A shiny dream meets a tough reality. Imagine a classroom in South Africa. It is crowded, loud, and hot. The teacher is trying to help dozens of children at the same time. Now, picture a sleek, high-tech robot standing in the corner.
On August 17, a brand-new factory and showroom opened at the Dube TradePort Special Economic Zone in KwaZulu-Natal. It was built for IRIS, a humanoid robot tutor powered by Artificial Intelligence (AI). IRIS is the creation of Thando Gumede from BSG and Technologies.
Gumede is not a detached tech executive; she is a former math and science teacher from the deep rural parts of KZN who lived the gruelling reality of teacher burnout.
She spent countless personal hours grading and lesson planning instead of giving children one-on-one attention. She built IRIS to fix her own broken reality, aiming to bring the tutor to nearly 23,000 public schools.
A robot that can speak all 11 of South Africa’s official languages is now being positioned as part of the answer to one of the country’s most stubborn education problems.
But there is a question that cannot be avoided: can artificial intelligence help a child learn to read when the deeper problem is not simply a lack of technology?
South Africa’s literacy crisis is stark. The 2021 Progress in International Reading Literacy Study (PIRLS), conducted locally by the University of Pretoria’s Centre for Evaluation and Assessment, found that 81% of Grade 4 learners could not read for meaning. South Africa ranked last among the 57 participating countries.
According to tracking by the Right to Read Campaign, if the country continues on its current path, it will take more than 80 years before 95% of Grade 4 learners can read for meaning. That statistic has not suddenly become relevant because a robot has arrived.
It has been sitting over the education system for years. Now, however, the country is watching a very different kind of intervention take shape.
The robot was first launched in August 2025 and is designed to teach learners from Grade R to tertiary level in all 11 official languages. It sounds futuristic. But for a child sitting in a crowded classroom, the future still has to meet the reality of today.
A robot cannot fix a problem that starts before Grade 4
Learning to read is like building a house. You cannot start with the roof. You have to start with the foundations, like learning the sounds of letters and how words fit together.
According to Neliswa Nyandeni, founder and director of Becoming Unlimited SA, a South African platform dedicated to youth development, literacy advocacy, and personal growth, AI tutors hold the potential to deliver what South Africa urgently requires: greater personalised attention for learners.
“AI robotic tutors have significant potential to support learners, particularly because they can create more personalised and engaging learning experiences,” she told The National Media Lifestyle.
Every classroom brings together students with a wide range of distinct abilities, background languages, and educational needs.
For reading comprehension, Nyandeni says AI could help learners build vocabulary, practise reading, ask questions and receive immediate feedback. But she is careful about what the technology can realistically do.
“A Grade 4 literacy challenge is often the result of gaps that have developed much earlier,” she says.
The University of Pretoria notes that comprehension depends on foundations such as phonological awareness, decoding, word recognition, and reading fluency. In other words, by the time a child is struggling to understand a paragraph, the story may have started years earlier.
The robot may help. But who gets one?
Technology is only fair if everyone can use it. In South Africa, the gap between the rich and the poor shows up clearly in our classrooms. An expensive robot does no good if a school does not even have working plugs.
Furthermore, a report shared by another publication reveals that 43% of households with young children have no books at all, meaning the school classroom is often a child's only gateway to words.
Ms Zhunique Wildeman, a Grade 9 and 10 English teacher, says AI could assist with pronunciation, personalised tests and faster feedback. But she has a simple concern: access.
“Realistically, because many learners are not privileged enough to access the internet, this will put them at a disadvantage.” That concern cannot be separated from the literacy debate.
The University of Pretoria's PIRLS findings already showed significant differences between learners depending on language and location. Rural provinces also tended to perform below the national average.
So putting sophisticated technology into schools without fixing electricity, connectivity, devices, teacher training, books, and other basic resources risks creating another divide: children who can use AI to learn and children who cannot.
Teachers are not the problem to be automated
A machine can give a child a test, but it cannot care about them. It cannot notice when a child feels bad or wants to give up.
Ziyanda Khwezi, a student teacher, has seen the literacy gap up close.
“Some can read the words aloud, but when you ask them what they have just read, they cannot explain it,” she said. She has also seen learners struggle to put their thoughts into sentences. That is why she does not believe AI robotic tutors should become the main solution.
“A robot cannot replace the understanding and relationship that a teacher builds with a learner,” she says.
There is another issue hiding in plain sight: teachers are already stretched. Lesson preparation, marking, assessments, meetings, learner records and administrative work compete with the time teachers actually have to sit beside a struggling child and say: try that sentence again. What do you think it means?
Gumede has previously made a similar argument about IRIS, saying the robot was developed to support rather than replace teachers. The machine can assist with curriculum information, learner results, and marking, potentially taking some pressure off educators. That may be where the technology makes the most sense.
Not teacher versus robot. Teacher plus robot.
At the end of the day, a robot is just a tool. It cannot fix education on its own. Nyandeni says the answer is not to throw technology at children and hope something changes.
“AI robotic tutors can be an exciting part of that solution, but their greatest impact will come when they are integrated into a broader ecosystem of quality teaching, reading opportunities, family engagement and early intervention.”
That ecosystem is the real story. A child still needs someone to notice that she has stopped participating. Someone needs to recognise that he is embarrassed because he cannot read aloud. Someone needs to understand the language spoken at home.
Someone needs to notice that a child is not struggling because they are lazy, but because they missed an important foundation years earlier.
A robot can provide practice. It cannot replace that human attention. That means books still matter. Parents matter. Teachers matter. Early childhood education matters. Language matters. Libraries matter. And yes, technology can matter too.
Wildeman puts it bluntly: “AI cannot serve as a magic wand. It is only made to assist in ideal circumstances.”
vuyile.madwantsi@nationalmg.co.za