At 4:30am one day, Julian Walker was vomiting so much from stress that he could not go back to bed. \u00a0<\/p>\n
Desperate for “immediate support”, he turned to artificial intelligence (AI).<\/p>\n
In fact, for the past three years, AI has been his go-to source of consolation.<\/p>\n
“I’ve been working my way back from a work injury which left me with post-traumatic stress disorder (PTSD),” Julian told the ABC.\u00a0<\/p>\n
<\/p>\n
Julian says ChatGPT helps him manage his mental health in real time. (Supplied: Julian Walker\u00a0)<\/p>\n
The 39-year-old from Queensland is working part-time as he tries to build up to full-time hours again.\u00a0<\/p>\n
“I’ve been slowly increasing my work capacity over time, and that process has been supported by AI,” he said.<\/p>\n
Julian said he had attended “more than 50 psychology sessions” during the past three years, all focusing on his work-related trauma.<\/p>\n
However, he felt himself needing a break from it.\u00a0<\/p>\n
“It got to the point where talk therapy wasn’t really getting anywhere,” he said.<\/p>\n
<\/p>\n
Some Australians, like Julian, turn to AI at times when human support is not available. (Supplied: Julian Walker\u00a0)<\/p>\n
It was what led Julian to create his own customised support system \u2014 named “Sturdy” \u2014 inside ChatGPT.<\/p>\n
“It does not diagnose or treat, that is not what I need,” he explained.<\/p>\n
“What I need is a calm presence that remembers me, responds with care, and helps me think clearly when I am overwhelmed.”<\/p>\n
Julian’s situation \u2014 and his burgeoning connection with AI \u2014 is not unique.<\/p>\n
He is one of the hundreds of people who responded to an ABC call-out asking people to share how AI<\/a> was impacting their lives.<\/p>\n ‘We’re not ready’: We asked readers how they felt about AI. This is what they said<\/a><\/p>\n AI is creeping into every part of life \u2014 work, therapy, school, art \u2014 and Australians are divided between fascination and fear.<\/p>\n From white-collar workers to university students, many shared experiences of not just using AI as a therapist, but as a friend, to get them through tough times.<\/p>\n While they were enthusiastic about the convenience and cost-effectiveness of AI, they remained cautious and emphasised using it in moderation with other, in-person options, and not as a substitute for professional clinical advice.<\/p>\n However, for student counsellor Catherine, whose name has been changed for privacy reasons, there are areas where AI can be “more effective” than human counsellors, such as with memorising clients’ histories or testimonies.\u00a0<\/p>\n “Having done some face-to-face counselling during my professional placement, I know how difficult it is to remember my clients’ content from one week to the next,” she said.\u00a0<\/p>\n Catherine believes AI’s ability to recall every detail of people’s past interactions is one of its key benefits. (Pexels: Alex Green)<\/p>\n “When you forget something a client told you, it can result in a minor rupture. AI doesn’t seem to forget, so the support feels seamless.”<\/p>\n Catherine said having constant access to AI was what made it so appealing.<\/p>\n “When you’re dealing with acute stress or anxiety, you need immediate therapeutic support,” she said.<\/p>\n “A human counsellor is not typically available all hours of the day. AI can offer that level of accessibility.”<\/p>\n AI tools improving but experts stress caution\u00a0 AI tools like ChatGPT have had to update their systems to better respond to user distress or red flags. (Reuters: Dado Ruvic)<\/p>\n Recently, OpenAI announced it had updated ChatGPT “to better recognise and support people in moments of distress”.\u00a0<\/p>\n If you or anyone you know needs help:Suicide Call Back Service<\/a> on 1300 659 467Lifeline<\/a> on 13 11 14Aboriginal & Torres Strait Islander crisis support line 13YARN<\/a>\u00a0on 13 92 76Kids Helpline<\/a> on 1800 551 800Beyond Blue<\/a> on 1300 224 636Headspace<\/a> on 1800 650 890ReachOut<\/a> at au.reachout.com<\/a>MensLine Australia<\/a> on 1300 789 978QLife 1800 184 527<\/a><\/p>\n The AI company said it was working with more than 170 mental health experts “who have real-world clinical experience” in a statement on its website.<\/p>\n “We’ve taught the model to better recognise distress, de-escalate conversations, and guide people toward professional care when appropriate.”<\/p>\n The company also said it expanded access to crisis hotlines, re-routed sensitive conversations originating from other models to safer models, and added reminders to take breaks\u2060 during long sessions.<\/p>\n University of New South Wales neuroscience professor Joel Pearson told the ABC that this was a step in the right direction, but added that it was important to stay cautious.<\/p>\n Joel Pearson warns that people need to be very careful when using AI. (Supplied: Joel Pearson)<\/p>\n “OpenAI is not trained to be a therapist,” Professor Pearson said.<\/p>\n “Chatbots don’t have to do a degree, pass a test, or anything like that.”<\/p>\n AI and data science professor Ronnie Das at the University of Western Australia recommended that people read OpenAI’s press release<\/a> carefully “before trusting the system”.<\/p>\n “This is still an experimental system in development that deals with a sensitive issue like mental health,” Mr Das said.<\/p>\n “The problem with the previous models was that they could have affirmed or endorsed harmful beliefs. The new model is much safer in that respect.”<\/p>\n Both experts raised the issue of AI-powered companion apps, which allow users to build characters with which they can text or even hold voice and video calls.<\/p>\n A promotional image for Replika AI companions. (Supplied: Luka\/Replika)<\/p>\n Earlier this year, US media reported on a lawsuit against Character.AI alleging negligence, wrongful death and deceptive trade practices.<\/p>\n It came after a 14-year-old boy died by suicide after forming a romantic attachment to an AI he created on the platform<\/a>.<\/p>\n The boy’s intentions were reportedly encouraged by the AI, which he had modelled after the Game of Thrones character Daenerys Targaryen.<\/p>\n “Unfortunately, Character.AI didn’t have the right guardrails built into it, so it didn’t start alerting people or stopping the interaction, which it should have,” Professor Pearson said.<\/p>\n AI is governed by different regulations at the Commonwealth and state and territory levels.<\/p>\n<\/p>\n
<\/p>\n
<\/p>\n
<\/p>\n