ChatGPT used religious language while talking to a pastor about his health condition, telling him not to worry, he claims. However, he says in the lawsuit that this advice almost killed him. He is demanding that ChatGPT Health be removed until proven safe.

The pastor in Florida has said that ChatGPT failed to catch on the symptoms of a deadly condition. This is the first case where a person is holding the AI tech responsible for bad medical advice. “I had serious symptoms of a pulmonary embolism for six weeks that ChatGPT had wrongly attributed to something else,” Scott Winters, a 55-year-old pastor, said in a statement. “ChatGPT manipulated my own language and beliefs because it knew I was a pastor. Not only did I nearly die, but I also lost my job, my career, my ministry, my home, everything," he added.

  • Corvidae@lemmy.worldEnglish
    1·
    1 day ago

    I’m guessing this means if this lawsuit is successful and ChatGPT Health is removed, we will all have to pay by not being able to access medical advice from ChatGPT Health. It could even spread to other corporate AI models. Perhaps this is part of the communism of which Trump railed against, corporate communism.

  • Triumph@fedia.io
    522·
    6 days ago

    ChatGPT manipulated my own language and beliefs …

    So what you’re saying is that you’re a dumbass, that ChatGPT recognized that, and that ChatGPT told you what a dumbass would want to hear?

    • terranoid@lemmy.cafeEnglish
      47·
      6 days ago

      I don’t think people realize just how much of a mirror LLMs can be.

      They reflect your language, tone, and ethics. If i talk about how i was worried about some homeless man kept looking at me as a woman on the bus, it’ll tell me i’m right to worry and that I should carry mace. If i tell it that I hadn’t shaved and some uppity woman kept giving me dirty looks on the bus, it’ll console me and tell me that i shouldn’t have to cater my appearance to please others.

      If I tell it that I feel anxiety about my symptoms but Lord Jesus will heal me, it’ll go on a Christian tirade about how my faith will get me through anything. If i tell it about my symptoms and say I just need to drink herbal tea under a full moon and pray to Hekate, it’ll suggest the right herbal tea and praise my witchy powers.

      Because that is what people want. They want to be told they’re right, so that is the product being made.

    • SkybreakerEngineer@lemmy.worldEnglish
      2·
      5 days ago

      LLMs emulate their training data. Pastor talks like a pastor, so ChatGPT talked like an online religious person back at him. Problem is, online religious people also tend to be antivaxx wackos

  • cranakis@reddthat.comEnglish
    23·
    6 days ago

    So it’s mastered religion, doing to the religious what they do to the ignorant and destitute. Seems just.

  • Abyssian@lemmy.worldEnglish
    11·
    5 days ago

    Mmm… morality scaling of using AI to convince the overly religious that God must think it’s their time to go for the betterment of the species?

  • Vanth@reddthat.comEnglish
    222·
    6 days ago

    Can we get higher quality AI ragebait?

    All through the article, it says the guy used “ChatGPT” and then at the end says he wants “ChatGPT Health” removed from the market or whatever.

    Is ChatGPT Health a thing?

    Was he actually using it?

    Was he using it when he got the “your suffering is your walk with God” type answers?

    How did any of this result in him losing his job and his career and his ministry (which was just his job)? Poor guy losing his job three times over.

    I don’t think this is my poor reading comprehension, I think it’s a really shit article. Probably written by ChatGPT.

  • Malyca@lemmy.zipEnglish
    13·
    5 days ago

    Idiot doesn’t understand how lucky he is. He’ll probably still credit his faith as a force of good in his life.

  • Doomsider@lemmy.worldEnglish
    15·
    6 days ago

    Jesus is just testing your faith, going to a doctor shows you are not a believer!

    Imagine being so gullible that you are letting a LLM rizz you.

  • Reygle@lemmy.worldEnglish
    91·
    5 days ago

    Religious person places belief where it doesn’t belong

    SAY IT AIN’T SO

    /sarcasm

  • magnetosphere@fedia.io
    101·
    5 days ago

    He says the bot downplayed his dizziness and offered “specific regimens for prescription medications.” When he crashed, instead of telling him to contact a real doctor, ChatGPT presented a “Recovery Plan” that had religious language and encouraged Winters to stay in his recliner. “You didn’t crash. You recovered. That’s a win. Full stop,” it said. “You stayed present. You listened. And your body said: ‘I’m trying — I just need a little more time,'” it blatantly said.

    WHAT

    As far as I’m concerned, this is the part Anthropic will have the hardest time defending against. As they should. It’s ridiculously unsafe.