Image: Nahrizul Kadri/ UnsplashThe use of artificial intelligence in tertiary education is often a much-discussed topic, but very little is taught about its responsible use.
At the University of the Western Cape (UWC), the conversation around artificial intelligence is not about whether students should use tools like ChatGPT, but rather how they should use them. The focus is on a future where AI supports learning without replacing the critical thinking and discipline-specific reasoning that higher education demands.
In an attempt to tackle the challenge of the lack of teaching of responsible AI use, Prof Derek Yu, from UWC’s Department of Economics, used AI tools to compile an AI writing workshop exercise series. The series takes the student on an educational research journey covering research idea generation, ways to search for literature, writing summary paragraphs and designing a primary survey questionnaire.
Prof Yu recently reflected on this challenge by comparing the work of two fictional honours students, which he based on real-life student approaches to research. Their assignments, similar in scope, revealed stark differences in outcomes depending on how AI was used.
The first student leaned heavily on ChatGPT, submitting an outline generated entirely by the tool. While the structure looked polished, it lacked academic depth: key concepts were misplaced, foundational theories were omitted, and data choices were weak. His work earned a modest 55%, underscoring the risks of relying on AI as a shortcut.
Image: Emiliano Vittoriosi/UnsplashIn contrast, the second student approached her assignment differently. She studied prescribed materials, conducted a preliminary literature search, and drafted her outline in her own words. Her submission demonstrated strong contextual framing, correct placement of theory, and a realistic data strategy. She earned 75%, showing that human-led engagement with materials remains the cornerstone of academic success.
Reflecting on these cases, Prof Yu emphasised, “AI tools are most effective when used after foundational understanding is established. AI-generated content can sound correct while still being academically weak.”
He said using AI as a replacement for learning often leads to superficial understanding, and to guarantee strong academic work, it is best practice to engage with prescribed materials and literature.
Prof Yu's emphasis is on guiding students toward the ethical and effective use of AI. “The University recognises that AI can be a powerful tool for refining ideas, exploring literature, and enhancing productivity, but use it critically, transparently, and responsibly,” said Prof Yu.
“The case study also raises important institutional questions: How can assessments be designed to reward genuine understanding rather than polished but shallow output? How can departments build AI literacy that strengthens disciplinary thinking instead of bypassing it?”
UWC is encouraging staff to share classroom examples like these to ensure the University develops a shared, realistic and pedagogically sound approach to AI. By doing so, the University can respond to the challenges of AI in education and shape ethical, responsible and future-ready learning.
- Use AI to simplify or make study notes on concepts and theories you struggle with, as a complementary tool, not a replacement. Rather, use it as an aid to traditional research methods (Google Scholar, keyword searches and established databases).
- Keep human judgment central and always apply your own critical thinking when deciding which sources are relevant and credible.
- Verify AI outputs carefully by screening suggested sources for relevance and accuracy before including them in your work.
- Consult supervisors and experts and seek guidance from academic mentors when identifying leading researchers or shaping your literature search strategy.
- Recognise AI’s limitations and understand that AI-generated lists may include irrelevant, less rigorous, or even inaccurate materials. Treat them as starting points to further stimulate critical thinking, not as final answers.
- Do not ask CoPilot to write your entire essay, and avoid using AI as your only research tool. It cannot replace established academic search methods.
- Don’t assume AI outputs are authoritative and never take AI-suggested sources at face value; they may lack academic rigour.
- Don’t bypass critical evaluation, since skipping the screening process risks including weak or irrelevant material in your work.
- Don’t treat AI as a shortcut and resist the temptation to let AI do the work for you. It should support, not replace, your own effort.
- Don’t ignore ethical responsibility by using AI without transparency or verification; it undermines academic integrity.
