In this example, Associate Professor Elizabeth Yuriev encourages students to compare personal responses with AI-generated ones to foster critical thinking, analytical skills, and research skills.
The inspiration behind this innovative approach resulted from the introduction of ChatGPT and considering how this would impact learning. There was an expectation that students would interact with generative AI without awareness of the inaccuracies it provides.
The goal was to demonstrate to students that generative AI, while a helpful tool, can make mistakes. The activity aimed to foster critical thinking by having students analyse and critique responses generated by ChatGPT, encouraging them not to rely blindly on such tools for problem-solving. This approach sought to prepare students for the challenges and risks associated with using generative AI in their academic work.
This task was made possible with the support of Betty Exintaris and Nilushi Karunaratne.
 | Facilitated a workshop focusing on developing problem solving skills rather than unit content. |
 | Began by testing existing chemistry problems and reviewing responses created by generative AI. Ensuring answers given contained inaccuracies or offered limited explanations or details. |
 | Identified suitable generated responses to use with students as part of a critiquing activity. |
 | Students applied problem solving and critical thinking skills to review the generative AI responses to analyse the inaccuracies it produced. |
Students initially found the interactions with generative AI surprising. While initially convinced by the responses, further investigation revealed inaccuracies in the information provided.
Without redirection, especially when students have limited content knowledge, misconceptions could become fact. Teaching staff play a vital role in guiding students on the right path by addressing misconceptions and identifying inaccurate information.
It was also observed that the students' critique skills require further development. Given their first-year status, enhancing these skills will be essential for upcoming tasks.
This exemplar requires a minimum level of effort to implement.
Recommended resources and training:
To set up this activity, create unit questions and generate prompts that answer these questions using ChatGPT. The responses should show inaccuracies for students to evaluate. These are shared with students via a slide deck during a workshop and can be posted as a resource on a Moodle site.
Encourage students to work in groups to critique ChatGPT's responses, focusing on inaccuracies. Working in groups increases students' confidence in using the GenAI tool and encourages discussions towards the limitations and inaccuracy of the responses generated.
Students individually reflect on the received responses and their engagement with ChatGPT, identifying actions and knowledge inaccuracies. Subsequently, they formulate general questions for future problem-solving tasks using ChatGPT.
Consider the content being used and the responses that are generated by ChatGPT. For instance, in generating Chemistry answers, there was little reasoning provided, resulting in inaccuracies. Ensure to test the inaccuracy of generative AI responses, to facilitate student critiques effectively.
Be mindful of potential accuracy differences between free and paid versions of GenAI tools such as ChatGPT. Ongoing developments in GenAI tools may influence the accuracy of the generated content.
Here are some additional resources that you can browse to help you implement this type of teaching practice.
