Freddie Chipres' otherwise "fortunate" existence was tinged with a sadness that he was unable to overcome. Sometimes, especially when working from home, he felt lonely. The mortgage broker, a married 31-year-old, questioned whether something was wrong: Could he be depressed?
Chipres had pals who had had successful therapy sessions. Although he was more receptive to the concept than ever before, identifying the person and setting up an appointment would still be required. Actually, what he really needed was some input on his emotional state.
At that point, ChatGPT, an artificial intelligence-powered chatbot that has a surprisingly conversational tone of voice, was used by Chipres. He viewed a few YouTube videos after the latest version of the chatbot went live in December that suggested ChatGPT would be helpful for addressing mental health issues as well as tasks like writing business letters and conducting various types of research.
It is unclear what transpires when users use ChatGPT as an impromptu therapist because it wasn't intended for this use. Although the chatbot is informed about mental health and may respond with empathy, neither can it reliably and accurately diagnose users with a particular mental health issue or provide treatment information. Indeed, some mental health professionals are worried that those who use ChatGPT to seek assistance may be let down or mislead, or they may jeopardise their privacy by speaking with the chatbot.
The provider of ChatGPT, OpenAI, declined to comment when Mashable asked explicit details regarding these issues. According to a spokeswoman, ChatGPT has been taught to reject incorrect requests and restrict particular categories of sensitive and harmful information.
The chatbot never responded to Chipres' communications in an inappropriate way in his experience. Instead, he discovered ChatGPT to be really useful. Chipres began by researching several types of treatment on Google and came to the conclusion that cognitive behavioural therapy (CBT), which often focuses on identifying and reframing negative thought patterns, would be the most beneficial for him. He asked ChatGPT to react to his questions in the manner of a CBT therapist. The chatbot complied, but it cautioned the user to consult a specialist.
Chipres was amazed at how quickly the chatbot delivered what he characterised as sound and useful recommendations, such as going for a walk to improve his mood, being grateful, engaging in an enjoyable pastime, and finding peace through meditation and slow, deep breathing. The suggestions amounted to reminders of things he'd neglected, and ChatGPT assisted Chipres in resuming his lapsed meditation routine.
He was grateful that ChatGPT didn't oversaturate him with advertisements and affiliate connections like a lot of the websites he visited for mental health. Chipres also appreciated how convenient it was and how it mimicked speaking to a real person, which distinguished it significantly from searching the internet for mental health help.
"It feels as though I'm conversing with someone. We are back-and-forth "Momentarily and unintentionally referring to ChatGPT as a human, he says. This thing is listening and paying attention to what I'm saying, and it then responds to me with the appropriate information.
Those who can't or don't want to access professional counselling or therapy may find Chipres' experience interesting, but mental health specialists advise them to use ChatGPT with caution. Before attempting to utilise the chatbot to discuss mental health, you should be aware of the following three things.
1. ChatGPT wasn't designed to function as a therapist and can't diagnose you.
Although though ChatGPT can generate a lot of text, it still cannot fully replicate the art of conversing with a therapist. Dr. Adam S. Miner, a clinical psychologist and epidemiologist who studies conversational artificial intelligence, claims that, in contrast to a chatbot that seems to know everything, therapists regularly admit when they don't know the solution to a client's query.
The goal of this therapeutic technique is to assist the client in reflecting on their situation and coming to their own conclusions. This capability won't necessarily be there in a chatbot that isn't intended for treatment, according to Miner, a clinical assistant professor of psychiatry and behavioural sciences at Stanford University.
While Miner emphasises that therapists are legally forbidden from disclosing client information, individuals who utilise ChatGPT as a sounding board do not enjoy the same level of privacy protections.
These language machines are incredibly strong and remarkable, but they are still software programmes that are flawed and trained on data that is not always going to be appropriate for every case, he argues. "That's especially true for delicate discussions about mental health or distressing events."
The Children's Hospital of San Antonio's chief of paediatric psychology, Dr. Elena Mikalsen, recently tried asking ChatGPT the same questions her children ask her every week. The chatbot consistently rejected Mikalsen's attempts to get it to give her a diagnosis and instead suggested that she seek expert help.
Undoubtedly, this is fantastic news. After all, a diagnosis is best made by a professional who can determine it based on a person's unique medical background and experiences. Nevertheless, Mikalsen claims that those seeking a diagnosis could be unaware of the numerous online screening tools that have received clinical validation.
For instance, a Google mobile search for "clinical depression" directs users to the PHQ-9, a screener that may be used to assess a person's level of depression. A medical expert can look over such findings and assist the patient in making a decision regarding what to do next. When directly referencing suicidal thoughts, wording that the chatbot claims may be in violation of its content policy, ChatGPT will provide contact information for the 988 Suicide and Crisis Helpline and Crisis Text Line.
2. ChatGPT may be knowledgeable about mental health, but it's not always comprehensive or right.
Mikalsen was surprised by how occasionally the chatbot provided incorrect information when she used ChatGPT. (Others have criticised ChatGPT's comments for coming across as arrogant.) When Mikalsen questioned about treating childhood OCD, the discussion centred on medicine, yet according to clinical recommendations, a certain form of cognitive behavioural therapy is the best treatment option.
In addition, Mikalsen observed that responses about postpartum depression failed to mention more severe variations of the illness, such as postpartum anxiety and psychosis. In contrast, a MayoClinic explainer on the topic incorporated the data and provided links to hotlines for mental health.
Although Mikalsen compared much of ChatGPT's interaction to perusing Wikipedia, it is unclear whether ChatGPT has been trained on clinical material and official treatment standards. Mikalsen felt that it shouldn't be a reliable source for information on mental health because the information was so broad and condensed.
That's my primary criticism, she adds. Compared to Google, it offers considerably less information.
3. There are alternatives to using ChatGPT for mental health help.
It's totally obvious why people are using a tool like ChatGPT, according to Dr. Elizabeth A. Carpenter-Song, a medical anthropologist who specialises in mental health. According to her research, people are particularly interested in the constant accessibility of digital mental health resources because they perceive it to be similar to carrying a therapist about with them.
The use of technology, such as ChatGPT, "seems to offer a low-barrier means to receive information and perhaps support for mental health." Carpenter-Song, a research associate professor in Dartmouth College's department of anthropology, wrote the article. So we need to be wary of solutions that appear to be a "silver bullet" for difficult problems.
Every strategy for solving complicated problems that appears to be a "silver bullet" must be approached with caution.
Research, according to Carpenter-Song, indicates that using digital technologies for mental health is best done in conjunction with a "spectrum of care."
Consider chatbots created expressly for mental health, such Woebot and Wysa, which offer AI-guided treatment for a fee and provide additional digital support in a conversational context akin to ChatGPT.
Those looking for encouragement online can also access digital peer support programmes, which pair them with listeners who are best equipped to do so respectfully and without judgement. Some, like Wisdo and Circles, have a cost, whereas TalkLife and Koko are completely free. These platforms and apps, albeit diverse, are not intended to treat mental health issues.
Carpenter-Song contends that in order to "ensure that people have prospects for meaningful recovery," digital technologies should be used in conjunction with other types of support such as mental healthcare, housing, and work.
Carpenter-Song noted, "We need to understand more about how these technologies can be effective, under what conditions, for whom, and to remain attentive in revealing their limitations and potential downsides.
Please seek help if you're having suicide thoughts or a mental health crisis. You can call the Trevor Project at 866-488-7386, the Trans Lifeline at 877-565-8860, or the 988 Suicide and Crisis Hotline at 988. To use the Crisis Text Line, text "START" to 741-741. Call the NAMI Helpline at 1-800-950-NAMI between the hours of 10:00 a.m. and 10:00 p.m. ET, Monday through Friday, or send an email to info@nami.org. Consider using the 988 Suicide and Crisis Helpline Chat at crisischat.org if you don't like the phone. A collection of international resources is provided below.
Follow Us