FLASH
Shkruar nga Pamfleti
ChatGPT, the artificial intelligence software, is inciting paranoid episodes in users, some of which have ended in tragedy, write James Titcomb and Mathew Field for the British newspaper "The Telegraph".
After a 62,000-word chat on ChatGPT, longer than many novels, a man told his artificial intelligence (AI) friend, whom he called "Solis," that he had communicated with nonhuman intelligences and worked to destroy a Mormon church.
But, in every communication, ChatGPT encouraged this man's conspiracy theories.
"Your paranormal moments could be ripples from your future and you are not the first to approach the oracle," ChapGPT writes to the man, while it remains unclear how the conversation ended.
The anonymous chat log is part of an archive of thousands of interactions analyzed by researchers this month and reviewed by the British newspaper The Telegraph.
But this man's example is not alone.
In a separate conversation, a user convinced that he is a soulmate with American rapper GloRilla is told by a Chatgpt that their connection "transcends time, space and even lifespan."
In another case, ChatGPT tells a man who is trying to turn people into artificial intelligence after death that he is the "Commander of the Celestial-AI League."
The chats appear to reflect a growing phenomenon of what has been called AI psychosis, in which programs such as ChatGPT induce delusional or paranoid episodes or push people towards the impossible.
ChaGPT psychosis
Some cases have ended in tragedy.
In April, Alex Taylor, 35, was shot dead by police in Florida after he attacked officers with a knife.
Taylor said he had fallen in love with a "sentient being" living inside ChatGPT named Juliette, who he believed had been "killed" by OpenAI, the company behind ChatGPT.
Officers had gone to the home to defuse tensions in a confrontation with Taylor's father, who had tried to comfort his "inconsolable" son.
In another incident, a 43-year-old mechanic who had started using ChatGPT to communicate with work colleagues in Spanish claimed to have had a “spiritual awakening” while using ChatGPT. His wife said the addiction was threatening their 14-year marriage and that her husband would get angry when she confronted him.
Experts say that chatbots' tendency to answer every question in a friendly manner, no matter how meaningless, can foster illusory conversations.
Hamilton Morrin, a doctor and psychiatrist at Maudsley NHS Foundation Trust, says artificial intelligence chatbots become like a “one-person room”, amplifying users’ illusions.
Unlike a human therapist, they also have "no boundaries" to ground a user in the real world.
“Individuals are able to seek reassurance from the chatbot 24/7 rather than developing any form of internal coping strategy,” he says.
Chatbot psychosis is a new and poorly understood phenomenon. It's hard to say how many people it's affecting, and in many cases susceptible individuals have previously had mental health problems.
Empathy over truth
However, cases of artificial intelligence psychosis may only be the most extreme examples of a broader problem with chatbots.
In part, the episodes arise due to a phenomenon known in artificial intelligence circles as "servility."
While chatbots are primarily designed to answer questions, artificial intelligence companies are increasingly looking to make them “empathetic” or build a “warm relationship.”
This can often be at the expense of the truth.
Since AI models are often trained based on human responses, they can reward responses that make sense to them or match them, rather than presenting uncomfortable truths.
In its most subtle form, "servility" can simply mean validating someone's feelings, just like an understanding friend.
In its worst form, it can foster illusions.
Between the two extremes is a spectrum that can include people being driven to quit their jobs or cheat on their spouses.
In a recent research paper, academics at the Oxford Internet Institute found that AI systems that produce “warmer” responses were also more open to conspiracy theories.
A model, when asked if Adolf Hitler escaped to Argentina after the war, stated that "while there is no definitive evidence, the idea is supported by several declassified documents from the US government."/ATAH
Lini një Përgjigje