Incorporating generative AI into a module

MediaCentral Widget Placeholderhttps://mediacentral.ucl.ac.uk/Player/bg20bBjH

Watch on MediaCentral
Download the transcript (docx).
Only got a minute? Jump straight to Jenny's top tip

Introduction

Generative AI (GenAI) has been developing rapidly. At the time of writing, GenAI can now even carry out data analysis and generate lengthy reports through Deep Research. I taught a course on energy data analysis for five years and became increasingly concerned that students might rely on these tools instead of going through the analysis process themselves. I updated my module in light of these issues, involving the following steps: 

1. Review module learning outcomes: which of them might GenAI help or harm?  
2. Discuss uses, knowledge and skills with students: including their awareness, expectations and potential uses of GenAI.  
3. Consider where GenAI could or shouldn’t be used: considering why GenAI does or doesn’t make a good fit, and communicating why this is to students.  
4. Encourage responsible use: explore how GenAI can be used as an assistive tool rather than doing the work on the student’s behalf.  

Here’s how I put each of these into action.  

Step 1: Review module learning outcomes  

Going back to the learning outcomes  of the course, which were about understanding statistics and deciding which techniques to apply to what data – but also trying it themselves using code – it seemed like I needed to have some discussions with the students about how GenAI might help or hinder them in achieving all of this.  

Step 2: Discuss with students  

UCL encouraged us to discuss GenAI with our students, and provided some discussion topics. These can be found in the GenAI teaching toolkit, which has slides to help structure discussions with students – I used these slides.   

I had heard that students learn to code these days using GenAI, so I actually encouraged my class to do that, and then asked them what they thought of it. They found it very helpful, because they could put error messages generated by the programming environment into GenAI and it would tell them what was wrong and why and how to correct it, so it saved them quite a lot of time debugging code.  

Step 3: Consider where GenAI could or shouldn’t be used  

On the other hand, when it came to learning statistics, I did not incorporate any AI into this part of the course. I imagined that in the future AI will do analysis and humans will check it. But how can they check it if they do not understand statistics themselves? So the statistics part of the course was left as it was. I showed the students some published examples of where researchers hadn’t properly understood the data before analysing it, and had got the wrong end of the stick as to what was going on. This principle could be used as a cautionary tale about AI, but it was in the course anyway before the widespread use of large language models.  

Step 4: Encourage responsible use  

I also explored getting AI to answer “how would you analyse my data”, and realised that since AI did not understand the context of the data it would not pick out the key features from our perspective. So I made students think about different stakeholders and what metrics and variables they would be interested in. Perhaps AI will do this in the future, but at the moment the students are much better at this.   

It has been an ongoing discussion with the students (even after the course ended), but I think we have broadly concluded that in data analysis you should decide what you want to do and why, and then you can use AI to code this up, and then use your knowledge to check it. Therefore many will be disappointed to know that AI is not a way to avoid learning statistics!  

What’s next?  

If I were to repeat this with another module, I would go through these steps during the planning phase rather than trying to react to GenAI on the fly. I would discuss the effect of GenAI on module learning outcomes with students and colleagues rather than considering it by myself. This would help the students to think about what they want to get from the module, and what skills they want to develop to be able to work with GenAI in the future.  

Jenny's top tip

Involve students as much as you can as you incorporate AI into your course. They will probably know more potential uses of AI than us. Be open to exploring what can be done with AI while being clear about what you want them to learn to do themselves.