mrdlm

coderman (of the systems kind), living statelessly, @iitmadras alum

[mrdlm @VioletStraFish]

@paraschopra but the universe isn't computing itself, it just is

[mrdlm @VioletStraFish]

@mutant1879 it's a decent yardstick, i would say, and compression often makes the prose more elegant; every rule, taken to the extreme, makes the writing sterile, but he is certainly smart enough to know that.

[mrdlm @VioletStraFish]

@malliktwts @paulg @thedankoe i have a writing club (online, whatsapp) wanna join? dm

[mrdlm @VioletStraFish]

@madiator it looks like a bait tweet doesn't it?

[mrdlm @VioletStraFish]

in this era especially, of machine-written prose and machine-written code, one must deliberately do them by hand

[mrdlm @VioletStraFish]

it doesn't matter how much you polish it afterwards, or add a personal touch; it still stinks of ai

[mrdlm @VioletStraFish]

@PythiaR it's a tired joke, it gets annoying. put yourself in his shoes. or maybe you just don't have the context.

[mrdlm @VioletStraFish]

@broadfield_dev @paraschopra implying what? if you run a world-simulator with the same input, it would also, regardless of when you run it, either now or a thousand years from now, run the same

[mrdlm @VioletStraFish]

@paraschopra saying it's merely a next-token predictor is technically true, but it is at a useless level of abstraction. it's like saying the writer is a pen-mover or a singer sound-producer.

[mrdlm @VioletStraFish]

@saidattax @paraschopra what tells you that you know more about what's right than the best llms? would you like to compete with claude and see? or have a crack at a creative problem with a double-blind judge?

[mrdlm @VioletStraFish]

@jaiadityax @paraschopra it depends on what is simulated. for anything informational in nature, such as intelligence, simulation amounts to instantiation; for others, like weather or bridge or consciousness, it never will.

[mrdlm @VioletStraFish]

@paraschopra the class of problems solved by next-token prediction subsumes all of computer science, and more

[mrdlm @VioletStraFish]

@harryjwang you write like ai bro

[mrdlm @VioletStraFish]

@zetalyrae you must be one of those geniuses

[mrdlm @VioletStraFish]

@nivi you know that freewill is an illusion right

[mrdlm @VioletStraFish]

@anieasyy @paraschopra and if it wasn't with some suite of tests, how else were we going to know it's AGI? doesn't mean we can't *talk* about it otherwise

[mrdlm @VioletStraFish]

@anieasyy @paraschopra you *can* test for intelligence (famously)

[mrdlm @VioletStraFish]

@paraschopra everything else about its experience may be unintelligible us; it won't be able to speak of it in a way we can understand. but count, an abstraction over experience, is a unique bridge between our worlds.

[mrdlm @VioletStraFish]

@paraschopra an intelligent system that's conscious can tell how many distinct modalities of qualia it has (smells, taste, vision etc.) we humans have about seventeen. even if AI qualia is alien to us, it should be able to give us a number.