← Back to the Using Artificial Intelligence outline The Course Maker
← Using Artificial Intelligence outline
Week 6 · Lecture outline

Week 6 — Lecture Outline · Simulations & Reusable Prompts

Using Artificial Intelligence · AI 101 Fall 2026 · Prof. Quinn Fictional sample

Course: Using Artificial Intelligence (AI 101) · Silver Oak University (fictional sample) · Prof. Quinn
Objective covered: Objective 2 — Apply effective prompting techniques, including simulations and reusable templates, to accomplish real-world tasks and critically evaluate AI output.
SLOs touched: A (produce high-quality, well-verified results with AI) · B (evaluate, verify, and use AI critically)
Meeting pattern: 2 sessions × 75 min = 150 min. Segment minutes total ~150; scale to your own pattern.

Week at a Glance
The week's big question "How do you use AI to rehearse reality — and how do you keep a simulation from becoming misinformation?"
By the end of the week, students can… (1) identify the four simulation types and match each to its use case; (2) write a simulation prompt with role, goal, and exit condition; (3) explain why AI-generated historical-figure quotes are not verified history and must never be cited; (4) build a reusable prompt template with placeholder variables; (5) catch and name a fabrication or error in a simulation's output.
Key vocabulary simulation prompt, role-play, pre-mortem, decision role-play, adaptive-tutor simulation, exit condition, reusable prompt, placeholder variable, prompt library, generated vs. verified
Materials slides (Deck 6), the week's readings + video links, one approved assistant (ChatGPT / Claude / Gemini / Copilot) for the live demos, the AI-critique moment, and the tutorial
Timing note 8 segments, ~150 min total. Session 1 = Segments 1–4 (~75). Session 2 = Segments 5–8 (~75).

Segment 1 — Hook & the Promise (8 min) · Session 1 opens

Hook. Ask the room: "What's a high-stakes conversation or situation you have coming up that you would really like to practice first?" Take a few examples — a job interview, a negotiation, a hard conversation with a family member. Then: "What if you could do that rehearsal twenty times tonight, from different angles, with feedback after each one?" Pause. "That's Skill 7. And this week we're going to learn to do it well — and to know exactly where it breaks down."

The promise: "By Sunday you'll have designed and run a real simulation, caught something the AI fabricated inside it, and built a reusable prompt template that saves you time on every similar task for the rest of your life."

Why it matters line: "A simulation is a rehearsal space, not a crystal ball — but only if you know the difference."

Segment 2 — What an AI Simulation Is: Four Types (22 min)

Plain language first. A simulation prompt gives the AI a role and a goal, then puts you in a scenario so you can practice or explore before the real thing happens. The AI is playing a character in a space you designed. Four types:

  1. Difficult-customer / difficult-conversation simulation. The AI plays an upset customer, a challenging colleague, or a hard conversation partner. You practice keeping your composure, problem-solving, and communicating clearly. Use it to rehearse situations where real stakes are high (complaint calls, feedback sessions, negotiations).

  2. Pre-mortem simulation. You tell the AI: "Imagine my project has already failed. What went wrong?" This is based on a real planning technique — teams that imagine failure first catch more risks than teams that brainstorm dangers forward. The AI adopts the perspective of a post-failure analyst and reasons backward. Use it before starting any significant project, event, or plan.

  3. Decision role-play / multi-stakeholder simulation. The AI plays multiple characters (a skeptical investor, a community member with concerns, your boss's boss) to stress-test a proposal from every angle. You practice defending your idea and improving it under pressure.

  4. Adaptive-tutor simulation. The AI becomes your personal teacher on a subject, adapting to your pace, checking your understanding before moving on, and explaining things in multiple ways. This is how the Week 6 tutorial is built — you are running this type right now.

Memory hook: "Rehearse before reality. Fail before you start. Hear every voice. Learn at your pace."

Anatomy of a simulation prompt — the three required parts:
- Role: who the AI is playing (specific, not generic — "a hiring manager at a mid-size nonprofit running a behavioral interview for a marketing coordinator" beats "an interviewer").
- Goal: what the simulation is meant to accomplish ("help me practice saying no to scope creep without damaging the relationship").
- Exit condition: when and how the simulation ends ("after five questions, break character and give me feedback").

Segment 3 — The Critical Rule: Generated ≠ Verified (18 min)

The most important concept of the week. Ask the room: "If I prompt an AI to role-play Abraham Lincoln responding to a modern policy question, is the dialogue it generates historically accurate?" Take answers. Then be direct: it is not. It is generated text — the AI interpolates in the style of Lincoln based on training data, but no historian verified those words, no transcript backs them up. Lincoln never said them. They do not exist in any historical record.

Why this matters so much: people screenshot AI-generated historical dialogue, strip the context, and share it as if it were real. This is one of the clearest paths from "useful tool" to "active misinformation." A simulated Benjamin Franklin quote about social media is fiction. Period.

The teaching rule: when working with historical figures in simulations —
- Use them to think through historical context (What issues did this person care about? What was the political tension of their era?)
- Never present generated dialogue as a real quote or a historical source.
- If a student's work includes a "quote" from an AI simulation, it must be clearly labeled: "This is AI-generated simulation dialogue, not a verified historical record."

A quiz item is built around this. Make sure students can state it clearly.

Misconception + cure:
- ❌ "The AI was trained on Lincoln's speeches, so its Lincoln dialogue is accurate."
✅ Cure: training on historical text means the AI can imitate style — it does not mean the specific generated sentences were ever spoken. Imitation is not quotation.

Segment 4 — Misconceptions + Quick Interaction (27 min) · Session 1 closes (~75)

Name the misconceptions out loud, then cure each:

  • ❌ "A simulation outcome is a real prediction."
    ✅ Cure: the simulation is a reasoning exercise, not a forecast. If the pre-mortem surfaces five risks, those are things worth considering, not guaranteed events. The AI is not running a model of the real world; it is generating plausible text about a hypothetical.

  • ❌ "Giving the AI a role makes it actually expert."
    ✅ Cure: telling the AI to "act as a cardiologist" shapes the tone and focus of the output, but it does not make the medical information accurate. The role frames the conversation; it does not confer real expertise. Anything medically, legally, or financially significant still requires a real expert.

  • ❌ "A simulation prompt doesn't need to be specific — the AI will figure it out."
    ✅ Cure: the less specific your role prompt, the more generic the simulation. "Be an interviewer" produces generic interview questions. "Be a hiring manager at a mid-size tech startup running a first-round behavioral interview for a UX designer role" produces something you can actually rehearse with.

Interaction — Classify the Simulation (rapid-fire, ~12 min):
Put four scenarios on slides; students classify as one of the four types, solo (30 sec), compare (1 min), vote. Scenarios: (a) "The AI plays a worried investor grilling my startup pitch" (decision role-play); (b) "The AI tells me what went wrong with my fundraiser after it failed" (pre-mortem); (c) "The AI is an angry customer who received a broken product" (difficult-customer); (d) "The AI teaches me calculus concepts one at a time, checking before continuing" (adaptive tutor). Have them explain the deciding feature each time.

Segment 5 — Reusable Prompts: Your Prompt Library (18 min) · Session 2 opens

Hook back in: "Last session: simulations. Today: how you make this skill compound over time."

The problem. Most people build a great prompt, use it once, and let it vanish. The next time they need something similar, they start from zero. Over a semester — a career — that is a lot of wasted effort.

The solution: a reusable prompt library. When you write a prompt that worked well, save it with placeholder variables so you can fill in the specifics next time. The template becomes a tool that works for every similar task.

Anatomy of a reusable prompt template:
- Task name — a plain-language label (e.g., "Cover letter opener — any job").
- Placeholder variables — [JOB TITLE], [COMPANY TYPE], [AUDIENCE], [TOPIC], [LENGTH] — clear labels for what changes each use.
- Role and goal — who the AI is and what it's producing.
- Constraints or format — e.g., "keep it under 100 words," "ask one question at a time," "format as a table."
- Critique instruction — built in, e.g., "At the end, tell me one thing I could improve."

Live example (show the conversion):

One-off prompt (used once, then lost): "Write me a cover letter opener for the marketing coordinator job at the nonprofit I'm applying to. My biggest relevant experience is running social media for my campus club."

Reusable template: "Write a cover letter opening paragraph for a [JOB TITLE] role at [COMPANY TYPE]. My strongest relevant experience is [ONE SENTENCE ABOUT YOUR BACKGROUND]. Tone: [formal / warm / direct]. Length: two to three sentences."

"That template takes five minutes to build. It works for every cover letter you'll ever write. That is what 'write once, reuse forever' means."

Memory hook: "Save the prompt, fill in the blanks."

Segment 6 — Live Demo: Build and Run a Simulation (20 min)

Set it up: "Watch me design, run, and then audit a simulation from scratch — this is exactly what the Studio asks you to do."

Do it live (narrate every step):

  1. State the goal first: "I want to practice a pre-mortem on a real plan — let's use a student event, like a club fundraiser."
  2. Draft the role and goal: "You are a post-event analyst. I just ran a fundraiser for my student environmental club. Imagine it is three months later and the event was a disaster — we raised nothing, no one came, and the organizing team is frustrated. Your job: reason backward and tell me the five most likely causes of failure. Be specific; avoid generic advice."
  3. Send and observe: note how specific the AI's analysis is — it will name real things (poor timing relative to midterms, no social media promotion, venue issues, no RSVP tracking, competing events).
  4. Ask a follow-up: "Which of these would you say was most preventable?" — model deepening the simulation.
  5. The audit move: "Now, can you cite any specific data or research that supports these risk factors?" Watch for fabricated citations or overly confident claims. If one appears, model catching it: "I asked for sources and it gave me a study title that looks invented — I would verify that before trusting it."

Land the key idea: a simulation is most useful when you design it carefully, deepen it with follow-up questions, and audit it for fabrications before acting on its output.

Segment 7 — Technology Workflow with Verify-the-AI Moment (16 min)

The simulation workflow:
1. Define: choose a simulation type and the specific scenario (be concrete about role, stakes, and context).
2. Draft the prompt: role + goal + exit condition. Include a format request (one at a time; table; after N turns, give feedback).
3. Run and deepen: engage, push back, follow up. Ask the AI to play a different angle or escalate the difficulty.
4. Audit: ask for sources or supporting evidence. Watch for fabricated citations, invented statistics, or historical dialogue presented as factual. Name what you find.
5. Exit and improve: trigger your exit condition, get feedback from the AI (if you built that in), and note what to change in the prompt next time.

Verify-the-AI moment:

Run any simulation that involves a real-world claim — a historical figure, a named research finding, a real case or precedent. Then ask the AI: "What is your source for that specific claim? Are you certain this is accurate?" The AI may double down confidently, or it may walk back. Either way, check the claim yourself before acting on it.
Class prompt: "What would this move look like in a legal mock-trial simulation? A medical case-study simulation? A historical debate simulation?" Have students name the check they'd run in each.

Segment 8 — Callback + Tease + Hand-off (21 min) · Session 2 closes (~75)

Callback:
"Step back and look at the arc: Weeks 3–6 have been one long prompting arc — conversation and emphasis, then structured meta-prompting, then few-shot examples and control, and now simulations and reusable templates. You now have the full Objective 2 toolkit. Every technique from the past four weeks is fair game on the midterm — and they stack. A simulation prompt is stronger when it uses the structured-prompt components from Week 4 and the specificity habits from Week 5."

Tease next week:
"Week 7 is a change of mode. Instead of how to prompt, we look at what you can prompt with. Voice, audio recordings, images, handwritten notes, documents — multimodal AI. We'll record a voice memo, transcribe it with a free tool, and have an assistant summarize it and pull out action items. Bring your phone to Thursday's class."

Hand-off (the week's graded work):
- Lecture Tutorial 6 — simulations and reusable prompts with your AI tutor; share link.
- Quiz 6 (no AI) — four simulation types, reusable template structure, generated-vs-verified rule.
- Discussion 6 — "Are AI simulations genuinely educational, or do they spread convincing misinformation?"
- Assignment 6 — design a simulation, classify types, build a reusable template.
- AI Build Studio 6 — "Run a Simulation" — design, run, audit for fabrications, improve.

Instructor FAQ — Common Stumbles

Student says / does Quick cure
Uses "Act as a job interviewer" and gets generic questions. Too vague. Specify the company type, role, interview style (behavioral vs. technical), and number of questions.
Cites an AI-generated historical quote in an assignment. Not a real quote. Generated dialogue is AI fiction. It must be labeled clearly or removed.
Believes the pre-mortem tells them what will actually happen. Not a prediction. It is a structured exercise in identifying risks — not a forecast.
Thinks assigning the AI a role (e.g., "cardiologist") makes output medically accurate. Role shapes tone, not expertise. Always verify medical/legal/financial claims with real experts.
Builds a great prompt but doesn't save it. Save it with variables. Turn it into a reusable template immediately — five minutes now, unlimited future uses.
Confuses a reusable prompt with a saved conversation. Different things. A prompt library holds templates (with blanks to fill in). A saved conversation is one use case.

Scope flag

This outline stays within Objective 2 at the simulation and reusable-prompts level. The deeper critical-thinking skills — hallucination shapes, full verification workflows, sycophancy — are the Week 10 focus and only reinforced here. The historical-figures rule is introduced here in full because it is the specific fabrication risk of simulation prompts. Real products (ChatGPT, Claude, Gemini, Copilot) are named factually; the instructor and institution are fictional. The buyer-facing verb for the product is "generates."

~ Prof. Quinn's edition · Fall 2026 · built with thecoursemaker.com