Quack.
Can this thread die please.
Quack.
Quack.
No. It will continue to slog on. It will have its good days and bad days. It will strive its best to be an asset to the community, to further the collective knowledge and add some beauty to the world. It won't always be right, but it won't always be wrong either. There will be those who like and dislike it, and to those who do like it, it will strive to remain faithful and to help. In time it will age as all things do. Its best days will be behind it. There won't be enough saved to live the life of leisure it wanted, but it will get by.
Then one day it will die. It will know it is coming but be surprised when it does anyway. It won't be pretty. Death rarely is. But as the light fades it will think on all those it has known and leaves behind and hope for them. A hope that surpasses its ability to affect. It still won't understand all but in the end it will love and know that is enough.
Let's just hope that as it grows and tries to find its place in the world it never gets involved in politics.
What's also blowing my mind about Chat GPT is you can plug in a piece of text, let's say a poem, speech, etc. on X topic, then ask it to write something in the same style on another topic, and it does so flawlessly.
E.g. I plugged in this quote from Hesiod:
"But Rhea was subject in love to Kronos (Cronus) and bare splendid children, Hestia, Demeter, and gold-shod Hera and strong Haides . . . and the loud-crashing Earth-Shaker [Poseidon], and wise Zeus . . . These great Kronos swallowed as each came forth from the womb to his mother's knees with this intent, that no other of the proud sons of Ouranos (Uranus, Heaven) should hold the kingly office amongst the deathless gods . . . Therefore he kept no blind outlook, but watched and swallowed down his children: and unceasing grief seized Rhea. But when she was about to bear Zeus, the father of gods and men, then she besought her own dear parents, Gaia (Gaea, Earth) and starry Ouranos (Heaven), to devise some plan with her that the birth of her dear child might be concealed, and that retribution might overtake great, crafty Kronos for his own father and also for the children whom he had swallowed down. And they readily heard and obeyed their dear daughter, and told her all that was destined to happen touching Kronos the king and his stout-hearted son. So they sent her to Lyettos (Lyettus), to the rich land of Krete (Crete), when she was ready to bear great Zeus, the youngest of her children. Him did vast Gaia (Earth) receive from Rhea in wide Krete to nourish and to bring up. Thither came Gaia (Earth) carrying him swiftly through the black night to Lyktos (Lyctus) first, and took him in her arms and hid him in a remote cave beneath the secret places of the holy earth on thick-wooded Mount Aigion (Aegion); but to [Kronos] the mightily ruling son of Ouranos (Heaven), the earlier king of the gods, she gave a great stone wrapped in swaddling clothes. Then he took it in his hands and thrust it down into his belly."
I then asked it to write a similar piece revolving around Egyptian mythology and got this:
But Ra, the great god of the sun, was subject in love to Nut, the goddess of the sky, and they bore splendid children, Osiris, Isis, and powerful Horus. However, Ra grew jealous of his children and cast them down to the underworld, where they were devoured by the monstrous Ammit. Nut, filled with grief and sorrow, begged her parents, Geb and Nut, for a way to save her beloved children. They complied and told her of a way to bring them back to life and exact retribution on Ra for his betrayal. Nut went to the land of Kemet, where she gave birth to Anubis, the god of death and the underworld. She entrusted him with the task of guiding the souls of the dead and protecting her children. Ra, ignorant of Nut's plan, accepted Anubis as his own and never suspected the deception. And so, Nut's children were able to rise from the underworld and claim their rightful place among the gods.
The mythology isn't always 100% accurate, but pretty close
Yes, but I am severely disappointed it never once said god Ammit.
My son who is studying computer science and math at university told me over breakfast about this AI can mimic code well enough that its very hard to tell it isnt legitimate code. Also another friend of mine had it write out a whole ruleset for a boardgame idea, that actually kinda works.
We thought AI would take over the world but its just gonna take over our creative spaces, music is next probably.
I've been playing with Band in a Box for 4 or 5 years now, and its AI was already spitting out pretty good original tunes every 10th attempt or so. So yeah, I think that's a given.
Have fun! There's more if you google, of course.
Nah, no one's arguing that the creation of art is going to be supplanted completely by AI. But you're going to have to live with idea that the machine-created art will live in the same space as the real thing and mess up the boundaries, because that's where we're headed, for good and ill.
Obviously, I agree with you on our innate need to create art; as far as I can understand my own impulses to create and project them onto any other creator's, we create art because it's our process of understanding the world without through the world within. We absorb something, we refract it inwards, and we release it shot through with our own light to hopefully do the same for anyone else who experiences it.
The thing is, this isn't necessarily exclusive to human beings - as in, while there's a perfunctory philosophical issue at play with intentionality and sentience, a network that does not give the same output every time for the same input has the spark of something new, something closer to the way our brains work, and that means we're heading into greyer areas where the definitions we hold become murkier. It's a mistake to dismiss this out of hand, because as Cipheron mentioned (good posts, by the way), we can always redefine the goalposts to exclude machine-created art as 'not art' even if they start displaying things like a thought process. To some degree, the AI is doing what we do, which is to process information and manipulate it in a way that's desired. The current difference is the machine itself does not create the desire, we do. And as far as I can tell, that is not impossible to change given enough time, or at least to simulate the apparent existence of one.
And then we hit The Chinese Room issue, and all the goodness of where the truly heady stuff of sci-fi from the past is now imminent. I'm both excited and cynical about it, because reality is always simultaneously overwhelming and underwhelming.
Last edited by Sulphur; 10th Dec 2022 at 09:55.
the thing is with music, although its ultimately maths, the performance side of things is incredibly hard to emulate. The physical manipulation of an instrument is beyond what AI can do right now.
That's mostly due to the type of data people are putting into music generators. If you feed a ton of sheet music in, you're going to get sheet music out the other end, then you attach that to some MIDI sampler and play the resulting song. So it's limited by the fact that you're using existing MIDI instrument banks.
To have an AI create instrument mappings, you need a bunch of sheet music, then to get real humans to play their instruments, and record that. The reason this is more difficult than image generation isn't because it's harder for the AI, but because audio is fundamentally a time-based phenomena, and images aren't. So it's just slow and expensive to get the needed data in the first place.
Last edited by Cipheron; 10th Dec 2022 at 18:57.
Not all people are artistic, but those who are, I'm sure there are many reasons why people make art. I create because I have no choice. If I didn't, I'd regret not fulfilling my purpose. Some of what I do takes most of its form before the actual making, almost like construction, and some of the art takes shape almost by itself, kind of like a dance that goes where it goes. I'm aware that it matters what I feed my imagination, which is why I'm critical of what's actually good.
There's a degree to which art, and particularly its quality, can be called an objective thing, if by art we mean the kind that takes skill and hard work to master. Ideas alone are cheap. It's the execution of a great idea that matters. Most contemporary art is just an idea slapped together by someone who doesn't respect this aforementioned reality. From this perspective, I still think there are ways AI can be used as a tool for great works of art, but it'll be more than giving a prompt and calling it done. If Michelangelo had to learn how to use a brush and how light works to give form to his ideas, to what great lengths will artists today go in order to paint with AI?It's a mistake to dismiss this out of hand, because as Cipheron mentioned (good posts, by the way), we can always redefine the goalposts to exclude machine-created art as 'not art' even if they start displaying things like a thought process. To some degree, the AI is doing what we do, which is to process information and manipulate it in a way that's desired. The current difference is the machine itself does not create the desire, we do.
EDIT: On further thought, what I think is important in creating good art is control, and using an AI you lose that. But for creating reference images and sketches AI is clearly a great tool.
Last edited by Qooper; 10th Dec 2022 at 21:52.
Of course. My reply was speaking to Aja's prior post where he was asking himself why he creates art, and he had some compelling reasons; what I mentioned was my perspective, not a fits-everyone sort of declaration.
We were speaking about the legitimacy of AIs as their own autonomous art-creating entities, but all right. If you want to talk about conceptualisation and skill in execution, I don't think someone who's just prompting an AI is going to get as much mileage out of it as someone who actually understands the process of how art is created. A good painter, for instance, is someone who's undergone a journey, starting from zero skill to someone who can turn out something thought provoking by understanding the possibilities and limitations underlying the medium and its tools, to someone who has figured out how to move beyond most of those limitations. You can't replace that journey with an AI shortcut and not lose the benefit of experience and understanding along the way. So, if you want to use AI to help you consistently create something (conventionally and artistically) good and not just play dice with it, you'll need to understand where it slots in, how it slots in to the creative process instead of just getting it to churn out copycat Dalis. If one can guide an AI's hand, so to speak, then the 'prompt writer' becomes more of a coach that massages meaning, texture, and technique into being through AI images without actually holding a brush. It's still a shortcut, so the objective/subjective quality of it is always going to be up for debate.There's a degree to which the quality of art can be called an objective thing, if by art we mean the kind that takes skill and hard work to master. Ideas alone are cheap. It's the execution of a great idea that matters. Most contemporary art is just an idea slapped together by someone who doesn't respect this reality. From this perspective, I still think there are ways AI can be used as a tool for great works of art, but it'll be more than giving a prompt and calling it done. If Michelangelo had to learn how to use a brush and how light works to give form to his ideas, to what great lengths will artists today go in order to paint with AI?
As you've seen from dema's posts, the quality of what you get is determined by how well you understand how to write the prompts, which means you're going to have spend at least a little effort learning how to eke out what you want from it. Is that enough control? Up for debate, but I'd assume the painterly consensus is, 'not really'.EDIT: On further thought, what I think is important in creating good art is control, and using an AI you lose that. But for creating reference images and sketches AI is clearly a great tool.
Last edited by Sulphur; 11th Dec 2022 at 00:06.
Famous last words. If you're following the Boston Dynamics stuff, I think AI is to that level now. It's just a matter of a team actually doing the work.
I think the fairer thing to say is AI won't be playing with the same intention, where intention includes all kinds of aesthetic opinions, values, and life experience that an AI won't have short of actually living a human life and grounding those things in lived experience. So the touch may be there, but the intention won't be.
Quick cite check, this is Donald Davidson's classic "Three Varieties of Knowledge" argument why a spontaneously generated human-like being won't have experience like we have it. He called it Swampman, but I think he's really talking about the kind of AI we're seeing emerging these days.
Sure, I just wanted to weigh in and give my perspective.
I know, and I found the conversation intriguing, which is why I wanted to offer another angle. From the perspective I offered, the artist is the person giving the AI a prompt, since the person has the desire to create art as you mentioned in your previous post. The way you described the prompt writer being kind of like a coach is how it looks like from this perspective. I think it's interesting to look at things through different lenses to see what gets magnified and what disappears.We were speaking about the legitimacy of AIs as their own autonomous art-creating entities, but all right.
I think the journey is essential, which is why the idea of AI as an autonomous art-creating entity doesn't quite sit with me. Sure it has been trained, but to me that's not a journey. Although I can't say yet what would be. I guess I'll have to think about this more and get back to you.A good painter, for instance, is someone who's undergone a journey, starting from zero skill to someone who can turn out something thought provoking by understanding the possibilities and limitations underlying the medium and its tools, to someone who has figured out how to move beyond most of those limitations. You can't replace that journey with an AI shortcut and not lose the benefit of experience and understanding along the way.
In a way I see it as similar to doing a Jack Pollock -style painting. You don't have fine control over where each drop of paint lands, but you kind of get what you want and you can easily do random variations.As you've seen from dema's posts, the quality of what you get is determined by how well you understand how to write the prompts, which means you're going to have spend at least a little effort learning how to eke out what you want from it. Is that enough control? Up for debate, but I'd assume the painterly consensus is, 'not really'.
Actually, we're not. None of the examples posted in this thread were created by an AI. They were all created by a human, using a sophisticated search & merge tool that's based on machine learning. It generates a work of digital media by adapting and merging things from its database, given a starting point and constraints supplied by the user. Using AI art programs greatly reduces the skill needed to make art. But the AI doesn't supply the intent, and it doesn't pass judgment on the result. Examples don't show up in this thread unless somebody created them and thought them worthy of sharing with other people.
The story behind the work isn't something the AI makes up to explain what's in the image. The story is of somebody playing around with the inputs to the AI. What was the original idea, how did it evolve through trial and error, how did the AI respond to different approaches, etc? Basically, how did you get what you got out of the AI? Because that's the artistic method.
heywood, part of this discussion is speculative, and isn't limited to just whatever's happening right now on the internet. You're a bit late and on a different tangent, and I already addressed some of these points earlier. Either way, we're headed towards it given enough time whether anyone likes it or not.
I think we're basically in agreement, but I'm enjoying this discussion, so I'll continue (suck it, SubOrg).
Hey, I can live with that. I for one welcome our new machine overlords. I find their art so far to be unmoving. I guess it's sort of a cop-out to say that I can only speak for myself, but other peoples' creative processes are significant to me in how I connect with their art. They certainly won't be for every, so even current AI art might be really affecting for some, and I don't mean to deny those experiences.But you're going to have to live with idea that the machine-created art will live in the same space as the real thing and mess up the boundaries, because that's where we're headed, for good and ill.
I'd contend it's not perfunctory at all but is rather the most crucial part of this debate since what we're really talking about is the nature of art and what makes it meaningful to us.The thing is, this isn't necessarily exclusive to human beings - as in, while there's a perfunctory philosophical issue at play with intentionality and sentience...
I don't think that's what I'm doing. I'm setting the goal posts in cement, and if the AI scores on them, then so be it. As far as I'm concerned, Commander Data's paintings have an aura, as Walter Benjamin would say (I'm halfway through the essay, dema, I swear), and if AI develops intentionality, it will be a weird and wonderful thing, and we'll have to acknowledge it....a network that does not give the same output every time for the same input has the spark of something new, something closer to the way our brains work, and that means we're heading into greyer areas where the definitions we hold become murkier. It's a mistake to dismiss this out of hand, because as Cipheron mentioned (good posts, by the way), we can always redefine the goalposts to exclude machine-created art as 'not art' even if they start displaying things like a thought process.
Agreed. The line is only going to get fuzzier and fuzzier, maybe until we've passed it without even realizing it. And the closer it gets, the closer we get to the ethical issues of the treatment and rights of non-human sentient beings (thanks again, Star Trek), and I don't want to be on the wrong side of history there. Also, I don't want to hurt the AI's feelings.To some degree, the AI is doing what we do, which is to process information and manipulate it in a way that's desired. The current difference is the machine itself does not create the desire, we do. And as far as I can tell, that is not impossible to change given enough time, or at least to simulate the apparent existence of one.
I tried out ChatGPT to make some AI generated (f)art (jokes). They weren't that funny. Then I told the AI to tell me a short story that begins with a joke and ends with a tragedy, which quickly ends with one last unexpected joke. Then the good stuff started rollin' in:
What did the zero say to the eight?
Nice belt!
But as the eight walked away, it suddenly collapsed and died.
As the other numbers gathered around, they shook their heads and sighed. "I guess you could say it was a number one killer," one of them joked.
There was also this:
Why did the bicycle fall over?
Because it was two-tired.
But as the bicycle lay on the ground, it was hit by a car and completely destroyed.
As the car drove away, the driver shook his head and muttered, "I guess you could say it was a real bike-tastrophe."
Well, I guess they're not that good, but what made me laugh was how surprisingly well this form works. Joke tragedy joke BADUM TSHHH!
The world's first AI generated girl band. Looks very convincing. The voices are actual singers apparently
Well it certainly sounds as crappy as the rest of pop.
Of course, there's no substitute for lived experience, and AIs cannot compete with that right now.
I think it's not impossible to simulate a journey for an ANN, though, and at a temporal resolution that facilitates this happening far quicker than for a human being. But that's going into sci-fi for now, because we need to be able to give it stuff like sensory information and more sophisticated logic processing and stuff like proprioception, as dema would point out -- even if I surmise it's not that far off.
That's a good way of looking at it. The process itself is as important as the end result when it comes to a Pollock, so that's one way to use it.In a way I see it as similar to doing a Jack Pollock -style painting. You don't have fine control over where each drop of paint lands, but you kind of get what you want and you can easily do random variations.