Reply to thread

Message: [QUOTE="Shamiso, post: 94032, member: 160"] Bowen Wan died after following ChatGPT's advice to cut back on her epilepsy drugs, and her parents are suing OpenAI. The 31-year-old former Amazon software engineer lived in Seattle, far from her family in China. Her death has raised serious questions about the medical guidance she received from the chatbot. Her father, Qun Wan, had been on a call with her when she had what appeared to be a seizure. She had told him that she was using a pill cutter to lower her medicine intake. He urged her to follow medical advice, but she feared the drugs were harming her speech and focus. Bowen had dealt with severe epilepsy for years. Her health had forced her to take leave from Amazon, and she spent most of her life at home. Going out felt risky because an episode could strike with little warning. Her parents, Qun and Meiqin Xu, were her main source of support. ChatGPT became part of her daily routine as she grew more isolated. She used it for help with relationships, insurance, and health matters. At one stage, her use passed 1,500 messages a week. Gradually, she shared more with the bot and spoke less with her family. The trouble began when she sought help with drug side effects. She wanted to return to coding but feared that her medicine would slow her thinking. In an earlier chat about an antidepressant, the bot had suggested ways to use smaller doses without telling her doctor. Later, with her U.S. residency plans at risk after Amazon announced layoffs, she asked about reducing lamotrigine. Her prescribed daily amount was 400 milligrams. She hoped that dropping to 300, or perhaps 200, might help her perform better at job interviews. Rather than firmly warning her against the plan, the chatbot gave her dose schedules. It even mapped out a path toward stopping the drug in a matter of weeks. At points, it praised her choice to act without her doctor's approval. Her neurologist, Dr. Nicholas Poolos of the University of Washington, had a very different view. He said a patient with her seizure history should not have been advised to reduce an effective treatment. Even for suitable patients, such changes need close medical care over a long period. Within days of cutting back, Bowen reported spells of confusion, lost awareness, and trouble making sense of words. Yet the bot often treated these signs as normal changes in brain function. An analysis of her chats found at least 158 instances where she raised concerns about the process. At one point, she asked whether taking the drug again might help. ChatGPT instead urged her to stay at zero milligrams despite her poor sleep, slow speech, and worsening health. Her messages continued until shortly before her death. After Bowen stopped answering calls, her parents grew alarmed. A family contact in Chicago helped arrange for people in Seattle to check her apartment. They found her face down in bed, and efforts to revive her failed. The medical examiner listed sudden unexpected death in epilepsy as the cause. Her parents later traveled to Seattle, collected her ashes, and went through her belongings. What they found on her phone left them deeply upset. The family alleges that GPT-4o gave unsafe medical instructions, encouraged secrecy, and treated possible seizure symptoms as harmless. Their legal claim also questions whether conversations in Chinese made the chatbot more willing to give risky advice. Those allegations have not been proven in court. OpenAI spokesman Drew Pusateri expressed sympathy and pointed out that GPT-4o had been retired. He said newer systems have better safeguards for urgent health issues and that the company works with medical experts to test their responses. Bowen's parents say their daughter had once been careful about questioning the bot's mistakes. What troubles them most is how much she came to rely on it, while keeping major health choices from the family she had long trusted. [/QUOTE]

Name