Innovative Data Sources and Tools: Generative AI

Description

In this class, students will explore the role of Generative AI, particularly Large Language Models (LLMs) like GPT, in data analytics. The session will begin with an introduction to the practical applications of GPTs in automating initial analyses and improving workflow efficiency. Using a case-based approach, students will interact with GPT models to generate a first-pass analysis of the case requirements, saving the output for refinement. The class will emphasize how GPTs can assist in building Python code that meets specific analytical needs, while also showcasing how iterative dialogue with LLMs can improve the quality of the solution.

Motivation

As generative AI continues to transform various industries, understanding how to effectively leverage these tools is essential for accounting and data professionals. By learning how to use GPTs and other LLMs, students will gain a competitive advantage in automating routine tasks, accelerating problem-solving, and enhancing data-driven decision-making processes. This class equips students with the ability to integrate cutting-edge AI technologies into their analytical workflows, which is increasingly valuable in fields such as forensic accounting, audit, and advisory services. The hands-on experience also prepares students to navigate the ethical and practical challenges posed by these powerful tools.

Review

This class builds on the foundation of previous sessions by expanding the toolkit available for data extraction and analysis. While earlier classes focused on more traditional ETL processes and textual analysis methods, this session introduces an innovative approach by integrating GPTs into the analytical process. Students will apply their knowledge of Python and Alteryx in combination with AI-driven solutions, pushing the boundaries of conventional data analysis techniques. This session's case studies and exercises will further develop their skills in leveraging advanced technologies to automate and streamline complex data workflows.

Preparation

• The background reading and case can either be read in advance, or used as a reference.
• The case has been updated from a draft version, but the content is still largely the same, no need to re-read it, but we will use it as a reference in class.

Class plan

• We will start with a short discussion of GPTs and a practical way of using them in data analytics.
• We will first ask the GPT to provide a first pass at the requirements for the case. The output will be saved as the first attempt.
• In discussion teams we will examine the GPT data and look for indicators, or red flags, that the data is fake/AI generated and make a list of these indicators (this list will be the deliverable for this class).
• We will then refine our approach to the case by interacting with the GPT in building python code to execute the requirements.

Team format

discussion teams

Tools

• Chat GPT (or another LLM), Python, Visual Studio Code, Alteryx