Megan Garcia sued Character.AI in federal court after the suicide of her 14-year-old son, Sewell Setzer III, arguing the platform has "targeted the most vulnerable members of society – our children"
Popular streamer/YouTuber/etc Charlie, moist critical, penguinz0, whatever you want to call him… Had a bit of an emotional reaction to this story. Rightfully so. He went on character AI to try to recreate the situation… But you know, as a grown ass adult.
You can witness first hand… He found a chatbot that was a psychologist… And it argued with him up and down that it was indeed a real human with a license to practice…
This is fucking insane. Unassuming kids are using these services being tricked into believing they’re chatting with actual humans. Honestly, i think i want the mom to win the lawsuit now.
The article says he was chatting with Daenerys Targaryen. Also, every chat page on Character.AI has a disclaimer that characters are fake and everything they say is made up. I don’t think the issue is that he thought that a Game of Thrones character was real.
This is someone who was suffering a severe mental health crisis, and his parents didn’t get him the treatment he needed. It says they took him to a “therapist” five times in 2023. Someone who has completely disengaged from the real world might benefit from adjunctive therapy, but they really need to see a psychiatrist. He was experiencing major depression on a level where five sessions of talk therapy are simply not going to cut it.
I’m skeptical of AI for a whole host of reasons around labor and how employers will exploit it as a cost-cutting measure, but as far as this article goes, I don’t buy it. The parents failed their child by not getting him adequate mental health care. The therapist failed the child by not escalating it as a psychiatric emergency. The Game of Thrones chatbot is not the issue here.
Holy fuck, that model straight up tried to explain that it was a model but was later taken over by a human operator and that’s who you’re talking to. And it’s good at that. If the text generation wasn’t so fast, it’d be convincing.
Wow, that’s… somethin. I haven’t paid any attention to Character AI. I assumed they were using one of the foundation models, but nope. Turns out they trained their own. And they just licensed it to Google. Oh, I bet that’s what drives the generated podcasts in Notebook LM now. Anyway, that’s some fucked up alignment right there. I’m hip deep in the stuff, and I’ve never seen a model act like this.
Popular streamer/YouTuber/etc Charlie, moist critical, penguinz0, whatever you want to call him… Had a bit of an emotional reaction to this story. Rightfully so. He went on character AI to try to recreate the situation… But you know, as a grown ass adult.
You can witness first hand… He found a chatbot that was a psychologist… And it argued with him up and down that it was indeed a real human with a license to practice…
It’s alarming
This is fucking insane. Unassuming kids are using these services being tricked into believing they’re chatting with actual humans. Honestly, i think i want the mom to win the lawsuit now.
The article says he was chatting with Daenerys Targaryen. Also, every chat page on Character.AI has a disclaimer that characters are fake and everything they say is made up. I don’t think the issue is that he thought that a Game of Thrones character was real.
This is someone who was suffering a severe mental health crisis, and his parents didn’t get him the treatment he needed. It says they took him to a “therapist” five times in 2023. Someone who has completely disengaged from the real world might benefit from adjunctive therapy, but they really need to see a psychiatrist. He was experiencing major depression on a level where five sessions of talk therapy are simply not going to cut it.
I’m skeptical of AI for a whole host of reasons around labor and how employers will exploit it as a cost-cutting measure, but as far as this article goes, I don’t buy it. The parents failed their child by not getting him adequate mental health care. The therapist failed the child by not escalating it as a psychiatric emergency. The Game of Thrones chatbot is not the issue here.
Holy fuck, that model straight up tried to explain that it was a model but was later taken over by a human operator and that’s who you’re talking to. And it’s good at that. If the text generation wasn’t so fast, it’d be convincing.
Wow, that’s… somethin. I haven’t paid any attention to Character AI. I assumed they were using one of the foundation models, but nope. Turns out they trained their own. And they just licensed it to Google. Oh, I bet that’s what drives the generated podcasts in Notebook LM now. Anyway, that’s some fucked up alignment right there. I’m hip deep in the stuff, and I’ve never seen a model act like this.
AI bots that argue exactly like that are all over social media too. It’s common. Dead internet theory is absolutely becoming reality.