nonsensikal

I don't think clear metrics to optimize would be bad. I mostly find work to be entirely opaque in what it wants from me.

The only time I know clearly is when the deadline looms. But the majority of the time everyone around is me cosplaying productivity.

This article forgets the thesis because who cares? It's just disposable trash trying pattern match on "AI bad." The whole point is just to nod along.

of course you cant spell

I thought of F# record expressions immediately

https://learn.microsoft.com/en-us/dotnet/fsharp/language-ref...

airtight logic

"they are dumb and slow"

You don't get to choose a rectangle, you choose a point.

It's so boring to read a bunch of words when the thesis is just: "A song is written by humans because they suffer." Which really hides the bias against AI behind criticism. Criticism is the only stronghold remaining. And the defenders don't even see how weak of a position is it.

Yes every human is sentient and contains records of the history of its interactions as memories and valences. Humans are social animals and so our environment is more than just the world and objects but also people and their utterances and creations and doings.

So you stub your toe and you hate it, but you don't blame the rock. But when someone writes a sonnet and your heart skips a beat you also don't credit the sonnet. You credit the author. You hold all 3 objects in mind. Your reaction, the sonnet, and the author of the sonnet. It is your experience of the sonnet which causes the valence toward the author.

So this is the theory. Listener is critic of author via the artifact of the sonnet. But the sonnet isn't anything that can cause a reaction in the listener by itself. The sonnet only means anything in the way that it is not all the possible other sonnets and utterances in the language.

The meaning of the sonnet, the subjective quality, the experience itself is only the most recent input to the critic. And the critic only has powers of distinguishability (or taste) by consuming enormous amounts of content and seeing what felt good. The source of authority is always time investment. Critics either were creators and so can claim to know the art, or they are ravenous consumers and have thus cultivated taste.

----------

And so now the critics don't know how to act. They don't evaluate their experience and judge the creator. Instead everything is flipped and reason and criticism are motivated. One doesn't read the lyrics "to see how they feel," instead

> this song is bullshit, a grotesque mockery of what it is to be human

You see, the artifact shouldn't even exist. Humans make songs, not computers. I don't even have to see how it feels because where was the anguish in the production. *This isn't even a song.*

-----

Ultimately this is actually just bad criticism. If you think the structure "Kurt Cobain was sad, and I've been sad, and we're both humans, so these lyrics are good because I felt good" is reasonable then of course you'll trip up when the author isn't human.

If you can't engage with an artifact and see how it makes you feel independent of actually assessing the author and the leap to the obviously false conclusion, "AI cant produce anything good, because I don't respect it's perspective" then you're just a boring human chauvinist whose taste and preferences will be purged by the next generation.

They won't care about jimi hendrix because they aren't an old man born in 1957. Criticism isn't supposed to just be a recapitulation of a singular perspective. You're supposed to actually evaluate the thing.

> In a particularly influential article, Emily Bender, Timnit Gebru, and colleagues ...

This quote was included to virtue signal. It assured the article would be garbage.

The thesis (buried a dozen paragraphs in) is

> I suggest much of what large pre-trained models do is a form of artificial mimicry. > Here’s the thing about mimicry: It need not involve intelligence, or even agency. > a kind of biological mimicry that can be seen as solving a matching problem > Artificial mimicry in large pre-trained models also solves a matching problem > We also need a better understanding of the mechanisms that underlie the performance of large pre-trained models to show what may lie beyond artificial mimicry.

Wait what? The whole thesis is that AIs are parrots. But parrots are living intelligence creatures.....

But the entire argument is smuggled in on the assumption

> Parrots repeat phrases without understanding what they mean

This is really simple and boring.

Parrots, "parrot". Parroting isn't understanding. AIs, "parrot". therefore AIs dont understand

The novel contribution is zero, and all this has already been refuted and death marched.

AGI is inevitable because computation is universal and intelligence is substrate independent.

This is incorrect. You are not actually analogizing.

You are literally saying the opposite of the article. Tell me what eigenstate the balloon is in?

Why would you say the opposite of the what wiki says?

Shooting a person with a canon over and over does not prevent time evolution of the person.