As AI tools become more common in mental healthcare, they are prompting new discussions around access, emotional support, ethics, and the evolving role of human therapists.

In recent years, artificial intelligence has become a powerful tool in healthcare, particularly in the field of mental health.

With the global demand for psychological support exceeding the availability of qualified professionals, some researchers have turned to AI, alongside psychobiotics, nanotechnology, as a potential solution, especially in psychotherapy.

Studies have demonstrated that AI can improve access to care and offer ongoing, personalised support for those experiencing emotional distress. Chatbots such as ChatGPT, Calmi, and Wysa now provide a rather affordable 24/7 assistance, symptom tracking, and cognitive behavioural techniques.

Research also shows that AI tools can effectively help manage conditions such as depression and anxiety, with some trials reporting diagnostic accuracy rates of up to 95% and a 30–35% reduction in anxiety symptoms in crisis-affected areas like Ukraine.

Yet, while these technologies present exciting opportunities, they also raise critical ethical, clinical, and relational questions. 

As AI becomes more deeply embedded in therapeutic contexts, pertinent questions often linger if it can truly replace the human therapist, or it works more like an additional support.

How do AI chatbots work as emotional supporters?

AI chatbots like ChatGPT and Calmi often use advanced natural language processing and machine learning algorithms to simulate human-like conversations.

Dennis Han, a computer science student, developed his AI therapist, Calmi, after going through a difficult breakup. He told Doha News that accessing traditional therapy was challenging due to “long waiting lists, difficulty finding the right therapist”.

Turning to ChatGPT revealed the potential for AI support, leading him to explore public therapy datasets and AI architectures. 

Crucially, these systems do not possess genuine emotional understanding; they simulate empathy based on patterns in their training data. As Han stated, “it’s mimicking. Large language models don’t feel anything. They can simulate empathy in a way that feels real”.

The primary goal for Han isn’t human-like comprehension, but providing responses that are emotionally useful.

He said, “the goal isn’t to create something that understands emotions like a human. It’s to make something that responds in ways that are emotionally useful. If it helps someone feel heard or work through a tough moment, that’s what matters.” 

Ralina Valiullina, a graduate of the Doha Institute for Graduate Studies and a regular ChatGPT user, has found AI chatbots to be helpful. During a period of emotional difficulty, especially after concluding her therapy sessions a year ago, she began using ChatGPT to reflect on certain topics, often after first discussing them with friends and family.

She told Doha News, “sometimes I felt like I was too obsessed with certain problems. I talked about them a lot with my friends. And yet, I felt that I still needed to talk. So, to avoid bothering my friends, I’d just go and talk to ChatGPT”.

Valiullina described the chatbot as a safe, convenient space. “It became a journaling space where I could get even more emotional support. It reminded me that I’m worthy, that it’s okay to feel the way I do. But it wasn’t a replacement for real conversations.”

She added that AI’s gentle tone, which she could guide and adjust, helped her stay kind to herself: “I told ChatGPT to talk to me sweetly, and it started doing that, and still does. But you have to be careful, it learns from you, and you train how it talks to you.”

AI chatbots: Threat or complement to therapy?

While users like Valiullina have found AI tools helpful for emotional support, mental health professionals caution against over-reliance on such systems, especially when it comes to treating deeper psychological issues.

Rachida Smaili, a psychological counsellor in Qatar, stressed that AI can be valuable, but only when used responsibly and in the right context.

“We wouldn’t say that it’s enough for emotional support, but we can use it for raising awareness or for those who cannot afford therapy,” she told Doha News. “The key is to use it the right way. It’s not a total threat if the person has enough awareness about how to use these tools.”

Smaili explained that AI may help users recognise early signs of distress and encourage them to seek professional help. In her words, “The AI itself has limits where they advise to go to professionals, especially if we’re speaking about self-harm or very severe cases.”

However, she warned that complete dependence on AI could have unintended consequences. “Avoiding human interaction is one of the hardest challenges in our field,” she said. “If people think, ‘I’m getting everything I want from AI, I don’t need to see a therapist,’ that’s totally wrong.”

Hala Awada, a psychological counsellor in Qatar, shared similar concerns. She acknowledged that AI’s accessibility and non-judgemental interface could make it appealing for those hesitant to seek therapy. “It’s fast, doesn’t require an appointment, it’s affordable, so it just makes it easier for people to get emotional support,” she said.

But Awada was quick to emphasise that this is not the same as therapy. “AI can offer support, but it can’t replace the deep understanding and ethical care that comes from trained professionals.”

She also expressed concern over the potential emotional risks, including dependency and social withdrawal. “One big risk is that it can make people feel more emotionally isolated,” she explained. “It could reinforce unhelpful thoughts or patterns because it doesn’t challenge or reframe them like a therapist would.”

One of her key concerns was AI’s inability to challenge unhealthy thought patterns. “AI could end up reinforcing some unhelpful thoughts or patterns,” she explained, “because it doesn’t have the ability to challenge or reframe those beliefs like a therapist would.”

For Awada, the emotional safety that therapy provides stems from the human connection itself. “AI might use the right words to make someone feel heard,” she said, “but it can’t feel the pain. It can’t empathise. That kind of connection only comes with another human.”

This was also highlighted by ChatGPT user Valiullina, who acknowledged that while AI offers a certain kind of emotional support, it remains limited in scope. “Even if it shows empathy and understanding, it’s still one-sided,” she told Doha News. “It’s being trained based on how you talk. If you tell it, ‘Be objective with me, analyse from several angles,’ it might help. But again, it’s not a real human.” 

While it is clear that AI cannot replicate the depth of human connection central to therapy, an inevitable question arises: could it ever come close?

The future of AI in mental healthcare

AI systems are already capable of mimicking empathetic responses, remembering user preferences, and adjusting their tone to reflect users’ emotional states. But with advances in technology and the ongoing revolution in AI, will it come closer to human understanding?

Psychological counsellor Smaili stressed that “AI will never replace the therapist or psychiatrist or any mental health professional,” and warned against relying on AI as a complete substitute.

Awada also emphasised the vital difference between AI emotional support and professional therapy.

She said, “AI can’t really replicate this therapeutic alliance or this alliance between a therapist and a client because this relationship is built on trust, the therapist’s attunement to the client, and a genuine human connection.”

She does not believe that AI, despite all its advancements, will ever reach that stage.

She explained, “The therapist, their experience, their body language, their use of humour, and most importantly the empathy, all of that creates a safe environment for the client.”

She warns that AI cannot guarantee confidentiality or ethical safeguards, as it is not bound by professional licensing or legal standards. This concern echoes wider professional insistence that therapy must be delivered by qualified, licensed practitioners to ensure safety and effectiveness.

Yet Calmi’s AI developer Han pointed out the limitations and variability within human therapy itself, noting that many struggle to find trustworthy, safe spaces with professional therapists. 

Han envisions a hybrid future where AI increases access to mental health support, especially for those who face barriers like cost or waiting times, allowing human therapists to focus on deeper, complex cases. 

He explained, “AI can offer consistency, nonjudgment, and emotional clarity,” while underscoring the need for ongoing collaboration with mental health professionals to ensure safety and responsibility.

He assured that user data is anonymised and not accessible to developers or professionals. 

“For data, we take privacy seriously. User information is not accessible to us. Even in rare cases where we’d need to look at something, the data is completely anonymous. Nothing is tied back to a real identity,” Han explained. 

He views artificial intelligence as a means to reimagine the mental health field, one in which therapists are able to concentrate on deeper, high-impact work, while AI tools offer accessible, real-time support for more routine needs. 

He emphasised that such a model cannot function without the involvement of qualified mental health professionals. He highlighted that “building these systems still requires real mental health professionals. We need their insight to shape the models responsibly, and we need humans involved to make sure these tools are safe, helpful, and grounded in real therapy practices.”