quotes

Wanna be the very best. Creator of @moxiescout /miw. ゼニゼニ

my core thesis on the whole thing is that these are models trained on text

they don't actually know what they're doing (yet?)

they can't even directly interact with base reality (you could argue this about humans in a pedantic way, maybe)

so anyway they can only regurgitate and infer from the corpus they are trained on, so far

with all that said there are a bunch of super sharp people working to change this. they will

"[llms] can't even directly interact with base reality (you could argue this about humans in a pedantic way, maybe)"

available for product & web design t.me/antimofm

i have to give some credit to @swabbie.eth on this tbh. he helped me sharpen my perspective