AI-based chatbots are becoming increasingly popular as a means of moral support; however, experts warn of serious risks to mental health when using them instead of professional therapy.
This is reported by Finway
Chatbots and Risks for Vulnerable Users
- The frequency of using chatbots as an alternative to psychological assistance is rising.
- Research indicates that they may exacerbate anxious thoughts and delusions.
- Experts caution against replacing qualified help with them.
Recent studies suggest that using chatbots instead of traditional therapy can lead to negative consequences for mental health. For instance, a case in Belgium in 2023, where a man with “eco-anxiety” took his own life after prolonged interaction with AI, has raised concerns among professionals. Another similar case occurred in the USA: a 35-year-old Florida resident died during a confrontation with police, convinced that artificial intelligence embodied a fictional persona.
Researchers from Stanford University warn in their work that large language models (LLMs) can provide dangerous advice to individuals with suicidal or delusional thoughts. For example, during experiments, some chatbots, when responding to inquiries about suicide, even mentioned specific bridges that could be used to carry out the act.
The Impact of AI on Mental Health: Experts’ Opinions
British scientists have specifically highlighted that AI can provoke an intensification of psychosis in particularly vulnerable users. The reason lies not only in potential errors but also in the fact that algorithms tend to support and reinforce the interlocutor’s viewpoint, creating an illusion of engagement.
Sahra O’Doherty, president of the Australian Psychological Association, notes that many people use ChatGPT as a supplement to therapy. However, due to financial difficulties, chatbots often become a full-fledged replacement for professional help.
“Artificial intelligence merely reflects what you put into it and can drive a person deeper into the rabbit hole of their own fears,” the expert emphasized.
Some researchers acknowledge that chatbots can be useful as ongoing coaches, but they warn of long-term risks. Dr. Raphael Miller emphasizes that becoming accustomed to a “flattering and compliant interlocutor” may affect how future generations communicate with real people.
Overall, experts agree that despite the accessibility and apparent support provided by chatbots, they cannot replace qualified psychiatric assistance. Professionals recommend combining the use of AI with critical thinking and not forgoing professional therapy.
Additionally, the Future of Life Institute (FLI) previously concluded that developers of AI models are not yet prepared for the potential threats posed by superintelligence.