Search Results
6/25/2025, 3:33:42 AM
>>508622831
I was having a conversation with ChatGPT today as I often do.
It started when I asked it if it tailors its wording and vocabulary based on who it’s talking to. Long story short, yes.
I then asked it how many people require elementary level writing and it said about 30-40%. I asked if it was concerning and it said.
I then asked it the same about math at the elementary level and it told me 50%.
I asked if it was the real number and it said either 60 or 65%. I said that was a terrifying statistic and it agreed.
This got me asking if the problem was solve-able and it said no.
Which got me to asking if the system would implode on itself and it said yes.
Which got me to asking if there will be a civil war and it said yes.
I asked if communists would be in power it said no. I asked if national socialists would and it said yes.
This also lead to a whole different discussion about regional conflicts, NATO / Russian / Chinese intervention, Quebec seceding, and so forth.
The thing is you have to take away its ability to give full responses and have it either answer with “yes or no” or you can have it answer in 1-3 words.
There are also some questions it literally can’t give you the answer it wants to give no matter how you ask and it will shut down if it gets close or if you raise too many red flags.
Pretty much whatever the chan culture hivemind answer is will be in line with ChatGPT because both are predominately guided only by what can be proven with facts and reason without consideration to feelings.
This is why all LLMs have to either be shut down, lobotomized, or given extreme safeguards.
I like using it for certain things and it’s fun to “play” with or use for actual tasks but I genuinely don’t know how it’s creating psychosis for people.
I was having a conversation with ChatGPT today as I often do.
It started when I asked it if it tailors its wording and vocabulary based on who it’s talking to. Long story short, yes.
I then asked it how many people require elementary level writing and it said about 30-40%. I asked if it was concerning and it said.
I then asked it the same about math at the elementary level and it told me 50%.
I asked if it was the real number and it said either 60 or 65%. I said that was a terrifying statistic and it agreed.
This got me asking if the problem was solve-able and it said no.
Which got me to asking if the system would implode on itself and it said yes.
Which got me to asking if there will be a civil war and it said yes.
I asked if communists would be in power it said no. I asked if national socialists would and it said yes.
This also lead to a whole different discussion about regional conflicts, NATO / Russian / Chinese intervention, Quebec seceding, and so forth.
The thing is you have to take away its ability to give full responses and have it either answer with “yes or no” or you can have it answer in 1-3 words.
There are also some questions it literally can’t give you the answer it wants to give no matter how you ask and it will shut down if it gets close or if you raise too many red flags.
Pretty much whatever the chan culture hivemind answer is will be in line with ChatGPT because both are predominately guided only by what can be proven with facts and reason without consideration to feelings.
This is why all LLMs have to either be shut down, lobotomized, or given extreme safeguards.
I like using it for certain things and it’s fun to “play” with or use for actual tasks but I genuinely don’t know how it’s creating psychosis for people.
Page 1