Student Perceptions of Generative AI in Teaching and Learning
As McMaster has taken the approach of having each instructor decide whether and how to incorporate generative AI into a course or assignment, you may be wondering why you might want to do so. What benefits, if any, does generative AI pose for student learning? What learning outcomes could its use support or enhance? This chapter assumes your familiarity with the risks and challenges of generative AI for post-secondary (e.g. academic integrity, assessment design, hallucinations) and imagines what benefits their might be and what opportunities for preparing students for a generative AI supported learning experience.
You can think of the possibilities in two domains:
- supporting personalized learning and
- generating academic content.
Generative AI has many capabilities in supporting personalized learning, some of which we detail below. Chief among them is providing actionable, timely and relevant feedback on drafted student content. This feedback might be focused on the grammar or style of the draft, or on the logic of the argument, organization of the piece, or further examples to consider.
With respect to generating academic content or performing academic skills, you want to think carefully about what the core learning outcomes are for the course, and whether and how students can demonstrate these outcomes. Those skills or knowledge that are not essential to the core learning outcomes might be appropriate for ‘cognitive offloading’ to a generative AI tool. Cognitive offloading refers to the use of external resources or tools to change the information processing requirements of a task so as to reduce cognitive demand.[1]
For instance, if your course learning outcomes require students to demonstrate abilities to generate multiple hypotheses to explain a phenomenon, using generative AI to generates these hypotheses would be inappropriate. However, if your course learning outcomes were focused on having students test a hypothesis it in a laboratory setting, having a generative AI tool generate the hypothesis which the student would then test would be an example of appropriate cognitive offloading.
Supporting Personalized Learning
Invite students to use a generative AI tool to:
Generating Academic Content
Invite students to use a generative AI tool to:
Expand or condense text
(e.g., expand bullet points to actual text) or condense longer text into shorter text (e.g., condense text into bullet points). A related function is to use AI to summarize the key points from a text, including academic articles. Example.
Brainstorm / Generate ideas
AI can be a useful starting point for students working to identify questions, topics, themes or arguments. A generative AI tool can also be asked to provide counter-arguments for a student-generated idea, that the student then needs to account for in their own work.
Find sources or references
This is a capability where you have to be extra careful. As we know, generative AI tools can “hallucinate” sources that do not exist. Generative AI tools that are integrated in search engines generally perform better at this task. Regardless of the tool used, it’s good practice to verify that any sources identified actually exist. Example.
Identify and analyze data
AI tools can analyze different datasets and structure tables with information based on inputted text or data samples with specific parameters offered Example.
Interact with spreadsheets
Generative AI tools like ChatGPT can easily read the .csv format. You can extract a CSV file and give it to ChatGPT to work with based on certain specifications (e.g., give me an overview of what’s in this CSV file and provide some insights into the information provided), as well as output a CSV file. Example.
Code with natural language prompts
Complete partially written code with suggestions, or translate code from one programming language to another. Example.
With all of these uses it’s important to remind students that what the generative AI tool generates may have hallucinations or biases. Students should be reminded to review and evaluate the output from the generative AI tool to ensure its accuracy and evaluate its effectiveness.
You may be wondering – or your students may wonder – what generative AI tool to use for these tasks. This review essay by Ethan Mollick summarizes the capabilities of the major generative AI tools and makes suggestions on the best tool to use for a specific task. You can also visit “There’s an AI for That” to find new generative AI tools for specific educational tasks.
