Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why A.I. Isn't Going to Make Art

newyorker.com

Post reply on HN
Print view View original

81–90 of 151 posts

Re: Why A.I. Isn't Going to Make Art

#81

Post by gmaster1440 »

ben_w wrote: post #76
gmaster1440 wrote: post #70

Earlier quoted context omitted.

Human beings are currently capable of productively searching through the space of possible knowledge and experience in ways no current AI systems are. This is not to say AI will never do this, but I think it's fair to say there are things human beings are capable of doing today that AI is not, and it remains very much unclear whether AI will ever be able to achieve important milestones like being conscious in the sen…

> Human beings are currently capable of productively searching through the space of possible knowledge and experience in ways no current AI systems are. I would say the opposite: AI are very good at searching through information spaces, much better than we are. They're terrible at learning from experience, the points made in the article about that are I think valid, but they're wildly super-human at searching. > like…

> I would say the opposite: AI are very good at searching through information spaces, much better than we are.

We are likely talking past each other here. By "searching" I don't mean how inference is currently carried out by efficiently analyzing the context window using weights trained on large data sets fine tuned on specific goals.

I mean the process by which novel information is discovered, which is why many proponents of AI will concede that it's not currently capable of "doing science" or making novel discoveries.

> we don't know what that is, it's just what we do.

Not sure I understand, we have a pretty good understanding of what qualia actually is, even if it can be difficult or awkward to talk about conceptually. The gap between having a subjective experience and not having one is a large one, just ask anyone who's alive but under general anaesthesia that induces loss of consciousness. Qualia is simply what arises from the quality and character of having a subjective experience.

Re: Why A.I. Isn't Going to Make Art

#82

Post by ben_w »

CuriouslyC wrote: post #10

I'm disappointed in Ted here. For a writer that likes to delve into the possibilities of tech, he's sharing a surprisingly underbaked view. People outside of tech seem to think that AI creation is just "fat finger a prompt -> take output and claim to be an artist on the interwebs" but the reality is that all the people I know who actually call themselves AI artists do photobashing, image2image, controlnets, inpaintin…

I believe he demonstrated awareness of the difference between lazy use and effortful use; he appeared to me to acknowledge the latter as art:

"""The film director Bennett Miller has used dall-e 2 to generate some very striking images that have been exhibited at the Gagosian gallery; to create them, he crafted detailed text prompts and then instructed dall-e to revise and manipulate the generated images again and again. He generated more than a hundred thousand images to arrive at the twenty images in the exhibit."""

Re: Why A.I. Isn't Going to Make Art

#83

Post by AlienRobot »

If you ever wonder why everything made with Stable Diffusion looks the same, it's because it can't generate images that are too dark or too bright. The denoising process involves recognizing shapes and shapes naturally have bright spots and dark spots. If you try to render "sea at night" you'll get some huge, bright moon for example.

The "AI artists" using this tool lack the technical and artistic competency to realize this. They didn't write the algorithm, draw the dataset, or train the model. They prompted. They have the smallest amount of creative input into this whole pipeline.

I do believe AI can be used in the process to create art as it's just an image generator like fractal art, but the problem is most people are going to use AI not as a means to create art, but as an end. You could fix the problem above by simply importing the image into GIMP and changing the brightness, but nobody does that because they aren't interesting into creating an art piece with a set goal in mind, they're just being entertained by generating dozens or hundreds of images with this technology.

Amusingly, you could also just type text in GIMP. Instead there is now something called "flux" that can do text literals.

While I see the point in making a prompt interpreter capable of generating text literally, if I were creating something, I wouldn't let an AI randomly pick a font, color, weight, serifs/slabs, etc. for me. These are creative choices in design that make all the difference. Prompting gives the illusion of (creative) choice. You get something that looks good, but "getting something that looks good" is the default state. Anyone can do that. It's the AI art equivalent of drawing a stickman. The prompters just don't realize it because they're comparing themselves to to artists of other media, not comparing themselves to other AI artists.

When everything is AI, and anyone can generate an image with a prompt, the whole market will be so saturated with this (perhaps it already is at the rate these are generated) that all the novelty will be gone.

It was cool when AI was able to generate video, just like it was able to generate text. But in my opinion, those are feats of the technology, not artistic feats. The piece itself isn't interesting. It could be any video. Just the fact that the tech can do this is impressive. But it's just the tech that is impressive, not its output. Once the tech can do it once, it can do it every time, so the second time AI generates video is never going to be as impressive as the first time. By the thousandth time it will be as impressive as my ability to send this message to the other side of the world at the speed of light.

Re: Why A.I. Isn't Going to Make Art

#84

Post by dennisalbie5 »

gmaster1440 wrote: post #81
ben_w wrote: post #76

Earlier quoted context omitted.

> Human beings are currently capable of productively searching through the space of possible knowledge and experience in ways no current AI systems are. I would say the opposite: AI are very good at searching through information spaces, much better than we are. They're terrible at learning from experience, the points made in the article about that are I think valid, but they're wildly super-human at searching. > like…

> I would say the opposite: AI are very good at searching through information spaces, much better than we are. We are likely talking past each other here. By "searching" I don't mean how inference is currently carried out by efficiently analyzing the context window using weights trained on large data sets fine tuned on specific goals. I mean the process by which novel information is discovered, which is why many prop…

[dead]

Re: Why A.I. Isn't Going to Make Art

#85

Post by CuriouslyC »

slowhadoken wrote: post #65
CuriouslyC wrote: post #13

Earlier quoted context omitted.

Art is perceived by people. It's not complicated.

Your statement isn’t complicated but it’s also not logical or accurate. If art is what is perceived by people then everything is art and art is a redundant term for everything.

My statement wasn't complicated, yet somehow you still didn't understand it, or you're purposefully misunderstanding it to try and strawman.

Re: Why A.I. Isn't Going to Make Art

#86

Post by doph »

swayvil wrote: post #67
gmaster1440 wrote: post #35

It's refreshing to see a science fiction writer underplay the capabilities of AI, but if anyone can speak to the nuances and implications of generative AI on art and writing it's probably someone like Ted Chiang. We can debate his generalized definition of art as making creative choices that carry subjective, intentional, and performative value for human beings (and therefore LLMs fall short of this), but I think he…

We should ask an artist if it's art. The testimony of an expert firsthand witness is vastly superior to interpretation of a secondhand abstraction by inexpert persons.

I'd argue that an LLM has a fuller understanding of what we mean when we say "art" than most individual users of the term (Chiang's definition sounds more like the one for "craft" to me). Indeed, asking ChatGPT about the difference between art and craft gives me a more nuanced contemplation on the subject than I found in the article. Maybe these models are closer to the experts than we are willing to give them credit for? At the vet least, they are the best mirror we've ever invented.

Re: Why A.I. Isn't Going to Make Art

#87

Post by ben_w »

gmaster1440 wrote: post #81
ben_w wrote: post #76

Earlier quoted context omitted.

> Human beings are currently capable of productively searching through the space of possible knowledge and experience in ways no current AI systems are. I would say the opposite: AI are very good at searching through information spaces, much better than we are. They're terrible at learning from experience, the points made in the article about that are I think valid, but they're wildly super-human at searching. > like…

> I would say the opposite: AI are very good at searching through information spaces, much better than we are. We are likely talking past each other here. By "searching" I don't mean how inference is currently carried out by efficiently analyzing the context window using weights trained on large data sets fine tuned on specific goals. I mean the process by which novel information is discovered, which is why many prop…

> We are likely talking past each other here. By "searching" I don't mean how inference is currently carried out by efficiently analyzing the context window using weights trained on large data sets fine tuned on specific goals.

> I mean the process by which novel information is discovered, which is why many proponents of AI will concede that it's not currently capable of "doing science" or making novel discoveries.

Huh. I think those are the same thing?

But then, I say that AI can do science. Not that I would recommend specifically an LLM for this, but what was AlphaFold doing if not science? Or even GOFAI having been used for the four colour theorem back in the day.

> Not sure I understand, we have a pretty good understanding of what qualia actually is, even if it can be difficult or awkward to talk about conceptually.

Hm. How to rephrase…

Can you create a testable definition of it?

> The gap between having a subjective experience and not having one is a large one, just ask anyone who's alive but under general anaesthesia that induces loss of consciousness. Qualia is simply what arises from the quality and character of having a subjective experience.

Purely from asking them questions, how will you differentiate between each of these cases?

1) A person under general anaesthesia that induces loss of consciousness

2) A person under the influence of a paralytic agent without anaesthesia, who is fully aware of their surroundings but unable to respond

3) A brain-dead person

4) A person with locked-in syndrome

5) A person in REM sleep who is currently dreaming but unaware of the surrounding real world

6) A person in deep (non-dream) sleep who also has no awareness of the surrounding real world

7) An unborn foetus (any species)

There are people with locked-in syndrome who later recover, who report that those around them treated them as if they were non-conscious.

Re: Why A.I. Isn't Going to Make Art

#88

Post by doph »

tkgally wrote: post #39

For an essay that hinges on the notion that “art requires making choices,” I wish the author had chosen to delve a bit more into the question of what choosing is. Do humans really make choices? If so, how free are those choices? Is there something about human choice that will never be convincingly imitated by computer? If so, what is it, and how do you know it cannot be imitated?

Exactly my thoughts. I think people who make arguments like Chiang's are unwilling to examine our own decision making process, and in particular are unwilling to entertain the idea that it is as mechanistic as an artificial neutral network.

Re: Why A.I. Isn't Going to Make Art

#89

Post by slowhadoken »

CuriouslyC wrote: post #85
slowhadoken wrote: post #65

Earlier quoted context omitted.

Your statement isn’t complicated but it’s also not logical or accurate. If art is what is perceived by people then everything is art and art is a redundant term for everything.

My statement wasn't complicated, yet somehow you still didn't understand it, or you're purposefully misunderstanding it to try and strawman.

Art is made by people therefore not everything we perceive is art. It’s not personal.

Re: Why A.I. Isn't Going to Make Art

#90

Post by gmaster1440 »

ben_w wrote: post #87
gmaster1440 wrote: post #81

Earlier quoted context omitted.

> I would say the opposite: AI are very good at searching through information spaces, much better than we are. We are likely talking past each other here. By "searching" I don't mean how inference is currently carried out by efficiently analyzing the context window using weights trained on large data sets fine tuned on specific goals. I mean the process by which novel information is discovered, which is why many prop…

> We are likely talking past each other here. By "searching" I don't mean how inference is currently carried out by efficiently analyzing the context window using weights trained on large data sets fine tuned on specific goals. > I mean the process by which novel information is discovered, which is why many proponents of AI will concede that it's not currently capable of "doing science" or making novel discoveries. H…

AlphaFold is really impressive and made scientific advancements and discoveries in the field of protein folding, and is now even expanding into more molecules and biology, but it was explicitly trained to do just that. You're not going to see AlphaFold write compelling science fiction.

We can build models that are specifically trained and fine tuned on scientific fields to make advancements in them, but that's different from what I'm talking about, which is building a model that forms its own hypothesis, designs its own experiments, and contributes to the wide and deep wealth of knowledge that, crucially, goes well beyond the scope of its training data.

Post reply on HN