Olivier Patrick LEFEBVRE

Department of Civil and Environmental Engineering,
College of Design and Engineering (CDE)

Editor’s Note: Olivier ponders the impact of generative artificial intelligence (GenAI) on teaching and learning, particularly its impact on the designing and implementation of authentic assessments and the challenges presented to educators. This reflection follows from what Olivier shared at HECC 2023 under the sub-theme “AI and Education”.

 Image by freepik

LeFebvre, O. P. (2024, July 29). How to make authentic exams in the age of AI? Teaching Connections. https://blog.nus.edu.sg/teachingconnections/2024/07/29/how-to-make-authentic-exams-in-the-age-of-ai/

 

I have had an interest in creating authentic learning environments—defined as teaching in conditions that mimic the real environment in which the students will apply their knowledge (Herrington & Oliver, 2009)–since I attended a workshop on that topic (Soong, 2015). One of the lessons from that session was the importance of setting authentic assessments as one of the nine elements of a true authentic learning environment (Herrington & Oliver, 2009). So, I started doing just that in my final exams, for example by asking students taking my course ESE5205 “Sludge and Solid Waste Management” to find creative solutions to our waste crisis that align with the national agenda on zero waste (Oh, 2019). Such exams proved particularly valuable when the COVID-19 pandemic hit us, as I found these assessments to be very robust when being taken from home as open-book exams with students having full access to the internet. Indeed, the use of the internet was not only a non-issue for me but in fact, I wanted students to browse the web and learn to differentiate relevant and valuable information that pertains to the course from all the incoherent (or sometimes plain wrong) facts that abound online. I also found it a valuable experience for my students to learn to cite their sources and avoid committing plagiarism, the act of claiming someone else’s work as your own.

 

And this is where generative artificial intelligence (GenAI) comes in as the new obstacle. Preventing the use of GenAI tools (e.g. ChatGPT) in teaching and learning feels like a step back, especially in designing authentic assessments. In the real world, we do have access to GenAI tools, so an authentic assessment would have to be designed with considerations of allowing students to access it too. Again, I was confident that GenAI would not have the capacity to answer questions that require higher order thinking skills, such as delivering a critical analysis of a report, interpreting a graph, or solving complex problems. So, I did the test and compared the performance of ChatGPT to that of my ESE2001 “Environmental Challenges in the Anthropocene” and ESE5001 “Environmental Engineering Principles” students in their final exam. Are you wondering what the outcome was? ChatGPT performed like an “A student”. It did not just do very well, it also answered questions much faster than I would have been able to do so.

 

How to solve this issue then? I do not have the answer, especially when we consider the pace at which GenAI is progressing, where the gap between its advances and our capability to even comprehend its implications is growing exponentially (see an illustration in Figure 1).

 

It seems like an impossible race to try and beat AI at its own game, and raising the difficulty of exams in a desperate attempt to try and “safe proof” them would end up leaving a large number of our students behind and feeling discouraged, as they too would likely be unable to solve them.

 

For now, all I can think of is to have a mixed environment. For example, the exam could be divided into two steps:

• Step 1 would be a preparation to the exam, consisting of an analysis of the problem at stake with unlimited access to Internet (and GenAI), but without the exact list of questions to be answered.
• Step 2 would then follow where the specific questions about the problem at stake would be given to the students, which they must now answer on their own, without access to the internet, and in their own words.

 

Meanwhile, I am aware that other teachers are evolving their practice towards making use of AI in their exams. Even though this may sound like the ultimate solution to the problem and definitely authentic, I am concerned that our high-performing, or “A students”, may become the ones that are able to give the best prompts to GenAI (or even worse the ones who can afford the most advanced version of ChatGPT), and are able to optimise making use of AI compared to their peers who may not perform as well academically. Some will argue that the ability to give prompts is the new skill of the future and therefore it is our duty as teachers to train our students for it. One thing is for sure, it is near impossible at this stage for most of us to envision how AI will impact our teaching; nonetheless, it truly is already a revolution that we, as teachers, cannot ignore.

 

References

Herrington, J., Reeves, T. C., & Oliver, R. (2009). A guide to authentic e-learning. Routledge.

Hogarth, I. (2023, April 13). We must slow down the race to God-like AI. FT Magazine. https://www.ft.com/content/03895dc4-a3b7-481e-95cc-336a524f2ac2

Oh, T. (2019, August 30). Singapore aims to cut daily waste sent to landfill by 30% by 2030 to extend Semakau’s lifespan. Today. https://www.todayonline.com/singapore/singapore-aims-cut-daily-waste-sent-landfill-30-2030-extend-semakaus-lifespan

Soong, A. S. K. (2015, November 4). Designing an authentic learning environment [Workshop presentation]. Centre for Development of Teaching and Learning (CDTL), National University of Singapore.

 

Olivier LEFEBVRE is an Associate Professor in the Department of Civil and Environmental Engineering at the College of Design and Engineering (CDE). His research work is in electrochemical technologies and advanced oxidation processes for industrial wastewater treatment and reuse. He is also an adjunct Professor at the Research Institute of Mines and Environment of the University of Quebec in Abitibi-Témiscamingue (UQAT). His teaching and learning interests cover active learning, creating authentic learning environments for his students, and classroom action research. Olivier can be reached at ceelop@nus.edu.sg.