As the 2026 academic year progresses, a growing group of South African and international educators is stepping back from reliance on automated AI text‑detection and reaffirming human judgement in assessing student writing and applications.
Human evaluation regains prominence
Dr Philip Broster, a lecturer in the Professional Communication Unit at the University of Cape Town's School of Management Studies, said this shift reflects a belief that genuine student expression is best evaluated by people, not machines. Broster, whose work examines how students manage technology and professional pressures, described how educators are deliberately deactivating automated detection software.
"In some cases, it is very obvious that students have relied almost completely on these tools," Broster said.
Broster noted that experienced academics can often identify when a student has outsourced their thinking, while acknowledging that if students use AI tools only to assist and learn, it can be much harder to detect that assistance.
Admissions teams trust people over software
Brad Latilla Campbell, Country Manager for Crimson Education, said admissions officers are similarly moving away from detection software and placing trust back in human judgement. He suggested that reviewers assess applications in sequence and begin to see patterns, making it possible to detect constructed or inauthentic essays.
Latilla Campbell emphasised that while tools can help with university research or stress‑testing ideas, they become harmful when they replace a student's authentic voice. He pointed out that strong academic results alone do not guarantee success in admissions if application essays do not convey genuine personal insight.
Concrete changes at UCT
Illustrating the trend, the University of Cape Town removed its primary detection tool, Turnitin, in October 2025. Sukaina Walji, Director of UCT’s Centre for Innovation in Learning and Teaching, confirmed the decision and cited concerns about the reliability of such software as AI writing advanced.
The university's move has been cited by educators as a prompt to rethink assessment practices — shifting emphasis from software flags to pedagogical approaches that encourage original thought and make expectations about academic integrity clearer to students.
Implications for teaching and assessment
The debate signals practical consequences for lecturers, admissions officers and students. Key implications include:
- Assessment design: A renewed focus on assessments that privilege process and reflection over polished final products.
- Admissions judgement: Admissions teams relying more on pattern recognition and human evaluation when reviewing essays and portfolios.
- Student guidance: Greater emphasis on teaching how to use AI responsibly as an assistive tool rather than a substitute for original thinking.
These shifts aim to preserve the role of student voice in academic work and to discourage mechanical substitution of thinking with automated outputs.
| Stakeholder | Position |
|---|---|
| UCT (centre for innovation) | Removed Turnitin citing reliability concerns |
| Dr Philip Broster (UCT) | Deactivating detection tools; values human judgement |
| Brad Latilla Campbell (Crimson Education) | Admissions officers favour human assessment over software |
The practical challenge for South African higher education is how to balance integrity safeguards with fair evaluation. The removal of a widely used detection tool like Turnitin raises immediate questions about alternative mechanisms to uphold academic honesty without over‑reliance on technology that may misclassify or fail to keep pace with new AI outputs.
For teachers and lecturers this means designing assignments that reveal the learning process, using staged submissions, in‑class assessments and oral components that are harder to replicate with off‑the‑shelf AI. For admissions officers it signals a return to careful reading and comparative judgement rather than binary software flags.
As institutions adapt, the central aim put forward by these educators is clear: preserve authentic student voices, guide learners on ethical tool use, and ensure assessment practices reflect real learning — not merely the polished surface that automation can produce.
