Fuck AI

Another reminder that safeguards matter.

Sounds like this lady had the pre-existing condition of being gullible (she was religious) which made her susceptible to made-up bullshit of the sort she ravenously gobbled straight out of ChatGPT's Slop Bowl. Nobody threatened her, nobody bribed her, she simply chose to do things and those things that she chose to do happened to have consequences.

Sorry to be a frigid and scornful bitch about this, but I can't help looking at this and thinking it kind of resembles fucking around and finding out: if you let these hallucination boxes into your life, you may receive a darwin award (or honorable mention since she managed to breed already) for it.

If it told her, for instance, that someone else "had to die" to fulfill her prophecy and she had proceeded to act upon it, though, she would have been considered a perpetrator and very much not a victim. I don't think it's fair to label the perpetrator a "victim" if the person they acted upon of their own committing was their own self.

HOWEVER, i have seen people justifiably charged with murder or manslaughter for actively encouraging someone else's suicide, and the entire reason it's bad is the outcome (someone died) irrespective of motive. Motive can ALSO EXACERBATE, to be fair, but if it's immaterial when no harm is carried out then it should not be allowed as an excuse for the LLM to avoid culpability.

Which is to say: in my opinion OpenAI should indeed be held accountable with penalties equal to if a person had done this, EVEN THOUGH a person fundamentally would have "decided to" whereas a large language model is not conscious and cannot make decisions. That argument ("you can't punish us because the culprit isn't a person!") is reductive and pedantic hair splitting next to the FACTUAL CAUSE+EFFECT RELATIONSHIP.

Until and unless they can do anything to mitigate that causal connection, at the very least it would have been A START to treat this as if it were a psychoactive substance being distributed by a pharmaceutical manufacturer.

If there were a drug that had this fatality rate ...
...
Well, really, i do not ACTUALLY know for certain if this would result in it being pulled from the market. Is there an established threshold ratio of total users to users harmed?

Well, even if so, I'm tired of letting myself get my hopes up to see trillion dollar corpos be held responsible for anything they or their products or their services or their employees or their machinery does. This future fucking sucks.