College training.
The MC Generative AI Guidelines and Use Cases are published by MC's Office of Information Technology. This session translates that guidance into plain language so it is easier to use.
Always review the official policy. Direct policy questions to the IT Service Desk.
Official source: info.montgomerycollege.edu GenAI Guidelines and Use Cases
Applicable MC policies: 76001 Acceptable Use of Information Technology · 76002 Confidential Data Management and Security · 70001 Responsible Use of Emerging Technologies.
Three tabs: which data is safe to use, which tools are approved, and the hard nos. Use the arrow keys to move between them.
Data approved for distribution to the public without restriction.
Free, non-enterprise, AND enterprise tools.
Internal-use data not protected by law. Includes employee, academic, planning, facility, emergency, or administrative data restricted for safety, competition, or development reasons.
Enterprise tools only, with additional review and approval.
Protected by federal, state, or local law, College Policy or Procedure, professional code, contract, or other binding mechanism. Includes PII. Identity-theft and fraud risk.
NO AI tool. Not free. Not paid. Not enterprise.
The Room Rule
Today we use public, fictional, or sanitized content only. Do not paste student records, employee issues, HR content, protected information, private emails, medical information, or any confidential college data into AI tools during this workshop.
Source: MC Policy 76002: Confidential Data Management and Security · MC Generative AI Guidelines
Your voice, your judgment
AI as assistant. Not author.
AI can clean up your message. It should not erase your judgment, personality, or responsibility.
If it sounds like AI
You have leaned on it too much. Read it again and put your own words back in.
If it reads like your own clear writing
You reviewed it and it carries your judgment. That is the goal.
A prompt cannot give the writing your voice. Your review and your edits do that. The prompts below are review moves: they keep you reading the output critically instead of handing the work over.
Three review prompts to keep you in control
Copy one. Use it when you review what AI writes.See every change
Improve only clarity, flow, and organization. Keep my main points and my wording where it already works. After revising, list exactly what you changed and why, so I can review each edit.
Surface the gaps first
Before you revise anything, point out where my draft is unclear or where you would have to guess. Let me answer before you change a word.
Use it as a thinking partner
I am thinking through an idea. Do not agree with me automatically. First, summarize my idea in your own words. Then identify what is strong, what is unclear, what assumptions I am making, what questions someone skeptical might ask, and what I should think through next.
The builder
The Five-Part Scaffold.
Five blanks that turn a vague ask into a clear one.
Trash in, trash out
Same tool. The prompt decides what you get back.
write an email about the parking changes
- Dear Team, I hope this message finds you well.
- I am writing to inform you of some important updates regarding parking.
- Please do not hesitate to reach out with any questions. Best regards.
Generic and padded. You still have to rewrite the whole thing.
Worked examples
Real prompts, built out
Real situations, real prompts. Each one is self-contained: copy the full prompt, paste it into Copilot, and see the effect for yourself. Hover any part to see why it is there.
What they were trying to do
You are running the monthly team meeting and have a jumble of topics in your notes. You want a clean agenda with time blocks so the meeting stays on track.
The prompt they built
Hover or tap a colored part to see why it is there.
Each color is one part of the Five-Part Scaffold. Together they turn a vague ask into a prompt that gets something usable. Hover a part to see the thinking behind it.
What it gave back
Copy the whole prompt into Copilot and you get a tight, ordered agenda from your own topics, ready for the invite.
Your turn
Now build your own
Fill the five blanks. Your prompt assembles live on the right.
Your assembled prompt
Role/Goal: You are a [role]. OR I want to [goal]. Context: [Situation or background.] Source: [What AI should draw from.] Task: [What you want it to do.] Format: [How the output should look.]
Source: Microsoft Learn: Get better results with Microsoft 365 Copilot Chat. Write effective prompts.
12 add-ons
12 add-ons that change what you get back.
Bolt these onto the scaffold to change what AI does.
Ask me up to 5 questions before answering.
Use when: Task has many right answers depending on context.
Do not draft yet. Tell me what is unclear.
Use when: You want diagnosis before production.
Give me 3 options with tradeoffs.
Use when: You are making a decision.
Preserve my voice. Improve only clarity.
Use when: You are revising your own writing.
Reverse engineer the structure of this example.
Use when: You want a reusable template.
Create a checklist from this process.
Use when: You want it to be actionable.
Challenge my assumptions before improving this.
Use when: You want pressure-testing.
Label what you know, assume, and need from me.
Use when: You want honesty about confidence.
Use only the source I provide. No outside info.
Use when: You want grounded, no-hallucination output.
Beginner version and expert version.
Use when: You need to adjust complexity.
Make this shorter without removing important detail.
Use when: Output is too long.
List possible errors or risks before finalizing.
Use when: Stakes are high.
Source: Microsoft Learn · Google Gemini developer guidance · NIST AI RMF 600-1
Your assembled prompt
Role/Goal: You are a [role]. OR I want to [goal]. Context: [Situation or background.] Source: [What AI should draw from.] Task: [What you want it to do.] Format: [How the output should look.]