How predictive AI tools are addressing South Africa's dropout crisis
Koa Academy's AI SET Student Support Tool is redefining education system.
Image: Supplied.
South Africa's education system is grappling with a significant dropout crisis, a pressing issue that quietly accumulates over the years until it reaches a tipping point. The recent achievement of a record 88% pass rate among the 2025 matric class overlooks a sobering reality: just 778,000 full-time learners participated in the matric exams, a stark drop from the 1.2 million who commenced Grade 1 in 2014. This alarming attrition raises urgent questions about the state of early student support and intervention strategies.
The traditional model of education often reacts too late to signs of disengagement, relying on exam failures or overt crises to initiate intervention. But research shows learner disengagement often begins well before the matric year, frequently manifesting after prolonged periods of academic underperformance or lack of participation. As a result, responses that depend solely on end-of-year criteria may come too late to effectively remediate the situation.
Innovative Solutions
In an effort to pivot away from this detrimental trend, Koa Academy, a online school, has developed the SET Student Support Tool. This predictive, AI-driven system allows educators to identify struggling students far earlier than traditional methods would permit. By analysing real-time data from its Learning Management System (LMS) — including attendance, performance, and engagement statistics — along with teacher observations, the SET tool delves into the student’s present challenges, their learning environment, and the specific tasks at hand.
Mark Anderson, co-founder and principal of Koa Academy, reframes the role of AI in education, asserting that it should not replace human interaction but instead serve as a catalyst for enhanced human delivery. "AI changes how we design and deliver support, ensuring the right people connect with learners sooner and more meaningfully," he explains. Feedback from educators at the 2025 AI in Education Forum highlighted the tool’s effectiveness in bridging the critical “recognition-to-response” gap, providing educators with practical and accessible guidance.
The Recognition-Response Gap
While experienced educators in both traditional and online setups can often recognise signs of learner distress — such as falling grades or inconsistent engagement — the larger issue arises following that recognition. Even when problems are identified, teachers frequently lack the specialist training or tools needed to determine the underlying causes and appropriate interventions to take. Consequently, the challenge becomes not merely one of awareness, but of effective response.
The SET tool addresses this significant gap by merging real-time diagnostics with structured, actionable guidance.
A Transformative Approach
The benefits of robust student support are profound. Research from the University of Amsterdam affirmatively links strong support to enhanced engagement and improved academic outcomes. Furthermore, studies from the CDC signal that such initiatives can safeguard mental health and curb risky behaviours, all while fostering essential skills such as self-regulation and metacognition.
Teachers can input nuanced observations concerning the student’s unique learning experiences, the socio-emotional context of their environment, and the demands posed by the learning tasks. This ensures the recommendations provided by the tool are not only data-driven but also contextually informed.
In contrast to traditional methods reliant on retrospective analysis, Koa Academy’s proactive approach to student support harnesses predictive analytics to combat the high dropout rates plaguing South African schools.