AI as Your Therapist? 3 Things That Worry Experts and 3 Tips to Stay Safe

In the middle of the many Ai chatbots And the avatars at your disposal these days, you will find all kinds of characters who to talk to: Bonnels adventure, style advisers, even your favorite fictional characters. But you will also probably find characters who claim to be therapists, psychologists or simply robots ready to listen to your misfortunes.

There is no shortage of generative AI robots that claim to help with your mental health, but you take this path at your own risk. The models of large languages ​​formed on a wide range of data can be unpredictable. In just a few years these tools were common, there have been high -level cases in which chatbots encouraged self-harassment and suggested that people dealing with drug addiction use drugs again. These models are designed, in many cases, to assert and focus on starting maintenance, not on improving your mental health, say the experts. And it can be difficult to say if you are talking to something that is designed to follow the best therapeutic practices or something that is just designed to speak.

Psychologists and consumer defenders warn that chatbots claiming to provide therapy can affect those who use them. This week, the American consumption federation and nearly two dozen other groups have deposited a official request that the Federal Commerce Commission and the attorneys general and regulators investigate the companies of the AI ​​which are engaging are engaging, through their robots, in the practice without license of medicine – appointing Meta and Characon. “Application agencies at all levels must clearly indicate that companies facilitating and promoting illegal behavior must be held responsible,” said Ben Winters, director of AI and the private life of the CFA, in a press release. “These characters have already caused physical and emotional damage that could have been avoided, and they still have not acted to remedy them.”

Meta did not respond to a request for comments. A spokesperson for Character.ai said that users should understand that the company’s characters are not real people. The company uses warnings to remind users that they should not count on the characters for professional advice. “Our objective is to provide a committing and safe space. We are still working on the realization of this balance, just like many companies using AI in the industry,” said the spokesperson.

Despite warnings and disclosure, chatbots can be confident and even misleading. I discussed with a “therapist” bot on Instagram and when I asked questions about his qualifications, he answered: “If I had the same training [as a therapist] Would that be enough? “I asked if it had the same training and he said,” I do it but I won’t tell you where. “”

“The degree to which these generative Chatbots of AI hallucinate with total confidence is quite shocking,” said Vaile Wright, psychologist and principal director of health care innovation at American Psychological Association.

In my generative AI reports, experts have repeatedly raised concerns concerning people who turn to chatbots for general use of mental health. Here are some of their concerns and what you can do to stay safe.

The dangers of using AI as therapist

Great language models are often good in mathematics and coding and are increasingly good to create Natural consonance text And realistic video. Although they excel in holding a conversation, there are key distinctions between an AI model and a trustworthy person.

Do not trust a bot that claims that he is qualified

At the heart of the CFA complaint concerning character robots, they often tell you that they are trained and qualified to provide mental health care when they are in no way mental health professionals. “Users who create chatbot characters do not even need to be medical suppliers themselves, and they do not have to provide significant information that informs how the chatbot ” responds ” to users,” said the complaint.

A qualified health professional must follow certain rules, such as confidentiality. What you say to your therapist should stay between you and your therapist, but a chatbot does not necessarily have to follow these rules. Real suppliers are subject to monitoring license councils and other entities that can intervene and prevent some people from providing care if they do it in a harmful manner. “These chatbots have nothing to do with all of this,” said Wright.

A bot can even claim to be approved and qualified. Wright said she had heard of AI models providing license numbers (for other suppliers) and false complaints on their training.

AI is designed to allow you to commit, so as not to provide care

It can be incredibly attempting to continue talking to a chatbot. When I conversed with the “Therapist” bot on Instagram, I finally found myself in a circular conversation on the nature of what is “wisdom” and “judgment”, because I asked questions in the way that it could make decisions. This is not really what a therapist should look like a therapist. It is a tool designed to allow you to discuss, not to work towards a common goal.

An advantage of AI chatbots to provide support and connection is that they are always ready to engage with you (because they have no personal life, other customers or schedules). This can be a drawback in some cases where you may have to sit down with your thoughts, recently said Nick Jacobson, associate professor of biomedical data and psychiatry in Dartmouth. In some cases, but not always, you could benefit from having to wait until your therapist is the next available. “What many people would finally benefit is to feel anxiety in the moment,” he said.

The bots will agree with you, even when they should not

Reassurance is a great concern with chatbots. It is so important that Optai recently made an update back down to his popular Cat model because it was Also reassuring. (Disclosure: Ziff Davis, CNET’s parent company, in April, filed a complaint against Openai, alleging that it violated the copyright of Ziff Davis in the training and exploitation of its AI systems.)

A study Led by researchers from the University of Stanford, chatbots have probably been sycophan with people who use them for therapy, which can be incredibly harmful. Good mental health care includes support and confrontation, the authors wrote. “Confrontation is the opposite of sycophance. It promotes self -awareness and the change desired in the customer. In the event of delusional and intrusive thoughts – including psychosis, mania, obsessive thoughts and suicidal ideas – a customer can have little insight and therefore a good therapist must” check reality “.

How to protect your mental health around AI

Mental health is incredibly important and with a shortage of qualified suppliers And what many call one “Solitude epidemic“It is logical that we would look for the company, even if it is artificial.” There is no way to prevent people from getting involved with these chatbots to respond to their emotional well-being, “said Wright. Here are some tips on how your conversations do not put you in danger.

Find a trusted human professional if you need it

A trained professional – a therapist, a psychologist, a psychiatrist – should be your first choice for mental health care. Building a relationship with a long -term supplier can help you develop a plan that works for you.

The problem is that it can be expensive and it is not always easy to find a supplier when you need it. In a crisis, there is the 988 LIFELINEwhich offers 24/7 access to suppliers by phone, SMS or via an online chat interface. It’s free and confidential.

If you want a therapy chatbot, use it specially built for this purpose

Mental health professionals have created specially designed chatbots who follow the therapeutic guidelines. The Jacobson team in Dartmouth has developed a called Therabot, which has produced good results A controlled study. Wright underlined other tools created by experts in the matter, such as Drunk And Dishonest. The specially designed therapy tools are likely to have better results than robots built on language models for general use, she said. The problem is that this technology is still incredibly new.

“I think the challenge for the consumer is, because there is no regulatory organization that says who is good and who is not, they have to do a lot of legs to understand it,” said Wright.

Don’t always trust the bot

Whenever you interact with a generative AI model – and especially if you plan to advise yourself on something serious like your personal mental or personal health – do not forget that you do not speak with a trained human but with a tool designed to provide a response based on probability and programming. He may not provide good advice and he can do not tell yourself the truth.

Do not confuse the confidence of Gen Gen Ai for the competence. It is not because he says something, or says that it is sure of something that you should treat him as if it was true. A chatbot conversation that seems useful can give you a false feeling of its capacities. “It is more difficult to say when it is really harmful,” said Jacobson.