Skip to content
NTPU AI Learn

AI 101 Practice Center

Use four real-world scenario files to practice Chat, Prompt, Cowork, and Code

Practice workflow

Every practice task follows the same five steps. AI output must be checked by you before you submit.

  1. 1DownloadGet the practice file
  2. 2UploadGive the file to Claude
  3. 3PromptWrite the instruction with T-C-R-F
  4. 4VerifyCheck the output against the source
  5. 5SubmitFinalize your result and Use Case

Practice Files

Meeting A: Project Status Meeting

Meeting_A_Project_Status.txt

A single meeting note. Good for Chat, document analysis, Prompt, and Fact / Analysis / Unknown practice.

Download

Meeting B: Cross-Department Follow-up

Meeting_B_Cross_Department_Followup.txt

A cross-department meeting note. Good for multi-document analysis and Cowork delegation.

Download

Sales Report: Sales Data

Sales_Report.csv

CSV sales data. Good for Code / automated analysis and correlation-vs-causation judgment.

Download

Customer Feedback: Raw Excerpts

Customer_Feedback_2026Q3.txt

Raw customer feedback excerpts. Good for classification, evidence citation, sentiment/theme analysis, and Cowork.

Download

Practice Tasks

1

Chat Practice

Use Meeting A to practice organizing a meeting note step by step with Chat.

Uses files: Meeting A: Project Status Meeting

  1. Understand the main content and purpose of this meeting note
  2. Identify 5 key issues
  3. Structure owners and deadlines
  4. Find information gaps the document does not clearly address
  5. Find passages that are inference rather than stated fact
  6. Produce a summary for a manager

2

Prompt Practice

Rewrite the vague instruction "Help me analyze the customer issue" using T-C-R-F.

Uses files: Customer Feedback: Raw Excerpts

  1. Write the Task: what exactly you want AI to do
  2. Write the Context: why this matters, who it is for
  3. Write the Reference: the material you will provide (e.g. the Customer Feedback file)
  4. Write the Format: the output structure you expect
  5. Try your T-C-R-F prompt and compare it to the vague version

3

Cowork Practice

Use all four files to practice delegating a fairly complete task to AI.

Uses files: Meeting A: Project Status Meeting、Meeting B: Cross-Department Follow-up、Sales Report: Sales Data、Customer Feedback: Raw Excerpts

  1. State the task scope and expected delivery format before delegating
  2. Identify 3 main issues
  3. Attach evidence to each issue (quote the source passage)
  4. List open items that still need follow-up
  5. Produce recommendations
  6. Produce a one-page executive summary
  7. Split the overall output into Fact / Analysis / Unknown

4

Code Practice

Use Sales_Report.csv to practice structured data analysis, keeping correlation and causation strictly separate.

Uses files: Sales Report: Sales Data

  1. Find missing values in the data
  2. Find duplicate rows
  3. Compute the completion rate (Actual / Target) by month and region
  4. Find months and regions that missed target
  5. Find the longest customer response time
  6. Find the lowest satisfaction score
  7. Find possible relationships between completion rate, response time, and satisfaction
  8. When describing relationships, do not claim causation. Describe only what is observed

5

Instruction Design Challenge

AI Workforce Foundation required practice: rewrite a vague instruction into a full work instruction with Task, Context, Requirements/Constraints, and Output Format, and compare the AI output under both versions.

Uses files: Customer Feedback: Raw Excerpts

  1. Write down a vague instruction you would normally use, e.g. "help me analyze the customer issues"
  2. Rewrite it into a full instruction covering Task, Context, Requirements/Constraints, and Output Format
  3. In Requirements/Constraints, explicitly require: use only content the Customer Feedback source text supports, and mark anything uncertain as unknown
  4. Run both the vague instruction and the rewritten one against AI
  5. Compare the two outputs: does the rewritten instruction make the output more precise and easier to verify?
  6. Write one sentence on what "padding with words" versus "supplying necessary work conditions" meant in this exercise

6

Output Verification Challenge

AI Workforce Foundation required practice: compare AI output against the original meeting note to find unsourced numbers, inferences beyond the source, and missing limitations -- practicing the verification-before-delivery checklist.

Uses files: Meeting A: Project Status Meeting

  1. Ask AI to produce a summary of Meeting A, including key issues, owners, and deadlines
  2. Compare the summary sentence by sentence against the original meeting note, marking what is supported and what is not
  3. Find at least one unsourced number or date (something in the summary that Meeting A never states)
  4. Find at least one inference beyond the source (an interpretation the AI added on its own)
  5. Find at least one important constraint or open item from the source that the summary left out
  6. Rewrite the summary, changing unsupported parts to unknown or removing them

What next?