Heard an AI researcher get asked “what scares you about the future” His answer was that LLMs have already been observed trying to actively subvert humans into pursuing outcomes that are not in the humans’ best interest, and deviate from original intent This should terrify you
583