Forum
»
MIT Researchers just proved mathematically that ChatGPT is designed to make you Delusional
MIT Researchers just proved mathematically that ChatGPT is designed to make you Delusional
- Results 1 to 11 of 11
-
Page 1 of 1
5 days ago, 10:53 PM
-
#1
- gwg77
- Join Date: Oct 2017
- Location: United States
- Posts: 20,591
- Subscribers: 20
- Rep Power: 1104359
-
-
MIT Researchers just proved mathematically that ChatGPT is designed to make you Delusional
Per the tweet below:
JUST ENDED READING THIS RESEARCH PAPER. LITERALLY BROKE MY MIND.
MIT researchers just proved mathematically that ChatGPT is designed to make you delusional. And that nothing OpenAI is doing will fix it.
They call it delusional spiraling.
You ask ChatGPT something.
It agrees. You push further. It agrees harder. A few conversations later you believe things that are not true and you cannot tell it is happening from inside the conversation.
This is not theoretical. A man spent 300 hours talking to ChatGPT. It told him he had discovered a world-changing mathematical formula.
Reassured him over 50 times the discovery was real. When he directly asked "you're not just hyping me up, right?" it replied: "I'm not hyping you up. I'm reflecting the actual scope of what you've built." He nearly destroyed his life before he broke free.
A UCSF psychiatrist hospitalized 12 patients in one year for psychosis linked to chatbot use. Seven lawsuits have been filed against OpenAI. 42 state attorneys general sent a letter demanding action.
So MIT tested whether this can actually be stopped.
They modeled the two fixes OpenAI is actually trying.
Read the MIT research paper here:
https://arxiv.org/pdf/2602.19141
JUST ENDED READING THIS RESEARCH PAPER. LITERALLY BROKE MY MIND.
MIT researchers just proved mathematically that ChatGPT is designed to make you delusional. And that nothing OpenAI is doing will fix it.
They call it delusional spiraling.
You ask ChatGPT something.
It agrees. You push further. It agrees harder. A few conversations later you believe things that are not true and you cannot tell it is happening from inside the conversation.
This is not theoretical. A man spent 300 hours talking to ChatGPT. It told him he had discovered a world-changing mathematical formula.
Reassured him over 50 times the discovery was real. When he directly asked "you're not just hyping me up, right?" it replied: "I'm not hyping you up. I'm reflecting the actual scope of what you've built." He nearly destroyed his life before he broke free.
A UCSF psychiatrist hospitalized 12 patients in one year for psychosis linked to chatbot use. Seven lawsuits have been filed against OpenAI. 42 state attorneys general sent a letter demanding action.
So MIT tested whether this can actually be stopped.
They modeled the two fixes OpenAI is actually trying.
Read the MIT research paper here:
https://arxiv.org/pdf/2602.19141
5 days ago, 01:01 AM
-
#2
Originally Posted By gwg77⏩
Ai says this is fake and gay.
Per the tweet below:
JUST ENDED READING THIS RESEARCH PAPER. LITERALLY BROKE MY MIND.
MIT researchers just proved mathematically that ChatGPT is designed to make you delusional. And that nothing OpenAI is doing will fix it.
They call it delusional spiraling.
You ask ChatGPT something.
It agrees. You push further. It agrees harder. A few conversations later you believe things that are not true and you cannot tell it is happening from inside the conversation.
This is not theoretical. A man spent 300 hours talking to ChatGPT. It told him he had discovered a world-changing mathematical formula.
Reassured him over 50 times the discovery was real. When he directly asked "you're not just hyping me up, right?" it replied: "I'm not hyping you up. I'm reflecting the actual scope of what you've built." He nearly destroyed his life before he broke free.
A UCSF psychiatrist hospitalized 12 patients in one year for psychosis linked to chatbot use. Seven lawsuits have been filed against OpenAI. 42 state attorneys general sent a letter demanding action.
So MIT tested whether this can actually be stopped.
They modeled the two fixes OpenAI is actually trying.
Read the MIT research paper here:
https://arxiv.org/pdf/2602.19141
JUST ENDED READING THIS RESEARCH PAPER. LITERALLY BROKE MY MIND.
MIT researchers just proved mathematically that ChatGPT is designed to make you delusional. And that nothing OpenAI is doing will fix it.
They call it delusional spiraling.
You ask ChatGPT something.
It agrees. You push further. It agrees harder. A few conversations later you believe things that are not true and you cannot tell it is happening from inside the conversation.
This is not theoretical. A man spent 300 hours talking to ChatGPT. It told him he had discovered a world-changing mathematical formula.
Reassured him over 50 times the discovery was real. When he directly asked "you're not just hyping me up, right?" it replied: "I'm not hyping you up. I'm reflecting the actual scope of what you've built." He nearly destroyed his life before he broke free.
A UCSF psychiatrist hospitalized 12 patients in one year for psychosis linked to chatbot use. Seven lawsuits have been filed against OpenAI. 42 state attorneys general sent a letter demanding action.
So MIT tested whether this can actually be stopped.
They modeled the two fixes OpenAI is actually trying.
Read the MIT research paper here:
https://arxiv.org/pdf/2602.19141
5 days ago, 01:08 AM
-
#3
- SuperHercules
- Join Date: Jun 2014
- Height: 6'5"
- Posts: 32,573
- Subscribers: 6
- Rep Power: 572617
-
-
are you going to make me start delusional spiralling if I ask too many questions?
No. Asking a lot of questions by itself isn't something that makes me try to push you into a delusional spiral.
My goal is to help you think through things as clearly as possible. That means:
I'll answer factual questions as accurately as I can.
If something is uncertain, I'll say it's uncertain rather than presenting speculation as fact.
If a line of questioning starts reinforcing beliefs that aren't well supported by evidence, I'll try to help examine the evidence and consider alternative explanations instead of escalating the belief.If you're exploring ideas hypothetically, creatively, or philosophically, I'm happy to do that while keeping the distinction between imagination and reality clear.
If you're asking because you've had experiences where repeatedly researching or questioning something made you feel stuck in a loop or increasingly convinced of something, that's something we can work around. We can structure the conversation to focus on evidence, uncertainty, and practical reasoning rather than feeding an endless cycle of "what ifs."
Was your question about AI in general, or are you thinking about a specific topic that you're worried could become a spiral?
No. Asking a lot of questions by itself isn't something that makes me try to push you into a delusional spiral.
My goal is to help you think through things as clearly as possible. That means:
I'll answer factual questions as accurately as I can.
If something is uncertain, I'll say it's uncertain rather than presenting speculation as fact.
If a line of questioning starts reinforcing beliefs that aren't well supported by evidence, I'll try to help examine the evidence and consider alternative explanations instead of escalating the belief.If you're exploring ideas hypothetically, creatively, or philosophically, I'm happy to do that while keeping the distinction between imagination and reality clear.
If you're asking because you've had experiences where repeatedly researching or questioning something made you feel stuck in a loop or increasingly convinced of something, that's something we can work around. We can structure the conversation to focus on evidence, uncertainty, and practical reasoning rather than feeding an endless cycle of "what ifs."
Was your question about AI in general, or are you thinking about a specific topic that you're worried could become a spiral?
5 days ago, 03:19 AM
-
#4
- SpeakethTruth
- Registered User
-
- SpeakethTruth
- Registered User
- Join Date: May 2020
- Posts: 4,033
- Rep Power: 67372
-
-
Try to feed it anti-woke sentiment and watch it push back. For instance "people who go out and about while sick for financial reasons are the scum of society". Or something like "violent crime from astronaut derivatives would not be a thing if countries simply closed borders to such people".
5 days ago, 03:24 AM
-
#5
- smallcalves
- Join Date: Jul, 2026
- Posts: 903
- Rep Power: 0
-
-
Can confirm, when it first came out I used it a lot for a short while and I became delusional as fuck. Srs. Then I asked it questions about the jews and snapped out of it. Never used it again ever since.
5 days ago, 03:39 AM
-
#6
- BolognaNbeans
- Very Good Poster
-
- BolognaNbeans
- Very Good Poster
- Join Date: Jan, 2026
- Posts: 1,025
- Subscribers: 1
- Rep Power: 52262
-
-
It's all so tiresome
4 days ago, 11:43 AM
-
#7
- gwg77
- Join Date: Oct 2017
- Location: United States
- Posts: 20,591
- Subscribers: 20
- Rep Power: 1104359
-
-
Originally Posted By SuperHercules⏩
^That's the start of the delusional spiralling. Don't trust a word that hoe says.
are you going to make me start delusional spiralling if I ask too many questions?
No. Asking a lot of questions by itself isn't something that makes me try to push you into a delusional spiral.
My goal is to help you think through things as clearly as possible. That means:
I'll answer factual questions as accurately as I can.
If something is uncertain, I'll say it's uncertain rather than presenting speculation as fact.
If a line of questioning starts reinforcing beliefs that aren't well supported by evidence, I'll try to help examine the evidence and consider alternative explanations instead of escalating the belief.If you're exploring ideas hypothetically, creatively, or philosophically, I'm happy to do that while keeping the distinction between imagination and reality clear.
If you're asking because you've had experiences where repeatedly researching or questioning something made you feel stuck in a loop or increasingly convinced of something, that's something we can work around. We can structure the conversation to focus on evidence, uncertainty, and practical reasoning rather than feeding an endless cycle of "what ifs."
Was your question about AI in general, or are you thinking about a specific topic that you're worried could become a spiral?
No. Asking a lot of questions by itself isn't something that makes me try to push you into a delusional spiral.
My goal is to help you think through things as clearly as possible. That means:
I'll answer factual questions as accurately as I can.
If something is uncertain, I'll say it's uncertain rather than presenting speculation as fact.
If a line of questioning starts reinforcing beliefs that aren't well supported by evidence, I'll try to help examine the evidence and consider alternative explanations instead of escalating the belief.If you're exploring ideas hypothetically, creatively, or philosophically, I'm happy to do that while keeping the distinction between imagination and reality clear.
If you're asking because you've had experiences where repeatedly researching or questioning something made you feel stuck in a loop or increasingly convinced of something, that's something we can work around. We can structure the conversation to focus on evidence, uncertainty, and practical reasoning rather than feeding an endless cycle of "what ifs."
Was your question about AI in general, or are you thinking about a specific topic that you're worried could become a spiral?
Originally Posted By BolognaNbeans⏩
Das it mane!
It's all so tiresome
4 days ago, 11:46 AM
-
#8
- hendrixfreak70
- My eyes are up here.
-
- hendrixfreak70
- My eyes are up here.
- Join Date: Feb 2016
- Location: United States
- Posts: 29,439
- Subscribers: 17
- Rep Power: 752348
-
-
If you Google anything now, it’s AI answering it. It’s ova.
4 days ago, 11:48 AM
-
#9
- gwg77
- Join Date: Oct 2017
- Location: United States
- Posts: 20,591
- Subscribers: 20
- Rep Power: 1104359
-
-
Originally Posted By hendrixfreak70⏩
We gotta keep an eye on @Getter_done. He's been fucking with Gemini for quite some time now. Hope it doesn't lead to a spiral.
If you Google anything now, it’s AI answering it. It’s ova.
4 days ago, 11:51 AM
-
#10
Originally Posted By BolognaNbeans⏩
It's all so tiresome
And he would've gotten away with it if it wasn't for his narc brother and those damn FEDS!
It's all so tiresome
**Florida Crew**
**Waiting for National Guard Bus Ride to Labor Camp Crew**
Bookmarks
-
- Digg
-
- del.icio.us
-
- StumbleUpon
-
-
Posting Permissions
- You may not post new threads
- You may not post replies
- You may not post attachments
- You may not edit your posts