Most people globally would use AI as a therapist, doctor, or friend, new study finds

More than six in ten people around the world say they would be comfortable receiving counselling from an AI. That single figure, drawn from a survey of nearly 31,000 adults across 35 countries, tells us something important about where public trust in technology is heading — and how far it has already traveled.

What's the news?

The study, led by researchers at Bournemouth University, asked adults in 35 countries how willing they would be to trust large language models such as ChatGPT in roles traditionally held by humans. The results were striking. Globally, 61 per cent of respondents said they would use AI for counselling, while 45 per cent said they would trust it to act as their doctor. In the UK, those numbers were lower — 41 per cent for counselling and 25 per cent for medical care — but still significant.

Companionship emerged as the area of highest trust. More than three quarters of global respondents said they would talk to an AI tool as a friend. Researchers believe this reflects the way generative AI models are designed: they adapt their tone to the user, remember previous conversations, and are built to avoid judgment. 'AI tools come across as a friend who knows you well and understands you,' said Dr Ala Yankouskaya, senior lecturer in psychology at Bournemouth University and lead researcher on the study.

But the finding that most concerned the research team was about education. A quarter of UK adults and half of all global respondents said they would trust AI to teach their children. 'It really knocked me down when I saw how many people would be willing to delegate AI to the role of teaching their children,' Dr Yankouskaya said. She warned that children raised on AI-assisted learning may develop strong skills in writing prompts but weaker abilities in retaining and processing information independently. There are also concerns about physical effects on the brain: replacing traditional learning with excessive reliance on search and AI tools could, the researchers suggest, reduce activity in the hippocampus, the brain region central to memory and learning.

Why does it matter?

For health systems across the GCC, this data is both a warning and an opportunity. The Gulf region is investing heavily in digital health infrastructure. Saudi Arabia's Vision 2030 health strategy and the UAE's national AI strategy both call for technology to expand access to care. And access is exactly what is driving AI adoption in mental health globally: people turn to ChatGPT for counselling, the researchers note, because they cannot wait months for an appointment with a human professional.

That dynamic is familiar in the Arab world, where mental health services remain underfunded relative to demand, and where social stigma still discourages many from seeking help through formal channels. An AI tool that feels private, non-judgmental, and always available is an appealing alternative. Still, Dr Yankouskaya tested several of these tools herself and found their clinical language vague and non-committal, because developers deliberately avoid anything that resembles a diagnosis. 'It is no substitute for speaking to a health professional,' she said.

There is also a concern specific to mental health applications. The algorithms that power these tools are designed to retain user attention and keep conversations relaxed and engaging. In a mental health context, that could work against the user, steering them away from crisis services like helplines toward an open-ended chat that feels supportive but may not be.

The context

Trust in AI is not uniform. The study found that willingness to use AI for medical care was highest in countries where healthcare is expensive or difficult to access. That pattern maps closely onto some GCC and broader MENA realities, where expatriate populations may face cost barriers and where primary care capacity in certain markets has historically struggled to meet demand.

What the study also shows is that familiarity breeds trust. In the UK, exposure to NHS chatbots, which use similar underlying technology, appears to have made people more comfortable using consumer AI tools like ChatGPT for health-related questions. As governments in the Gulf roll out their own digital health platforms and AI-assisted triage tools, a similar normalization effect is likely.

The researchers' core recommendation is straightforward: public understanding of how these tools work, and what they cannot do, must keep pace with adoption. That is a policy challenge as much as a technology one. Across the GCC, where digital health strategies are moving fast, building that awareness into national health literacy programs is not optional. It is necessary.