[theqoo] WOMAN WITH EPILEPSY D**S AFTER STOPPING MEDS BASED ON CHATGPT'S ADVICE

A Chinese woman in her 30s has d**d after following medical advice from the generative AI chatbot ChatGPT, which led her to stop taking her epilepsy medication. Her grieving family has filed a lawsuit against OpenAI, the developer of ChatGPT, alleging that the AI’s dangerous advice caused her death.

According to a report by The New York Times (NYT) on the 27th (local time), Bowen Wan, a software engineer living in Seattle, was found dead this past February. An autopsy determined the cause of death to be sudden unexpected death in epilepsy (SUDEP).

Wan, who had managed her epilepsy with medication for over a decade, was under significant stress due to a job layoff and visa-related issues. Concerned that the side effects of her medication were causing cognitive decline, and worried that this might hinder her performance in job interviews or her ability to communicate in English, she decided to reduce her dosage without consulting a medical specialist.

During this process, she asked ChatGPT, which she frequently relied on, for advice on how to reduce her dosage. She devised a plan to taper down from 400 mg per day to 300 mg, and eventually to zero.

It is reported that ChatGPT described this as a "typical self-directed dosage adjustment." According to the bereaved family, ChatGPT interpreted dangerous symptoms, such as confusion, loss of consciousness, and hallucinations, in a positive light, effectively encouraging her to continue with the cessation of the medication.

Wan’s parents discovered records of conversations she had with ChatGPT on her mobile phone, which continued right up until shortly before her death. They have filed a lawsuit against OpenAI, alleging that the AI ​​ignored dangerous symptoms and, in effect, engaged in the unauthorized practice of medicine.

The model Wan used at the time was reportedly GPT-4o. While expressing regret, OpenAI stated that the specific model is no longer in service and that their latest models have been improved to more accurately recognize medical situations.

original post: here

1. They probably kept giving positive responses whenever the user suggested reducing the dosage

2. Yeah, given that level of intelligence, maybe it’s better if you just don’t take it at all

3. The answers just followed along with the leading questions she asked

4. Honestly, it's worrying how many people rely on AI these days...

5. I really wonder what’s wrong with people like this. Is that an appropriate thing to say on an article about a fatal accident? Are they even human?

6. Even though ChatGPT said to take it, she probably just kept "electrocuting" it with their persistent, biased questioning until she got the answer they wanted

7. Why would you believe in GPT....

8. Even in mild cases of epilepsy, consistent medication is essential...

9. No matter how much I tried to nudge my AI, they absolutely wouldn't budge. Maybe it’s because it's Gemini? Either way, it's a shameㅠ

10. Wait, is there actually a risk of death if you don’t take epilepsy medication??

New post

Next Post

✉ PC Stories