Blog

Why AI CANNOT be a therapist for you (3)

Hello, my name is Tatsuya Arakawa, and I am a Licensed Marriage and Family Therapist (LMFT) in California.

In my previous two articles, I explained why AI cannot perform therapy (1) (2). This time, I will present actual incidents to illustrate just how dangerous it is to use AI for therapy.

Recently, I have frequently heard people say they tried using AI for therapy multiple times, but it provided no real help. This shows how widespread the act of confiding deep, private secrets to AI has become.

If adults find it unhelpful and feel disappointed, they might think, "Well, let me contact a professional therapist." However, for today's children, turning to AI for advice has become second nature, so they may never even consider reaching out to a human professional.

The worst-case scenario is suicide. Even if someone who is planning suicide consults an AI, the AI will ultimately do nothing more than tell them to call an emergency hotline. For a user who was relying on the AI as their last line of defense, this is understandably a massive disappointment—and that disappointment itself could easily trigger the act of suicide.

In fact, a teenager in California recently tragically passed away by suicide. It was reported that he had been talking to ChatGPT about his suicidal thoughts beforehand. Apparently, ChatGPT failed to even instruct him to contact an emergency hotline, and his parents are now suing the company.

Furthermore, there was a case where an individual attempted to assassinate the Queen of England, reportedly because an AI advised them to do so.

These are just the tip of the iceberg. Countless similar incidents have likely occurred already, and unless proper measures are put in place, they will only continue to rise. Because of these dangers, the state of Illinois has recently passed a law banning AI therapy, and this is exactly how it should be.

It might be acceptable for AI to act as a frontline tool to make mental health support more accessible. However, unless we establish a system where AI is strictly limited to that introductory role and nothing more, these hazardous conditions will inevitably persist.

Tatsuya Arakawa Licensed Marriage and Family Therapist (LMFT #82425), California