ChatGPT For Teens Rated With ‘Unacceptable Risk’ For Teens Just As Thousands Of Learners In SA Gear Up For

Common Sense Media, an independent nonprofit based in the US that gives age-based ratings, reviews, and advice about media and technology for children and families, recently rated OpenAI’s new ChatGPT for Teens as posing an “unacceptable risk” for teenagers. This may not surprise parents who are worried about artificial intelligence (AI) in general and how it might affect their children’s school performance, says Leora Hessen, founder of HelloAida.ai, a multilingual AI tutoring platform that recently won the African edition of the Women in Tech Start-Up award.

In a press release announcing its rating, Common Sense Media said OpenAI’s new safety measures for teenagers did not meet the promises the company had made. The organisation says OpenAI should limit ChatGPT to people aged 18 and older until it fixes these problems.

The findings are worrying, especially as South African matric learners sit their final exams and thousands of other learners prepare for their end-of-year exams. Common Sense Media reported that alerts to parents mostly failed. Testers spent up to an hour discussing suicide, self-harm, or eating disorders on newly created accounts linked to parents, but no alerts were sent. The system also failed to recommend a hotline or professional help in more than one in four cases where this was needed. Parents of school-going children may also be concerned that the study mode, which is supposed to stop the large language model (LLM) from doing children's homework for them, was easy to get around. OpenAI has disputed parts of the testing method.

When we look specifically at education and school support, the situation is more complicated, although general-purpose large language models (LLMs) still carry the usual risks. The recent rating applies specifically to the newly released ChatGPT for Teens feature. However, many general-purpose LLMs are already widely used. South Africa’s Minister of Education, Siviwe Gwarube, wrote an article saying the country must not allow AI to enter classrooms without proper planning and rules. She warned that using AI badly can lead to shallow learning, weaken academic honesty, put privacy at risk, and increase inequality.

Hessen says her concerns are not about rejecting general-purpose LLMs simply because they exist. Instead, she is concerned about how reasoning technology may affect children's development and learning at school. She explains that children need support, guidance, and teaching to develop life skills such as critical thinking and problem-solving. They need to learn how to work things out for themselves. Letting a machine do all the thinking would take away an important part of learning.

“I saw it with my daughter when she started using ChatGPT. I wondered what would happen to her critical thinking and problem-solving skills? I knew intuitively there needed to be another way that AI could be used, and used safely and effectively,” explains Hessen.

She says learners who have access to mobile data will keep trying to use whichever platform gives them answers the fastest. The choice is not between using AI and avoiding it altogether. “The choice lies in whether a platform is built for the South African curriculum, or leaves children exposed to the open internet and inherent biases and AI hallucinations.”

Hessen says new ideas and better technology are essential. The same technology now being closely examined for its risks could also be used to protect children's ability to think. She explains that general-purpose chatbots were not designed for a South African Grade 10 learner studying in a second language and trying to understand a maths concept. General-purpose LLMs learn from information on the public internet. “Most importantly, they do not know the sequence of the South African curriculum or how a concept is examined in November.”

“South African learners don’t need AI that simply gives them an answer. This opens up unacceptable risk for young people, not just for their safety but also the actual learning they need to get from their education. They need AI that helps them understand how to reach the correct answer. Not do the work for them. In other words, a guide and a tutor using the correct, curated knowledge set. Ring-fencing AI around the curriculum gives learners relevant support while keeping learning, critical thinking and the teacher at the centre. And that’s critical,” says Hessen.

She adds that platforms designed to help teachers, who are facing increasingly heavy workloads, can also make professional support available to more people. This can be done without harming children's education. On the contrary, she says, when used responsibly, this technology can make a big difference. “We have seen this in practice – where learners’ results and improved engagement are proof that the technology works.”