Learning about AI in the data science classroom

My plans for Fall 2024

for professors
teaching
Author
Published

June 26, 2024

Background

If you, like me, always feel behind in everything especially when it comes to catching on the newest trends in education, you might also possibly been stressed about catching up on AI in the classroom. To change that I dedicated my Spring quarter catching up on learning about AI in the classroom. I tried to attend every event at the intersection of statistics + AI + education that fit my schedule and tried to read as much as possible. I am not an expert on the use of generative AI but as a learner, I want to share a few resources that I found helpful. In addition I will share my plans for my classes.

Resources

I started the Spring term firstly by using GitHub Copilot and testing it out. A short video by Melissa Van Busen shows the process of using Copilot within RStudio. For those interested a longer video by Tom Mock is also available.

I also read a few manuscripts including

In addition, American Statistical Association hosted a fantastic webinar on Generative AI in Statistics and Data Science Classrooms.

I also tried to attend events on campus. What I really enjoy about these events is that they bring educators from different disciplines together. For instance, my colleague Renée Link teaches in the chemistry department. We absolutely have no connection in terms of the content we teach but seeing examples from her classroom, even if the tools (e.g. feedback generated by AI) are not necessarily available for data science classes, it helps me envision the future of the data science classroom. It has been a perfect way for me to see what faculty in other departments are doing in terms of AI use.

My Plan for Fall 2024

Different educators have different views on generative AI in the classroom. Some are scared and some are relaxed. This is by no means a way to dictate my opinion on others’ classrooms. However, I would like to share what I will be doing in my classroom.

In Fall 2023, I had started the term with the assumption that my students will be using tools like ChatGPT. In my syllabus I had included the following snippet which will mainly stay similar for Fall 2024:

  • Use tools and discuss with people to LEARN. Do not use tools or discuss with people to simply copy and submit the homework.
  • Do not copy as is from the internet or your peers or any other person.
  • Do not pay someone or a website to do the work for you.
  • The work you submit should be reflective of your understanding. In other words, the teaching assistant and I hold the right to question you anytime about your submission. If you cannot answer about, discuss, articulate what you have submitted then we might look further into if the work submitted is not yours or copied and take appropriate action.
  • At the end of each homework, you will be asked whether you have done the homework by yourself, with a peer(s), or used tools like ChatGPT. If using ChatGPT, then you will be asked to submit your prompts and how it supported your learning. If you do not acknowledge any help from a person or ChatGPT, we will consider this cheating.

During the 2023-2024 University of California Irvine has released in-house AI tools including ZotGPT. In Fall 2024, I will be directing my students more to these tools. My main reason for doing that is students’ data privacy. I also plan on including tips on how to write better prompts for ZotGPT and alike. These tips will come together in Fall 2024.

I have always been a big fan of paper-pencil exams completed in-person when the rest of the course is also balanced with different assessment types (e.g., take-home data analysis portions for the exam or projects). I really see the value of the in-person exams, especially in the class following the exams. Students really take the time to review and study the course content. I am also aware of many limitations that come with paper exams as well and write exams with these limitations in mind. For instance, there are many students who are undiagnosed for learning disabilities. Thus, I never write long exams and give ample time for completion.

I have been a big fan of using Gradescope’ AI assistance in image recognition. I will continue using Gradescope when grading exams.

My plan is to introduce GitHub Copilot after the midterm. I would like students to have a basic understanding of R before meeting Copilot. I expect students to use it extensively especially for their final projects that requires an extensive data analysis.

If I can manage the planning in the summer, I will also be using AI assistant that was developed by my colleagues Cristina Videira Lopes and Alberto Krone-Martins. This tool is currently in early stages. In a nutshell, the model is trained by using previous posts from the discussion forums, previous homework assignments, answer keys etc. of the same course. AI assistant then answers students’ questions synchronously without revealing them final answers. The tool itself is new. It is new to me as well. I will see how this goes.

Closing Remarks

It is important to note that AI will be assisting my students (e.g., ZotGPT, AI assistant) and me (e.g., Gradescope). I am also aware that some instructors use tools like ChatGPT to generate their own examples and exams. I have learned this really well from my colleague Zhaoxia Yu’s example on how she asked ChatGPT to write a homework question from one she used the previous year. I have not done this yet but this is something I might also try using and getting better at.

I have a lot to learn. More importantly we have a lot to learn as a community. I am testing tools. I would like my students to have the best learning experience possible. However, I am also precautious on impacts of AI. As I learn more on this, I will try to update readers on this blog.

No matching items