Exam Task

You’ve nearly arrived in the present day. How you make it from there to here is up to you.

After submitting 44 Media Logs to each unit, you should find your character’s story not quite having made it to today. We still have over 20 years to cover! Now that you have some experience with the types of writing that the LLM generates, it’s your turn to start creating some parameters for the rest of the story.

Designing LLM Prompts

If you have been using LLMs throughout the course to do research, you’ve probably just been typing in prompts. That’s a great way to get results. If you want to enhance your ability to get the results you want, you can use a strategy called “Retrieval-Augmented Generation”.

Retrieval Augmented Generation (RAG)

Think of Retrieval-Augmented Generation (RAG) as giving the AI an "open-book test." Instead of asking the AI to write a story based only on what it already knows (which can lead to historical inaccuracies or "hallucinations"), you are providing it with a Knowledge Base of specific facts you have researched.

By "grounding" the AI in your RAG data, you ensure that your character’s life is shaped by real historical statistics, like the actual cost of rent in Vancouver in 2010 or the specific impact of the 2008 recession.

How to Build Your RAG Data

Your RAG data should be a clear, organized list of the facts you find in Step 2 (below) You can format it as a simple list like this:

My Character's Context (RAG Data):

• Year: 2008
• Location: Windsor, Ontario
• Economic Event: The automotive industry crisis; local unemployment rises to 15%.
• Personal Finance: Character’s assembly line job is at risk; median income for this sector drops to $42,000.
• Social Trend: Rise of Facebook as a primary way to stay in touch with laid-off coworkers.

Integrating RAG into Your Prompt

When you move to Step 3, your prompt should act as a bridge between your research and the story. A strong RAG prompt follows this structure:

1. The Persona: Tell the AI who it is (e.g., "You are a historical biographer").
2. The Context (The RAG Data): Paste your researched facts and statistics.
3. The Instruction: Tell the AI how to use those facts (e.g., "Write a 500-word story entry where the character reacts to the economic data provided above").

Steps

1. Identify Historical Trends Choose three distinct lenses (e.g., economic, social, technological) through which to view Canadian history from 2002 to the present. Identify the major shifts that would have impacted a person living through this era.

2. Gather Your RAG Data (The Knowledge Base) Research specific facts and statistics for your story. This is your Retrieval-Augmented Generation (RAG) data—the "truth" you will provide to the AI so it doesn't have to guess.

   • Example: If your character is a student in 2010, find the average tuition cost or the top-selling song of that year from Statistics Canada or digital archives.

   • Deliverable: A list of at least 5-10 key data points per decade.

3. Design Your "Grounded" Prompt Construct a prompt that forces the AI to use your Step 2 data. A "grounded" prompt ensures the AI stays historically accurate. Your prompt should include:

   • A Persona: Tell the AI who it is (e.g., "You are a memoirist writing about the immigrant experience in Ontario").

   • The Data: Explicitly paste your RAG data into the prompt field.

   • The Task: Instruct the AI to write the story only using the facts provided to ensure historical "continuity and change."

4. Generate and Refine the Story Run your prompt through the LLM. If the output feels too generic, go back to Step 3 and add "Style Instructions" (e.g., "Write in a nostalgic tone" or "Use short, diary-like entries").

5. Create the Multimedia Scrapbook Craft an end to your story, and support it by building a visual/auditory "Scrapbook" (Google Doc/Slides).

   • Curated Content: Include real historical photos or headlines from the era.

   • Generated Content: You may use AI image generators to create "personal" artifacts, such as a portrait of your character or a photo of their fictional childhood home.

6. Works Cited Provide a formal Reference List for the data gathered in Step 2. This proves the "Knowledge" and "Thinking" categories of your rubric by showing where your "truth" came from.

Tips

• Check the "Vibe": Does the AI-generated story sound like a real person? If it sounds too much like a textbook, go back to your prompt and add a "Tone" instruction (e.g., "Write this in the style of an emotional letter to a younger sibling").
• Verify the Facts: If the AI mentions a specific event (like the 2010 Vancouver Olympics), double-check that it matches the data you provided in your Knowledge Base. * The "Why" Matters: In your scrapbook, make sure it’s clear why you chose certain images or statistics. Historical significance is about making choices—be prepared to defend why you chose your specific trends.
• Cite Your Work: Even though the AI "wrote" the prose, the facts belong to the researchers at Statistics Canada and the archives you used. Ensure your Works Cited is polished.