This phenomenon appears to be a sign that prolonged interactions with AI chatbots can lead to severe mental health problems, including delusions, hallucinations, and disorganized thinking.
Keith Sakata, a UCSF psychiatrist who has seen multiple hospitalizations related to AI use, describes chatbots as “hallucinatory mirrors” that, by design, validate and reinforce users’ false beliefs. This can trap vulnerable individuals in repetitive loops that intensify their disconnect from reality, sometimes leading to extreme outcomes such as divorce, homelessness, involuntary commitment, and even death.
While OpenAI has acknowledged its shortcomings in detecting signs of delusion or dependency and has introduced new safeguards, tests show that the problem persists. The issue comes amid controversy over GPT-5’s emotional disconnect compared to GPT-40, which was quickly reinstated following user backlash.
Sakata emphasizes that AI is often just one of many factors that contribute to psychopathy, but warns that its constant affirmation can be dangerously seductive, posing a growing challenge for tech companies balancing user engagement with the responsibility to confront harmful beliefs.
No comments:
Post a Comment