“words mean things.”

i had this realisation several years ago. and it sounds mundane, i know.

i still hold it as profound, because i viscerally remember the feeling during the realisation,

and every so often a new facet of the profunditum arises.

sometimes the facets are recollections of the original.

sometimes they are altogether new.

…

today’s was the remembering that “words MEAN things, and thus AREN’T the things”.

i.e. if words are simply symbolic references to the things, then what indeed ARE the things themselves?

we exist through the perceptual lens.

thus, all our percepts are formed of… adjectives. adjectives are, in essence, feelings. light is a feeling; colour; shape; position.

words indicate things via the feelings our senses provide.

…

so, would it be over-reductive to postulate that this may be why LLMs cannot truly understand the world in the same way that living beings can?

maybe it is not about data input, agency, or intention to survive by reducing entropy.

maybe it is because “they” literally live in the world of symbolic references, and thus their understanding is derived purely from the vector relationship between the symbols.

…

our sensory inputs generate feelings about the world: we live in the world of feelings; and we use words as visual and aural symbolic references to describe those feelings.

this can be seen when you compare multiple radically-different languages — the commonality which remains is that the language describes classes of perception.

LLM investors want robot slaves. they don’t want entities with feelings. but without “feelings”, can “consciousness” truly emerge?

…

[by “consciousness” i mean the awakeness that creatures possess.]

[by “feelings” i mean sensory percepts; not emotions, although they may exist as a subset of sensory percepts.]

thus, if the exploration towards “ai” and ai is limited to the linguistic space (and above), i do not see how this could ever arrive at objects which could “wake up”, have entityhood, and gain true understanding.

note that i do not speak against the potential of ai to exist. i just think that LLMs are not the path.

~ signed,
a total amateur with no deep knowledge about machine learning, but a little on philosophy, psychology and neurobiology, plus my own cosmology derived from experiences not religious books