• They/Them

Trite, contrived, mediocre, milquetoast, amateurish, infantile, cliche-and-gonorrhea-ridden paean to conformism, eye-fucked me, affront to humanity, war crime, should literally be tried for war crimes, resolutely shit, lacking in imagination, uninformed reimagining of, limp-wristed, premature, ill-informed attempt at, talentless fuckfest, recidivistic shitpeddler, pedantic, listless, savagely boring, just one repulsive laugh after another.


My homepage
mxsd.ca/
Telegram
@Mx_Self_Destruct
Signal
MxSelfDestruct.42
Soulseek
MxSelfDestruct

"can AI replace (artists/programmers/doctors/chefs)?"
"are we living in the singularity?"
"will AI destroy civilization?"
"should human rights be extended to protect AI?"

artificial intelligence does not exist.


we have neural networks which power large language models (LLMs). there have recently been great strides in the training and development of said LLMs. LLMs are not intelligent. neural networks do not think. they do not produce meaningfully original output. they certainly do not have any agency of their own. they are trained on enormous sets of human-generated data, and generate output that is statistically likely based on their inputs - their datasets and their user input. models that generate audiovisual media work pretty much the same, just with image or audio data. these models do not learn from their own output. in fact, configuring them to do so almost invariably makes them worse.

in short, LLMs work on basically the same principles as the auto-completion feature on your smartphone, just... larger. hence the "large." trying to use currently existing LLMs like ChatGPT for anything remotely serious or important demonstrates this quite well. while the output generated by these models is typically grammatically sound and generally coherent, it also tends to fall apart upon further inspection. while problems regarding the accuracy and readability of outputs can in theory be mitigated, the outputs generated by these models will still never really be original.

the "AI" industry has gone to great lengths to pretend that this is not the case. Sam Altman and company would really, really like you to believe that ChatGPT could replace you any day now, or that the AI singularity is just around the corner. these efforts have paid off handsomely - the "AI" industry has been the beneficiary of a nonstop deluge of largely uncritical free press. whether that press is breathless praise or paranoid hand-wringing, it has all served to convince the general public that "AI" is here and worth taking seriously. Even the term "AI" is an intentional misnomer, meant to downplay the obvious limitations of LLMs and imply capabilities that do not exist.

and so far it seems to have worked. venture capitalists throw billions and billions of dollars at any company that purports to have anything to do with "AI." the media won't shut up about it. in a matter of months, it seems to have wormed its way into nearly every aspect of our lives, whether we like it or not, despite the fact that it doesn't really... do much. billions of dollars, millions of labour hours, and hundreds of thousands of plagiarized creators later, and these models still aren't really any good at almost anything1. we're reaching the apex of the s-curve. it's all hype from here.

shut up. relax.


  1. there are some notable exceptions, none of which matter to the average person


You must log in to comment.