It is available at midnight, it doesn’t judge you and you don’t need medical aid to use it. It is perhaps unsurprising, then, that South Africans are increasingly turning to AI chatbots when they are anxious, overwhelmed or simply need somewhere to put their thoughts. But mental health professionals warn that there is an important line between using AI as a tool and treating it as a therapist.
For many people, typing “I’m having a really bad day” into ChatGPT or another AI chatbot can feel considerably easier than telling another person.
And in South Africa, where access to mental healthcare remains a serious challenge, that accessibility matters.
Mental Health Counsellor and CEO of The ICHAF Training Institute, Devan Moonsamy, says he has encountered people who spoke to ChatGPT about what they were experiencing before speaking to another human being.
“They weren’t looking for technology. They were looking for someone who would answer,” he says.
That distinction is important.
South Africa already faces a significant mental health burden alongside a shortage of professionals, particularly within the public healthcare system. For someone without medical aid, living far from appropriate services or afraid of the stigma associated with asking for help, an AI chatbot can appear to fill a very real gap.
But should it?
So, is it okay to talk to AI about how you’re feeling?
The short answer is yes, but it matters what you expect it to do.
AI can be useful as a supporting tool. You might use it to organise your thoughts before a difficult conversation, create questions you want to ask a therapist or doctor, learn general information about mental health, find grounding exercises or help put feelings into words.
What it cannot do is replace a qualified mental health professional.
Professor Christopher Szabo of the South African Society of Psychiatrists has warned that AI cannot replicate the personal interaction, clinical judgement and accountability of a trained professional.
Moonsamy shares this concern.
“These systems aren’t built to assess suicide risk,” he says.
A chatbot cannot observe changes in your body language, facial expression or behaviour. It does not know you in the way a healthcare professional or trusted person in your life does, and it cannot physically intervene when someone is in danger.
Know when the conversation needs to move offline
There is an important difference between asking AI for ideas to help you unwind after a stressful day and relying on it during a serious mental health crisis.
Consider moving the conversation to a qualified professional or trusted human being if:
- you are thinking about harming yourself or someone else
- your mental health is making it difficult to function normally
- you are experiencing severe or worsening symptoms
- you are using AI instead of speaking to people in your life
- you find yourself becoming emotionally dependent on a chatbot
- you need a diagnosis, treatment plan or advice about medication
- something an AI tells you makes you feel more frightened, hopeless or distressed
You should also remember that AI can be wrong. A confident or compassionate-sounding answer is not necessarily an accurate one.
Parents should know what their children are asking AI
The issue becomes particularly important when children and teenagers are involved.
South Africa’s Department of Women, Youth and Persons with Disabilities has described suicide among young people as an escalating concern and says suicide is the second-leading cause of death among South Africans aged 15 to 29.
For parents, the answer is unlikely to be simply banning AI.
Instead, it may be worth adding AI to the conversations families are already having about online safety.
Ask your child whether they use chatbots, what they use them for and whether they have ever asked one for advice when they were sad, frightened or overwhelmed.
Most importantly, make sure they know that talking to AI is not the same as talking to someone who can actually help them.
A child should know who the real-life people are that they can approach when something feels too big to handle alone.
Don’t give a chatbot everything
There is another consideration that can easily be forgotten when a conversation with AI begins to feel private and personal: it is still technology.
Avoid treating a chatbot like a confidential medical consultation.
Think carefully before sharing highly sensitive personal, medical or identifying information, particularly information about children or other people.
If you want to use AI to help organise your thoughts, you can often do so without including names, addresses, identifying details or information that isn’t necessary for the conversation.
AI can be the first step, but it shouldn’t be the last
The usefulness of AI and the risks associated with it can exist at the same time.
For someone who has never spoken about their anxiety before, typing it into a chatbot may help them finally put a name to what they are feeling. Asking AI, “How do I tell my mom I think I need help?” could even make the next conversation easier.
That is very different from asking a chatbot to be the help.
Moonsamy believes South Africa’s mental healthcare gap is therefore as important to this conversation as regulating the technology itself.
“Close the gap that’s driving people to chatbots in the first place, not just regulate the chatbots,” he says.
He argues that greater investment in community-based counsellors, school support and accessible mental healthcare needs to happen alongside discussions about how AI platforms respond to people experiencing a crisis.
Perhaps the simplest rule for users is this: use AI to help you get to human support, not as a replacement for it.
Where to get help in South Africa
If you are thinking about suicide or are concerned that someone may be at immediate risk, contact the SADAG Suicide Crisis Line on 0800 567 567.
You do not have to wait until things feel unbearable before asking another person for help. A GP, psychologist, counsellor, teacher, trusted family member or other appropriate professional can also be a starting point.
AI may be available 24 hours a day, but when it comes to serious mental health concerns, the person on the other side of the conversation still matters.
Leave a Reply