Yann LeCun, VP & Chief AI Scientist at Meta, made an interesting post on LinkedIn about AI world models: Post | Feed | LinkedIn.
The short post may just be one of the most important posts on AI.
“World models should not be generative.”
I always knew this was the case, but it is great to see empirical confirmation. If one is willing to go beyond the mere empirical, he may start to see the ontological meaning of the universe being revealed, AI being just an interesting illustration.
In the real sense of the word “generative”, not just world models but all AI models should not be generative. In fact, they cannot be generative in the real sense.
Human minds are generative not because they are computationally superior (they are not, as far as the formal algorithmic logic is concerned, and are left behind even further with time), but because they are not closed systems. There is inspiration coming from outside of the material brain, and it’s the inspiration that is truly generative.
The triune nature (spirit, soul, and body) of a human being is real, not a mere religious teaching.
Reality is a selective subspace of the infinite information space of unlimited mathematical possibilities.
Reality is a very special convergence (or collapse) of an unlimited mathematical space. It is this convergence that provides the meaning of existence.
This is because the convergence is not random but is selective.
The meaning is in the selection. Selection means purpose. Purpose means a Mind behind it. Unless we believe we are the Master of the universe (I know I’m not), we are the object of the Purpose, not the subject of it.
Therefore, choose wisely. Our subjective feeling is not necessarily the right meaning.