Sarah Gravett is a professor of teacher education and development; Dean van der Merwe a senior lecturer, and Lerato Ndabezitha a lecturer in the department of childhood education at the University of Johannesburg.
They recently published an opinion article that first appeared in the Sowetan Live on 01 April 2025
Our goal is to ensure future teachers know how to leverage these tools thoughtfully, critically and ethically so they can integrate them effectively into their own classrooms
The rapid advancement of artificial intelligence (AI) has sparked widespread discussions about its impact on education. In particular, generative AI (GenAI) applications have drawn significant attention.
Millions of people now use GenAI tools in their daily lives, and while the implications for education are enormous, they are not yet fully understood.
As highlighted in Unesco’s Guidance for Generative AI in Education and Research, the growing capability of GenAI to automate aspects of, for example, writing and artwork creation and individualising of learning is reshaping our understanding of learning and teaching.
It challenges educators and policymakers to reconsider fundamental questions: Why do we learn? What do we learn? How do we learn? How should we teach?
These questions demand urgent attention, yet responding thoughtfully is difficult given the rapid acceleration of AI technologies. Recently, Professor Kerry Kennedy, a distinguished visiting professor at the University of Johannesburg (UJ), argued that schools and governments must make proactive decisions about AI in education before technological change overtakes them.
The importance of establishing guidelines for AI use in schools was also raised at the recent DBE (department of basic education) lekgotla, signalling the need for policy-level interventions.
But what about pre-service teacher education? Are future teachers being prepared to engage with the affordances of GenAI? Do they understand how to use these tools ethically and responsibly in their teaching practice?
These questions lie at the heart of an initiative by the Department of Childhood Education at UJ, which began experimenting with ChatGPT in two teacher education courses in 2024 — one for first-year students and another for fourth-years.
The first-year course focuses on preparing pre-service teachers to use guided play as a pedagogical approach, while the fourth-year course helps them refine their skill in incorporating principles from the science of learning into lesson design.
In both courses, we explored whether integrating a large language model like ChatGPT could enhance the learning experience of pre-service teachers — and if so, how.
The intervention was designed as follows: pre-service teachers were given a prompt to design a specific play activity (for first-years) or a lesson component (for fourth-years). They also learnt how to craft effective prompts to generate relevant responses from ChatGPT.
After receiving ChatGPT’s suggestions, they had to critically adapt the activity or lesson component to align with the SA context, curriculum requirements and, in the case of lessons, also the specific lesson design guidelines we use.
Students submitted both versions — the AI-generated activity or lesson component, as well as their adapted version — along with a reflection on the process and their learning experience.
To understand the students’ experiences and inform future work, we collected data through an analysis of a sample of student-generated play activities, lesson plans and interviews.
The findings were revealing: students who used ChatGPT as a consultative tool generally produced more purposeful and thoughtful activities and lessons than those who did not in previous years. Many found ChatGPT useful as a starting point, helping them structure their ideas more effectively.
However, they also noted that the AI’s suggestions were often too generic and not well-suited to the SA context and the lesson design approach that we use.
This required students to engage in critical thinking — analysing the AI-generated content, identifying gaps, and making informed modifications to ensure relevance. In other words, they weren’t simply using AI; they were learning how to work with AI in a way that supported good teaching practice.
Students expressed their intention to use ChatGPT or similar GenAI tools in their future teaching, recognising its value as a “thinking buddy” or a consultative tool. However, some also voiced concerns about over-reliance, warning against using AI to “do your thinking for you.”
These concerns highlight a common challenge: when students start assignments at the last minute, they may be tempted to use ChatGPT as a shortcut rather than as a tool for deeper learning. In such cases, AI can hinder rather than enhance understanding and meaningful practical application.
However, students who engaged thoughtfully with the tool found that, when used correctly, ChatGPT encouraged deeper reflection and critical thinking. It allowed them to produce stronger work while maintaining full ownership of their ideas.
This work is ongoing and will be expanded because we believe that pre-service teachers must be equipped to engage with GenAI tools in an informed and responsible manner.
Ignoring AI in teacher education is not an option — our goal is to ensure future teachers know how to leverage these tools thoughtfully, critically, and ethically so they can integrate them effectively into their own classrooms. If we fail to do this, we risk producing teachers who either fear or misuse AI, rather than those who can harness its potential to support learning.
