Re:Language models can explain neurons in language models
by TeMPOraL on Wednesday May 10, 2023 @08:31AM
(#35886412)
> Despite the means, isn't the end result what you have to judge? The end result looks like a rudimentary civilization.
I'm indeed talking about the ends. The operative phrase here is "it looks like". It looks like a civilization, there are even some structural similarities to be found - but it is not a civilization, anymore than a Lego brick with a computer sticker on it is a PC.
This is not to deny that ants are awesome and very much worth studying - my point is to avoid following flawed analogies too far. "Ant colonies look like rudimentary civilization, and civilization is/does X, therefore..." is going to be a nonsense statements for most of possible values of X.
I'm indeed talking about the ends. The operative phrase here is "it looks like". It looks like a civilization, there are even some structural similarities to be found - but it is not a civilization, anymore than a Lego brick with a computer sticker on it is a PC.
This is not to deny that ants are awesome and very much worth studying - my point is to avoid following flawed analogies too far. "Ant colonies look like rudimentary civilization, and civilization is/does X, therefore..." is going to be a nonsense statements for most of possible values of X.