3DBeerGoggles

AI "art" does not create, it imitates and it regurgitates.

Humans made art before art existed, but a diffusion model could not make art without being shown millions of examples of art first.

Anyways, as much fun as it is reading "AI art copium 101", it's getting a bit tedious so I'm going now.

But I like how you've now pivoted between the extremes of "AI is real cognition for making art" and "AI isn't making art, it's actually the guy punching "Cute girl 4K sunset cyberpunk neon art motorcycle posing High Quality Portrait" into stable diffusion that's the artist.

...and with the reliability of the sun in the East, we're back to "Actually, cribbing off of millions of images to statistically generate something is exactly like using a paintbrush"

Yeah sure, and if I screenshot the results from a google query I'm an artist.

Clearly this conversation is wasting both of our time, because you want to conflate "it makes the product" with being the same a person making something.

Wolfram alpha is not a mathematician, Excel isn't an accountant, and Stable Diffusion isn't an artist.

...and again, you conflate "output" with art.

A diffusion model doesn't have intent, it has statistical models. It doesn't have opinions, intents, preferences, or a personal style.

Again, the comparison is reductionist to the point of making art itself worthless, which is a theme for AI techbros.

My critique was responding to the notion that CURRENT AI systems and architecture were at all comparable to an actual thinking being rather than what they are: a collection of statistical models

Saying "well some day we might emulate a brain" wouldn't retroactively make Stable Diffusion a person

I'm talking about reality, today, not speculative what-ifs about some day building a 1:1 neuronal model of an actual thinking creature and getting it to make art for me.

1) You linked a locked article, and the article I could find about it didn't mention emulation
2) This still has fuck all to do with GENERATIVE AI.

I didn't say "it's literally impossible to ever emulate neurons" I said it's not how genAI works.

Comparing the current state of GenAI to this is BS

Actually, on that topic - with that whole "looks like an answer, but the source doesn't back it up" topic, are you perchance just having an AI write these responses for you?

Also, and I can't believe I have to point this out:

1) It's a map, not an emulator.

2) Fruit flies don't have opinions about art.

That you thought this proves anything in the topic at hand feels more like you're just throwing anything that is vaguely proof-shaped at the screen.

No, what's "overly simple" is conflating something like diffusion models with the complexity of a biological brain.

And conflating leading-edge research into modelling the least complex biological systems with current GenAI ranges between dishonest and delusional.

"Guy that works for Google overstating parallels between social cohesion and AI models" engaging in vague handwaving and sophistry strains that "may" in the headline to the breaking point.

Neither this nor Eliza-effect hallucinations mean genAI has actual opinions.

Daily affirmations:

"I have what it takes to cause cancer and reproductive harm"

"I am recognized by the state of California"

....and that aside, if you were person looking at a million pictures of a hat, you're STILL making intentional choices as to what you make a picture of.

A human artist is acting with intent, not a statistical model.

This is such a reductionist take on the human brain I'm not sure if I'm more bothered by grossly overstating what genAI is doing or grossly understating how complicated the human brain is.

Neural nets are a shortcut to emulating some things a brain does, it's not an actual brain.

There's as much artistic intent in a prompt as there is sending an email to commission an artist to paint you something.

Or, for that matter, to perform a google search. It too, is using an algorithm to generate a statically-significant match to your query.

The fact that, ethically, AI isn't some little artistic hobby where someone "rolls their own" to make variations of their own work but instead a multi-billion-dollar economic flywheel based on stealing work of actual artists, only reinforces my opinion that it deserves to fail and is a net negative

The "what is art?" debate will never end, but to me, art is at minimum 'intent'.

A splash of paint can be a spill, or it can be Jackson Pollock. A gold-plated toilet can be a gauche decorating choice or it can be a statement.

And as far as I'm concerned, AI images are too divorced from intent.

I think the problem is that it assumes that art is just a thing people can look at. It is merely OUTPUT.

I don't do photography because I want to see a picture of an eagle. I do photography because I wish to find an eagle and try and capture its beauty. I want to share some of what my eye sees.

If I built my own camera and did nothing but take pictures of other people's work, I'm not an artist, I'm a xerox.

Unless you're making a model trained entirely on your own work, anything coming out of that model is just stolen credit.

The way I see it, I can build a camera. I could, given the motivation, even coat my own glass plates with emulsion and process them.

I can plan, scout, compose, and expose the photograph I want to make.

I can do all of that without running every photo in history through a blender.

"Some people were wrong about something else, that means I must be right today" always feels like one of those last-resort arguments, like a conspiracy theorist insisting we'll all feel stupid when they're proved right about the moon being made of cheese.

Kinda like how they managed to turn Saturn from a successful subsidiary into yet another tombstone: execs that always need to be seen "doing something" without care for what the something actually is...

I think Wil is really overestimating the chances that manufacturing and maintaining an entire separate model line of under-sized, non ADA-compliant taxis with overcomplicated doors is going to make money on marginal energy savings.

The cybertaxi simultaneously claiming it's ready for prime-time while alienating anyone that has more than one friend and is good physical condition.

If there was ever an example of Techbros building for Techbros, it's this.

Really not shaking the allegation that Tesla stans don't want to engage with the real world there, Crissa

Tesla Stans really have to maintain that bubble, don't they?

"They're behind on the qualification. Not the tech."

Big "I totally could've aced that test if I actually cared enough to show up to class" energy.

Yeah, it's only like the third time he's announced plans for a bus, I'm sure he's totally going to do it this time.

Dude is like Lucy holding the football.

It's on the shelf next to the stainless-steel "exoskeleton" truck (still love that they completely rebranded stressed-member design and STILL failed to deliver on it...)