Teaching with AI: Insights and Recommendations

Elizabeth Stanny

Background

BUS 320 - Business Data Analysis and Interpretation

  • Required for Business majors

  • 40 traditional college-age Juniors and Seniors

  • Most students don’t have any programming experience

  • Structure:

    • Mondays: Lecture and examples
    • Wednesdays: Lab assignments

Course Learning Objectives and Tools

Learning Objectives:

  • Determine question want to answer with data analysis
  • Extract, transform and load relevant data (ETL process)
  • Apply appropriate data analytic methods
  • Interpret and present the results

Tools and Platforms:

  • R/Posit Cloud for data analysis
  • Student post assignments to their blogs

Challenge

  • Students struggle with coding errors
  • Error messages are not always helpful
  • Time-consuming to diagnose and correct errors

Lecture on AI in Business Data Analysis

  • Role of AI in Business Data Analysis: Data analysis, predictive modeling, decision-making
  • Benefits of AI: Automation, efficiency
  • Challenges of AI: Hallucinations, Bias, ethics, privacy
  • Examples: Using ChatGPT to diagnose coding errors

Lab Activity and Student Feedback

  • Wednesday Lab: Diagnosing coding errors with ChatGPT
    • Students used ChatGPT to identify errors in provided code chunks
    • ChatGPT provided a response, and students corrected the error and re-run the code
    • Discussion on the effectiveness of ChatGPT
  • Student Feedback:
    • Found ChatGPT helpful in diagnosing errors
    • Provided detailed explanations, and suggestions were mostly helpful

Student Feedback

Overwhelmingly students found the response from ChatGPT to be more helpful than the error messages from R

I found ChatGPT to be incredibly helpful to diagnose errors in codes. The error message tells you where the error is however it doesn’t show you how to fix it. Sometimes you are given a recommendation, however not every time. ChatGPT shows you what is wrong and gives you instructions to fix your code. On one of the instructions it wasn’t as clear as I wanted it to be so I asked ChatGPT to explain further, which it did. This tool is very helpful when trying to diagnose problems and further understand why they happen.

Mixed responses on how likely use ChatGPT in programming

On a scale of one to five, I would give it a solid four. It is a very helpful tool in assisting with programming. I would give it a 5 out of 5 if I was not wary about the technology and the problems that can come from it such as its accuracy, but nonetheless, it is a helpful tool.

Yes and no. It was nice to have more flushed out feedback than a simple error message. But, due to some confusion on part of the chatbot and unclear/misleading responses, I would not trust my life or grade to this machine.

Reflections

  • What to keep:
    • Continue using ChatGPT to diagnose coding errors
    • Incorporate more AI tools in the course
    • Provide more opportunities for students to use AI tools
  • What to change:
    • Address accuracy of responses
    • Provide more guidance on using AI tools
    • Integrate AI tools with course content

Recommendations

  • Incorporate AI tools early in the semester
  • Provide opportunities for students to use AI tools
  • Discuss benefits and challenges of using AI tools
  • Encourage students to use AI to enhance learning

Thank you!