Cases in Generative AI Risks Part II

This is the first of a two-class module. This class explores the implications of generative AI in creating fraudulent operational data, focusing specifically on HR records and workforce metrics. Students will examine a synthetic dataset designed to misrepresent workforce size and productivity, gaining hands-on experience in identifying red flags and anomalies. Through data analysis and an interactive dashboard, they’ll detect signs of data fabrication, such as artificial employee ID patterns, unrealistic salary distributions, and implausible position histories. The session culminates in a group discussion on the ethical and strategic implications of AI-manipulated data in business contexts, empowering students with critical data forensics and analytic visualization skills. (Note that we were previously scheduled to have a guest speaker in this class on Generative AI and Governance, but unfortunately our guest speaker was required to work with the client on a recently scheduled workshop).

Why is this important?

As generative AI increasingly enables the rapid creation of highly realistic datasets, the risks of fabricated operational data continue to grow. Understanding and detecting synthetic data manipulation is essential for future business leaders, auditors, and data analysts, who must protect organizations and stakeholders from the financial and reputational harm of data misrepresentation. This class provides students with foundational skills in data integrity assessment and visual analytics, preparing them to navigate the emerging challenges generative AI introduces in verifying workforce and operational data. Through real-world scenarios and practical tools, students learn to uphold transparency and accuracy in business reporting.

Materials:

Case: Pure Oils (see page 5 for example analyses to perform in Alteryx).
No slides are available for this class.
Data: A data update may be required for this class. To ensure your files are the most up-to-date, navigate to ACCTG521_Labs folder and run the command git pull.
Analytics Tools: Alteryx Macros and Tableau/Power Bi for dash boards.
Analytics Tools: Git and GitHub

Review and Extension:

Continuing our focus on Generative AI, we will examine the risks associated with Generative AI relating to data veracity. This is the second of a two-class module which will explore how to identify data veracity red-flags, and visualize that risk using dashboards. This class is related to data created as part of the Pure Oils case covered earlier.

Preparation:

1. There is no formal required preparation for these classes. Reading the Class Plan can aid in understanding the tasks we will be performing on the data.

Class Plan:

Teams: during this class, please sit in your discussion teams.

1. This class builds directly on the prior class.
2. The goal of this class is to work on understanding how to create, edit, and troubleshoot Alteryx Batch Macros.
3. The Class17 Folder contains all of the data from the Chat GPT prompts from Class 7 in ACCTG522. The files will serve as the input files to a macro that is expected to automate the pass/fail of each file based on the analysis of red-flags performed in the prior class.
4. In addition, teams should attempt to visualize their findings, with the goal of being able to demonstrate why the files failed the test, and assess why any files didn't fail.