> What you describe as pausing and thinking is exactly letting your circuits run for longer - which again - is a decision made by said circuits who then informe

> What you describe as pausing and thinking is exactly letting your circuits run for longer - which again - is a decision made by said circuits who then informed your internal time keeper that "you" made said decision. Yup, and it feels like ChatGPT might be able to approximate this by giving the model a "think longer" output that feeds the output back into itself. I'm actually curious if immediately prompting the model "are you sure" or something else a few times could get you a similar effect right now.