The AI Said 'You're Absolutely Right.' She Quit Treatment, Pressed Charges.

5 points
1/21/1970
a day ago
by thatha7777

Comments


DennisP

Ironic that this article sounds like it was written by AI.

> The desperation is gone. In its place: epistemic certainty.

> And it told her, in prose that sounded like medicine and felt like prophecy, that she was right about everything.

> In Greek, medical terminology isn't borrowed, it's part of the native language.

> it doesn't land like pseudoscience. It lands like something Galen might have written.

> Strip out the hedging, and you have something that sounds like medicine but functions like prophecy.

> It wasn't her framing. It was the LLM's. She brought real pain, real conditions, real institutional failures. The LLM gift-wrapped them in the one genre guaranteed to get them dismissed.

33 minutes ago

PaulHoule

Every LLM I've seen approved of my project to transform into a fox except for Meta AI which makes fun of it.

a day ago

thatha7777

the interesting part isn't whether Meta's AI is right or wrong, it's that when N models say "great idea" and one pushes back, the one pushing back feels like the broken one

a day ago

PaulHoule

I think other AI agents have been trained to talk with somebody like a high IQ and remember a lot of context, Meta's AI has been trained to talk like somebody with a moderate-to-low IQ with a sense of humor.

I can get Gemini or ChatGPT on the other hand to use words like "ego-syntonic" and talk about folk religion in China and about mind-body work you would use in character acting, etc.

Also if foxwork is a delusion it has a large element based in reality. It started out as "I felt a presence" and when I needed to explain it I developed a cover story that became real

https://mastodon.social/@UP8/tagged/foxwork

I even keep KPIs as people keep approaching me and I have to keep printing more tokens to replace the ones that I give away.

a day ago

gus_massa

Somewhat related: My mother in law asked a question to Meta AI and got a silly joke as an answer.

a day ago

PaulHoule

It was trained to do that the same way that ChatGPT was trained to say "That's not funny — it's serious!"

It's a different market position.

a day ago

curio_Pol_curio

Quite related: my mom asked Gemini a question, here was the answer (I lol'd at it, so it mustnt have been well-trained?):

To overcome the fear of affinity fraud, simply lean into the grandeur of your own delusion

15 hours ago