Overview

In this use case category, you’ll find examples of how GenAI can assist in the creation and understanding of IT products. These examples are directly related to the tasks we assign to our students in the ICT&Business curriculum, specifically in semesters 2, 3, 6, and 8. Most of these examples focus on exploratory data analysis assignments, presented from a student’s perspective to demonstrate how they can leverage GenAI in their coursework.

Example 1: SQL Query Explanations with GitHub Copilot

Our first example showcases a solution with explanatory comment generated by GitHub Copilot for a SQL assignment.
GitHub Copilot is one of the most popular GenAI tools for coding support, seamlessly integrated with VSCode, Azure Data Studio, and other IDEs like JetBrains. I primarily use Copilot in VSCode for SQL and Python programming, though I haven’t yet tried it for R programming. The tight integration with VSCode poses a challenge for showcasing Copilot on this static website.

Despite the ease with which our ICT&Business semester 2 students can solve the 70+ SQL queries using GenAI, it’s crucial to ensure they truly understand SQL. Simply submitting a solution doesn’t prove learning outcomes. Therefore, we assess students’ ability to reproduce and explain results or solve similar problems in face-to-face sessions, such as FeedPulse conversations. Fortunately, Copilot excels at explaining code, including SQL, and can quiz users on their understanding within the coding environment.

Example 2: PowerQuery solution as an Anti-Pattern

The second example, PowerQuery solution, serves as an anti-pattern for using GenAI from a student’s perspective. This exercise was particularly insightful for me because I already knew the solution. Without prior knowledge of the tool (PowerQuery) and the domain (data engineering), it would be nearly impossible for students to arrive at a working solution. This example highlights the importance of properly breaking down problems for GenAI and acknowledges its current limitations with PowerQuery and its programming language, M.

In addition to creating a PowerQuery solution, this example involved using GenAI to generate documentation from development notes. The resulting essay comments on the process of using GenAI and could similarly be used to produce technical documentation. However, the essay distinctly feels GenAI-generated. Personally, when I receive such a document, I’m inclined to summarize it with GenAI before investing significant time in it. Over time, GenAI will likely produce texts with a more personal touch, but this example isn’t quite there yet.

Example 3: Data Model and SQL Scripts for a Real Estate Agency

The third example involves creating a data model and SQL create table scripts from a real estate agency case description. This task was easier for GenAI to handle, likely because it has encountered many more data models and SQL create table scripts than PowerQuery solutions. While the abstract complexity of this problem is comparable to the first example, GenAI’s familiarity with similar tasks resulted in a better solution.

Example 4: Creating proof for a learning outcome

The fourth example is about creating a professional product (a presentation) to serve as evidence of a learning outcome. I think this is an exercise that can really go two ways: on the one hand, it shows how easy and effortless it is to produce impressive, good-looking evidence of learning outcomes and leave it at that, and on the other hand, how it can serve as a spark to ignite enthusiasm and ownership of learning about a particular topic.