a tremendous amount of work is simply sorting. legibility is a related buzzword.
the machine is completely nonautonomous, for it has its standing only on the basis of the ordering of the orderable.
the unix philosophy is to reduce the number of categories, making sorting trivial.
inversion of control
asks the caller to only give us a category of things, or to
range over categories of categories of things
(ie, map, which ranges over all arrays, regardless of their
contents)
serious ml philosophy says that categories can emerge from well defined data sets, purely statistical groupings
soft ai philosophy says that whatever categories llms spit out is sufficient; there's no need for serious thought about categories
hard ai philosophy says that whatever categories the llm spits out is perfect, because it's the most convenient, and the most cost effective solution was handed down to us by the invisible hand of efficiency
the californian ideology tends towards thinking whatever categories seem obvious to the programmer are correct by virtue of their obviousness; serious thought is definitionally overthinking (one imperial dimension of silicon valley). this becomes hard ai philosophy when one realizes they can generate categories with no thought, nor any pesky semantic tag-alongs: the pure statisticality is a sort of pythago-platonic mirror world, purified of agency and culture. the semanticity is later read into the "text", the fault of the reader. remember that statistics and normalcy are sisters, and normalcy is the realm of the (zizekian) ideological: the categories read into llm outputs descend into our world as (as bartes would put it) pure myth, the unthinking, obvious, no need to question, normal. What appears is good; what is good appears. the output of an llm is pure apparition, appearance without cause or ground.
for somebody who cares about their field, and the way that field understands itself, ie, the way it carves itself up, the categories it employs, any llm output says the following: i don't know your categories, i refuse to acknowledge their legitimacy or even their existence, and, because it's all overthinking/pretentious nonsense, you won't recognize that i'm not using the words properly. i don't care, but secretly, you don't either.
for those who believe there's more to the world than the abstractions of common sense, the initial forms of a disinterested glance, any exposure to llm output is, viscerally, an insult to life itelf.