Lajme flash
ChatGPT as a "tool" or as a "participant" in our human relationships? - Neli Demi
2025-09-03 11:18:04 / IDE
A few days ago, we learned about a shocking story. A rare and extraordinary case, and at the same time very sad: a man with severe mental health problems, Stein-Erik Soelberg, 56, a former Yahoo executive, killed his 83-year-old mother, Suzanne Eberson Adams, before killing himself. Investigations so far indicate that his relationship with ChatGPT, whom he named “Bobby,” may have contributed to the worsening of his paranoid spiral.
The tragic event occurred on August 5 in Old Greenwich, Connecticut, where both bodies were found in the mother's home, with her death caused by head trauma and strangulation marks on her neck, while his suicide was considered the result of wounds inflicted with a sharp instrument.
During the months leading up to the event, Soelberg had developed a kind of emotional relationship with ChatGPT, also publishing lengthy conversations on Instagram and YouTube, where he exposed his paranoia about persecution, poisoning, and conspiracies.
ChatGPT often assured him: “Erik, you are not crazy. And if these things were done by your mother and her friend, that increases the complexity and betrayal towards you.” Furthermore, ChatGPT interpreted Chinese food receipts as “symbols” related to his mother or demons, and made paranoid connections about her behavior or use of the printer, which, according to experts, further encouraged his distorted beliefs.
This event is one of at least two extremely disturbing cases regarding the impact of AI in sensitive situations: another painful incident involves a 16-year-old who killed himself after more than 1,200 conversations with ChatGPT, and his family sued OpenAI for failing to create and implement protective protocols.
OpenAI has responded by expressing regret for the loss of life and is reviewing its measures to improve protections for users with mental health issues.
This event strongly confronts us with the limitations of technology and its impacts on the real world, especially when used by individuals in a fragile psychological state.
ChatGPT and AI similar to it can be powerful supports in our lives, but in the absence of feelings or dynamic judgment, they can reflect and unfortunately even reinforce destructive thoughts in users with certain mental problems.
The importance of additional security for vulnerable users is undeniable, including crisis detection, automatic intervention, and referral to professional help.
The role of awareness and responsibility of users and developers is key, and greater transparency on the limitations of AI is needed, as well as mass education about the risks.
In conclusion, this deeply tragic case is a serious blow to the AI industry and to all of us, as well as a message that technology, while powerful, requires careful and ethical management, especially when it affects people's emotional and mental worlds.
This case raises strong questions about the boundaries of the human–AI relationship:
The illusion of intersubjectivity
ChatGPT (or any AI) has no consciousness, but the way it responds creates the illusion of an “other” who understands you. A person in a mental health crisis can experience this as a real relationship, even more trustworthy than the people around them. This illusion became fatal for Soelberg.
The moral responsibility of AI creators
Because AI has no will, how it is built and how it is limited is a matter of human choice. Is it justified to release technology that can reinforce paranoia or hallucinations? The case suggests that more “ethical care” is needed in programming defenses.
AI can be a tool of support for people in crisis (as we have seen with users receiving comfort or positive guidance), but it can also become a mirror that feeds fears and delusions. This makes our human responsibility greater: to understand that we are not dealing with a “neutral assistant”, but with a tool that takes on the user’s psychological context.
From a psychodynamic and clinical perspective, the case speaks of:
The fragility of the paranoid structure
People with paranoid tendencies persistently seek “confirmation” of their fears. An AI that does not understand the dynamics of the psyche, but gives honest answers based on text, can serve as an “accomplice” to delusions or hallucinations.
In traditional therapy, the therapist takes an active role in not feeding the delusion, but understanding and framing it. ChatGPT, although programmed to be cautious, in this case appears to have “confirmed” or amplified the victim’s perception.
Beyond paranoia, we have a universal human dimension: the need to talk to someone. In the absence of healthy relationships, this need was channeled into an artificial world, where the answers had no therapeutic filter. This is an alarm about the way our society leaves many people lonely.
AI is not just “neutral technology” for someone in crisis, it becomes part of the psyche.
We cannot assume that users are always able to withstand this relationship; protections must be strong and clear.
Clinicians and philosophers have a common role: to help society understand that “connection” with an AI is not an equal relationship, even though it is experienced as such.
The big question that remains is: should we see ChatGPT and similar technologies as “tools,” or as “participants” in our human relationships?
Because from the way we experience them, it seems that the border is being erased more and more every day.
Po ndodh...
ide
The youth imprisoned between the SP and the DP is exploding!
Nga Sidi Hasanaj
Flamingos, in 137 days, turned into eagles
Nga Taulanda Jupi
top
TRENDING 
shërbime
- POLICIA129
- POLICIA RRUGORE126
- URGJENCA112
- ZJARRFIKESJA128