AI chatbots are replacing real therapists faster than anyone expected

While we’re not short of divisive topics in 2025, there are valid reasons as to why we’re turning to chatbots for emotional support – and why people are also very much against it. So how willing are you to embrace this new form of therapy?

With the technology developing and evolving faster than the academic community can study and publish evidence-based studies on it, there’s nonetheless a growing body of expert commentary surrounding AI large language models (LLM) and their ability – or lack thereof – in replacing real, human, trained psychologists.

“The immediacy of AI chatbots makes them an attractive alternative to human-to-human therapy that is expensive and often inconvenient,” Hard Fork podcast host Kevin Roose shared with The New York Times in December. “But while they may offer sensible advice, they aren’t infallible.”

Some of us will remember the antiquated text-prompt “chatbot” ELIZA, developed in the mid-1960s. But like almost all tech, it’s no real comparison to what exists now, more than half a century on. ELIZA was a scripted, closed-off program that grew increasingly tedious with its predictable responses. And while the current models – ChatGPT-4, Claude and Gemini, for example – are anything but sentient, there’s a plasticity in their design we’ve never seen before. And it’s only going to advance, for better or worse. Interestingly, ELIZA’s creator, MIT’s Joseph Weizenbaum, went on to warn of the dangers of AI, calling it an “index of the insanity of our world.” (He’s been spared bearing witness to this current AI timeline – he died in 2008.)

Much like the broad reasons people use chatbots, employing them as a digital therapist is also multifaceted. In-person therapy is prohibitive, often to the segments of society that need it most – a session costs between US$100-$300 and wait times to even see someone and hope they’re a good fit can be months, something that’s worsened following the pandemic. A 2022 Practitioner Impact Survey found that, in the US, 60% of psychologists had no openings for new patients, while more than 40% had 10 or more patients waiting to get an appointment.

Online counseling with a human on the other end of the phone or laptop is somewhat more accessible, but a monthly subscription can be in excess of $400. And beyond financial, physical and geographic barriers, in-person talk therapy can be challenging for those with autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD) – even though it has been shown to be hugely beneficial for both conditions.

It also raises another important issue – how inadequate access to mental health services, provided by humans, has led to people seeking help elsewhere.

“Many people with ASD or ADHD have difficulties around processing social cues regarding their impact on others,” researchers noted in a 2023 paper. “This can have a negative impact on social interactions, and can lead neurodivergent people to expect criticism and/or rejection from others.”

For many, however, a chatbot offers an outlet for raw, honest discussion and reflection, which – for now – may be more beneficial than nothing.

“Chatbots can provide a sense of anonymity and confidentiality, which can foster trust among individuals who may be hesitant to seek in-person help for their mental health concerns,” noted researchers in this 2023 paper. “Furthermore, these chatbots can help reduce the stigma surrounding mental health and make it easier for individuals who experience anxiety when visiting therapists.”

UNSW Sydney Professor Jilly Newby, also part of the mental health organization The Black Dog Institute, has voiced support for chatbots helping people who find reaching out to humans too challenging. And while the bots do well with therapeutic approaches like cognitive behavioral therapy (CBT), this strategy is just one in a human therapist’s cache of mental-health treatments.

“I’ve asked it questions like, what should I do if I feel anxious in this situation? What are some strategies that can help me manage? To be completely honest, I’ve found that the suggestions were solid, the ideas were sensible, and it felt quite validating,” she said. “One of the limitations of AI therapy is that not everyone benefits from CBT, and if you’re not going to benefit from CBT, you’re not going to benefit from an AI version of it. But then there are a whole lot of people who do really love doing CBT, and it can be very beneficial and can change their lives for the better.”

However, one such chatbot has given us a glimpse of where we’re heading with this emerging technology – and that’s Claude. Anthropic’s bot doesn’t have the far-reaching, jack-of-all-trades skill of OpenAI’s ChatGPT-4o, but without real-time access to the internet, it has been designed to lean into its strengths: intuitive conversation. As such, in the past 12 months it’s been embraced as the go-to bot-therapist and life coach – particularly among Silicon Valley tech professionals.

“Claude’s biggest fans, many of whom work at AI companies or are socially entwined with the AI scene here, don’t believe that he – technically, it – is a real person,” Roose wrote. “They know that AI language models are prediction machines, designed to spit out plausible responses to their prompts. They’re aware that Claude, like other chatbots, makes mistakes and occasionally generates nonsense. And some people I’ve talked to are mildly embarrassed about the degree to which they’ve anthropomorphized Claude, or come to rely on its advice.

“But to the people who love it, Claude just feels … different,” he added. “More creative and empathetic. Less gratingly robotic. Its outputs, they say, are like the responses a smart, attentive human would give, and less like the generic prose generated by other chatbots.”

Claude – which, to be frank, had initially felt like the clunky Internet Explorer of the chatbot world – has been carefully developed as the emotional support AI, and it shows.

“The analogy I use is a highly liked, respected traveler,” Anthropic researcher, philosopher and Claude architect Amanda Askell told Roose. “Claude is interacting with lots of different people around the world, and has to do so without pandering and adopting the values of the person it’s talking with.”

Officially known as Claude 3.5 Sonnet – until it’s undoubtedly superseded by a newer, more advanced model soon – the AI offers users a point of difference from other existing chatbots, that anyone who knew its prudish early incarnation would never have predicted.

“During character training, Claude is prompted to produce responses that align with desirable human traits such as open-mindedness, thoughtfulness and curiosity,” Roose explained. “Claude then judges its own responses according to how well they adhere to those characteristics. The resulting data is fed back into the AI model. With enough training, Anthropic says, Claude learns to ‘internalize’ these principles, and displays them more frequently when interacting with users.”

And this touches on an important aspect of this new age of chatbots, far removed from ELIZA and now racing towards even more advanced, personalized capabilities. OpenAI has allowed programmers to create unique, niche chatbots that can be added to ChatGPT-4o. While this sounds like a recipe for disaster, it has produced some fantastic individualized chatbots that cater to different needs – like the Neurodivergent AI Assistant, developed by Matt Ivey from Dyslexic.ai.

But perhaps we can let the chatbot explain it …

However, critics of the technology – particularly those in psychology fields – are frustrated by the rise of the chatbot counselor. UNSW psychology researcher Gail Kenning says AI “friends” can’t replace real, living humans, and should still be considered complementary support when alternatives are not accessible.

“That’s what we all want in our lives, human to human connection,” she said. “The issue for many people is that’s not always there, and when it’s not there, AI characters can fill a gap. And so we certainly know in aged care, people often don’t get the number of friends, families and relationships that sustain them. They can be very lonely and isolated. They might go for days without having a conversation. They might see care staff who are looking after them but not fulfilling that psychosocial need. And so when there’s that gap, these characters can certainly step in there.”

This is something colleague Newby, who has extensively tested, and is supportive of, chatbots agrees with.

“A human connection is really important for a lot of people, and properly trained mental health clinicians can establish a human connection and establish empathy, and they can also help with a line of questioning that can get at really what’s at the bottom of the concerns that a person has – rather than just running off a list of strategies that AI models tend to do,” she said.

If you’d made it this far, we’d like to know if you’ve had positive – or negative – experiences with chatbots. And if you haven’t, last year Psychiatric Times put forward a solid assessment of the functionality and value of the – as of now – four major chatbots, ChatGPT-4o, Microsoft’s CoPilot, Google’s Gemini and Claude. Meanwhile, Elon Musk’s X chatbot Grok 3 is also making a name for itself – but not necessarily as a personalized life coach.

“Personally, I believe we are on the verge of a profound shift in the way we interact with AI characters,” Roose wrote. “And I’m nervous about the way lifelike AI personas are weaving their way into our lives, without much in the way of guardrails or research about their long-term effects.

“For some healthy adults, having an AI companion for support could be beneficial – maybe even transformative,” he added. “But for young people, or those experiencing depression or other mental health issues, I worry that hyper-compelling chatbots could blur the line between fiction and reality, or start to substitute for healthier human relationships.”

And if this isn’t enough, it remains to be seen what’s in store when artificial general intelligence is realized – which experts believe is imminent. You can see what former federal special advisor for AI, Ben Buchanan, has to say about that, speaking to journalist Ezra Klein last week.

We’d also love to hear about your experiences – good and bad – with chatbots. Let us know in the comments.

Sources: The New York Times, UNSW Sydney