Do Models Doubt?
psychologytoday.comi wonder how many psychology today articles have been shared to hacker news ... my reason for posting is:
maybe computation IS the same thing as doubt? or at the very least, doubt is an ingredient that goes into making a computation. it sure feels similar to me.
> i wonder how many psychology today articles have been shared to hacker news
You can see submissions from a domain by clicking the domain’s name next to each submission’s title. psychologytoday.com: https://news.ycombinator.com/from?site=psychologytoday.com
surely you can almost hear it. “As a language model, I do not experience doubt like humans do.”
some take that as proof that no doubt is experienced. others, that no human doubt is experienced. there’s a chasm in the middle.
these words are all malleable as they are but crude carriers of shadows of ideas.
To me, doubt is the pause that occurs when the negative consequences of an action outweigh the probable success. So it's a probability threshold