1 of 18

AI Dialogue for Business Students

Fraud cases and Human decisions

Professor Mei-Rong Alice Chen

人類

2 of 18

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

3 of 18

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

4 of 18

4

CONCEPT

Why Prompt Design Matters

Same case. Different task design. Different usefulness.

Vague prompt

“Analyze this fraud case.”

Likely problems:

• no role

• no audience

• no evidence boundary

• no required output shape

Designed prompt

“Act as a junior risk analyst. Use only the provided case facts. Identify red flags, unknowns, and next verification steps for the audit manager. Return a three-column table.”

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

5 of 18

5

CONCEPT

Why Prompt Design Matters

Same case. Different task design. Different usefulness.

Vague prompt

“Analyze this fraud case.”

Likely problems:

• no role

• no audience

• no evidence boundary

• no required output shape

Designed prompt

“Act as a junior risk analyst. Use only the provided case facts. Identify red flags, unknowns, and next verification steps for the audit manager. Return a three-column table.”

Better prompting does not guarantee truth. It makes the task easier to inspect.

好的 Prompt 不保證答案正確,但會讓查核更容易。

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

6 of 18

6

AI

✓

DATA

→

DECISION

AI Dialogue for Business Students

From Fraud Cases to human Decisions

20-min guided demo · students fill along

Professor Mei-Rong Alice Chen

  • use fraud-related business cases to learn a practical way to work with AI.
  • design a clearer prompt, inspect the response, preserve uncertainty, and make a human decision about what should be checked next.
  • start with a very simple prompt and see how AI gives different responses when we change the way we ask a question.
  • try it yourself and learn how to include role, context, audience, task, and output requirements in your prompt.
  • be able to have a more effective dialogue on your own with AI.

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

7 of 18

7

ORIENTATION

Today’s Learning Goal and Workshop Flow

Learning goal

Use a structured prompt to analyze a fraud-related business scenario, audit the AI response, and prepare a concise supervisor briefing.

20 min

Guided demo

Watch, then fill the same field

40+ min

Continue

Build, test, and diagnose

Share

Compare

Explain one useful revision

Output

TXT file

Save the full process

Fill along with the demonstration. The worksheet auto-records your role, scenario, terms, prompt, AI output, verification, reflection, and score.

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

8 of 18

8

FILL-ALONG GUIDE

學習單怎麼填

I demonstrate one step → you complete the same field → we compare → you continue.

我示範一個步驟 → 你同步完成相同欄位 → 一起比較 → 再繼續下一步

1

WATCH / 看示範

Do not copy my answer. Watch how I define the task.

不要抄老師答案,觀察我怎麼把任務說清楚。

2

FILL / 填

Complete the same field with your own case.

用自己的案例,立刻完成相同欄位。

3

RUN & CHECK / 測試檢查

Run the prompt and paste the raw AI response.

執行 Prompt,保留並貼上 AI 原始回答。

4

REVISE / 診斷再改

Identify a problem, revise, and keep both versions.

找出問題再修改,前後版本都要保留。

Keep the raw AI output. Do not overwrite it. / AI 原始回答要保留,不要直接覆蓋。

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

9 of 18

9

WORKSHEET STEP 1

Step 1 · Define Your Role and Fraud Scenario

Start with a concrete business role and a bounded scenario. / 先選角色,再把案例範圍說清楚。

Choose a role

• Internal Audit Intern

• Junior Business Analyst

• Business Case Competitor

• Compliance Assistant

Describe the scenario

Include transaction anomalies, amounts, timeline, and relevant employee or vendor actions. Avoid inventing facts that are not in the case.

Name the audience

Examples:

• Senior Audit Manager

• Department Risk Team

• Professor and classmates

​

Audience changes vocabulary, detail, and format.

Example fictional case: duplicate invoices totaling $120,000 were paid to a vendor sharing the procurement manager’s registered address.

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

10 of 18

10

CORE FRAMEWORK

The Five Building Blocks

1

Task Goal

任務

What should AI do?

2

Role

角色

Who should AI act as?

3

Context 情境

4

Output

輸出

Audience, language, format

5

Constraints

限制

What must AI avoid?

A useful prompt is clear, specific, checkable, and revisable.

  • using only the supplied facts
  • marking unverified claims,
  • avoiding conclusions about
  • criminal culpability.

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

11 of 18

11

WORKSHEET STEP 2

Step 2 · Select 3-5 Audit and Fraud Terms

Use terminology to focus the analysis, not to decorate the answer. / 術語是分析工具,不是裝飾。

Red Flag

abnormal indicator

Internal Controls

control mechanism

Segregation of Duties

division and checks

Financial Exposure

potential loss

Dual Approval

two-step approval

Audit Trail

transaction record

Shell Company

paper entity

Whistleblower Protection

internal reporting

Chain of Custody

evidence handling

Rule: choose only terms that help you examine this case. You will be asked to use them explicitly in the prompt.

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

12 of 18

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

13 of 18

13

WORKSHEET STEP 3

Step 3 · Assemble the Investigation Prompt

You are acting as a [ROLE].

Target audience: [AUDIENCE].

[Context & Fraud Incident Scenario]

​

[CASE FACTS]

[Mandatory Audit Terms]

[3–5 SELECTED TERMS]

​

[Constraints]

Use only provided facts. Mark unconfirmed claims “Requires Supervisor Verification.”

​

[Output Format]

[SELECTED FORMAT]

The prompt combines your choices, but you still judge the prompt. / 系統幫你組合,但最後仍由你判斷。

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

14 of 18

14

WORKSHEET STEP 4

Step 4 · Diagnose the AI Response

!

Missing numbers

No concrete estimate of financial exposure

!

Overconfidence

Treats suspicion as established guilt

!

Generic advice

Repeats textbook recommendations

!

No priority

Does not identify urgent next checks

Ask “Can I verify this claim?”

問「我能不能查證?」

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

15 of 18

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

16 of 18

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

17 of 18

17

HUMAN VERIFICATION

Human Decision: Brief the Supervisor

Three briefing points

1. What should the manager check first?

2. What evidence is still missing?

3. What immediate control or follow-up action is justified by the available facts?

Reflect on blind AI adoption

Possible risks include:

• false accusations

• HR or legal disputes

• reputational damage

• incorrect loss estimates

• misplaced operational priorities

AI organizes information. People verify facts, justify action, and take responsibility.

AI 整理資訊;人類查證、判斷並負責。

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen

18 of 18

18

FINISH

Rubric, Score, and Submission

Worksheet completion score

Learner profile + scenario

35 pts

3–5 audit terms

15 pts

Raw AI response

20 pts

Supervisor briefing

15 pts

Critical reflection

15 pts

Before you leave / 離開前

1. Click Evaluate & Score / 計算分數

2. Review missing fields / 補齊欄位

3. Download Text Report (.txt) / 下載 TXT

4. Open and check the full process / 打開檢查

5. Submit the TXT file / 繳交 TXT

Continue your worksheet / 繼續完成學習單

Keep the original AI response. Your revision history is part of the assignment.

原始 AI 回答與修改歷程都要保留。

AI IN CRIMINOLOGY · BUSINESS STUDENTS

Prof. Mei-Rong Alice Chen