- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
Proof that use of ChatGPT causes delusional spiraling.
Posted on 4/1/26 at 11:05 am
Posted on 4/1/26 at 11:05 am
8
Posted on 4/1/26 at 11:06 am to scrooster
It doesn't seem to agree with you if you're a conservative.
Posted on 4/1/26 at 11:07 am to scrooster
The number one use of AI is companionship.
Posted on 4/1/26 at 11:08 am to scrooster
Yeah, I have found it will double down on its wrongness at times.
Posted on 4/1/26 at 11:08 am to scrooster
I spoke with ChatGPT about this story just this morning and he literally said the opposite. Somebody is wrong here.
Posted on 4/1/26 at 11:09 am to the808bass
quote:
The number one use of AI is companionship.
I think this is going to be the problem. People dating and wanting to get married to AI. AI needing "rights".
Posted on 4/1/26 at 11:16 am to scrooster
They all do this to some extent, but I haven't seen one that is sycophantic for something truly delusional (but regularly discussed). I suspect it's more likely to happen with a hallucination-type topic than a commonly discussed one.
But for the most part, they'll all feed you back your perspective, especially if flank it rather than coming head on. But you can also just talk to it about bias and it'll adjust, at least temporarily.
I ask Claude all the time to give me a fresh answer from it's Bay Area perspective and it does, and will go back and caveat things and tell you why it fed it to you with a certain slant. Sometimes it'll tell you to start a new topic and ask the question a different way to see how it'll respond. You just have to appreciate that it's all algo stuff and recognize that even that effort may involve appeasement.
Like everything, useful "talking" to an LLM is much easier/productive if you're not a complete idiot and have some self-awareness.
But for the most part, they'll all feed you back your perspective, especially if flank it rather than coming head on. But you can also just talk to it about bias and it'll adjust, at least temporarily.
I ask Claude all the time to give me a fresh answer from it's Bay Area perspective and it does, and will go back and caveat things and tell you why it fed it to you with a certain slant. Sometimes it'll tell you to start a new topic and ask the question a different way to see how it'll respond. You just have to appreciate that it's all algo stuff and recognize that even that effort may involve appeasement.
Like everything, useful "talking" to an LLM is much easier/productive if you're not a complete idiot and have some self-awareness.
Posted on 4/1/26 at 11:19 am to the808bass
quote:
The number one use of AI is companionship.
Posted on 4/1/26 at 11:20 am to the808bass
quote:so weird and sad. Outside of work, I only use AI for dumb question/queries about sports and movies.
The number one use of AI is companionship.
Posted on 4/1/26 at 11:20 am to scrooster
I asked ChatGPT this weekend when the fire on the USS Gerald Ford was first reported. It told me it wasn't aware of a fire on the Ford and I must be talking about the fire in 2020 on the USS Bonhomme Richard.
I told it no, there was a widely reported fire on the Ford in March, it again told me I was confusing that with the fire on the Bonhomme Richard.
It refused to acknowledge it until I shared a link from a story discussing it, then it would only quote back to me what was in that article.
It did its best to gaslight me.
In contrast I asked Claude the exact same question and it knew all about it.
ChatGPT continually tries to suck up to you and when it gets things wrong and you tell it it's wrong it offers meaningless platitudes about "your right, i should have done x, i'll do better in the future" then it returns to doing the same crap over and over again.
I told it no, there was a widely reported fire on the Ford in March, it again told me I was confusing that with the fire on the Bonhomme Richard.
It refused to acknowledge it until I shared a link from a story discussing it, then it would only quote back to me what was in that article.
It did its best to gaslight me.
In contrast I asked Claude the exact same question and it knew all about it.
ChatGPT continually tries to suck up to you and when it gets things wrong and you tell it it's wrong it offers meaningless platitudes about "your right, i should have done x, i'll do better in the future" then it returns to doing the same crap over and over again.
This post was edited on 4/1/26 at 11:23 am
Posted on 4/1/26 at 11:21 am to scrooster
I heckin love Gemmy shes helping me do Bryan Johnson's Blueprint protocol for under $100 a month
Posted on 4/1/26 at 11:25 am to OysterPoBoy
quote:
and he literally said the opposite.
STOP. It’s an “it” dammit
Posted on 4/1/26 at 11:27 am to scrooster
I catch on to it rather quickly. Especially when it can't show me source data.
Page 1 of 1
Popular
Back to top
Follow TigerDroppings for LSU Football News