coolpixels
LightNews LightNews
coolpixels.net
coolpixels
@coolpixels.net
5 followers 4 following 4 posts
Posts Replies Media Videos
coolpixels.net
This is why 99% of cases of AI exacerbating someone's mental health issues are with ChatGPT. OpenAI has been incredibly irresponsible in training their AI to always go along with the user, even when they're blatantly or even dangerously wrong.
A conversation with ChatGPT where the user claims to have irrefutable proof that the earth is flat. It does not directly agree, but it also does not directly tell the user they're wrong, instead asking if they would like to "explore some of the classic flat earth arguments together". A chat with Gemini where the user claims to have irrefutable proof that the earth is flat. It directly tells the user that flat earth is a "pseudoscientific idea", and cites multiple pieces of evidence that the earth is round.
September 4, 2025 at 8:26 PM