AI has certainly brought human creative agency into sharper critical focus. An excellent argument suggests we look closely at the human unconscious whenever we discuss AI creations (See: Babel, published in n+1). Though AI-developed artworks can defy critical scrutiny (aka useful criticism in the sense of lit/art crit, not whining or complaining), they largely seem to be based on the manipulation of the syntactical. Machines do not seem to bother with the semantics of the works they scrape (learn) for study or the works they create as a result of this scraping. They may even have ethos, logos, and pathos, but these aspects are generated by the inevitable semantic value of syntax. The human unconscious may operate similarly, but with a more attuned sense of the sematic. In other words, AI-developed artwork is largely combinatorial (manipulation of symbols), which ends up having meaning. Whereas the human unconscious reckons with meaning, and not just the syntactical.
What’s your definition for creativity?
For me, it means the ability of the human mind to spontaneously create based on random inspiration through memory, emotion, and life experience. The ‘creative spark’ or ‘Eureka’ moment. Something often fleeting or nebulous. The unconscious mind. A machine can never replicate that unless certain theories or sci-fi storylines become reality(as with the fascinating exploration of consciousness in the original GITS)
Note: I’m purely talking about artistic creativity here, not the creativity that drives engineers or mathematicians, or theoretical physicists - as that’s grounded in logic and mathematical rules, and can be replicated somewhat by a machine/algorithm.
I initially disagreed with your previous post:
However, you are correct by definition:
To me, a piece of art is creative or has a creative component to it, if it is perceived as such by people. To me, how it was created or who/what created it is not relevant.
Because according to your definition, only humans can be creative, any discussion in that direction is pointless.
That is the dictionary definition of creativity - “using imagination to create something”. Machines don’t have imagination, only humans do. That’s not a perspective based position, it’s a definitional one. Imagination is the ability to construct something unique devoid of external data- AI creates something “unique” reliant on external data. Or, as Wikipedia says:
Imagination is the production or simulation of novel objects, sensations, and ideas in the mind without any immediate input of the senses.
Nobody knows how imagination and/or creativity ‘work’ in a human being, or even if there is some process that could be discovered. So how can we say that it can’t be reproduced in a non-human, even non-organic entity?
There are many definitions for creativity. Quite a few of those rely on a human factor, which kind of makes sense because that was pretty much the only way so far to create something new and unique, especially from an artistic point of view.
Just to give a different example:
creativity
the ability to produce or use original and unusual ideas:
[…]
and
creativity
the ability to produce original and unusual ideas, or to make something new or imaginative
Or when you ask misses Wiki directly about creativity, it starts off like this:
Creativity is a phenomenon whereby something new and valuable is formed.
To me, those definitions allow non-human creativity, which makes more sense in my personal opinion. Whether we reached a point where machines/neural networks or whatever can be creative according to those definitions is another question.
We’re talking about art here, though, and a machine being ‘creative’ in an artistic sense is really just a meaningless parody. A machine can not view a subject under different lighting conditions and ‘feel’ a certain way about them. A machine cannot wake from a dream and write down a poem or a piece of piano music based on inspiration from that dream.
I think we’re talking about different things here, that’s why I stated above that I was specifically focusing on art, rather than more logic-based pursuits.
Only humans can currently create(as I also stated above) I have to disagree here, I think that it’s a pertinent discussion to have during this era of machine learning infancy, and also something fascinating to ponder.
I am not talking about that.
There is no way a discussion can take place if your definition does not allow it.
You’re not talking about art, or you’re not talking about logic-based creation?
I didn’t say I didn’t ‘allow’ it. You asked my opinion and I gave it to you. Why would you think I would shut down discussion when I’ve enthusiastically invited the discussion?
I’m not trying to be obtuse. I honestly am interested in the possibilities - hence my shoutout to GITS.
I’ll happily use any tool that I find on the shelf – including AI. Anything that enables me to do more work in less time and with less drudgery. Because, there is a lot of “tedious drudgery” in doing 3D work right now. Please give me another “power tool.” I’m not afraid of it.
Your question have hidden aspect and unless you name it, the discussion will go nowhere.
Namely: do machines have consciousness?
There is philosophical and theological consensus that this is not the case, and discussing aspects of human nature (like creativity) in machines is pretty much pointless.
Before you attribute any human trait to a machine, first you need to untangle and sort things like consciousness and unconsciousness that are de facto the driving force behind most of the human activities - partially even including aspects of biological needs.
Right now DALL-E and similar systems are nothing more than sophisticated automata. We tend to view them as something original because they are essentially black-boxes. But underneath there is only pure logic.
Human dreams, and most importantly human psyche are not logical. This condition very often flies over the heads of the people who are too technical and have underdeveloped spiritual side of their personality.
This is irrelevant to the over-arching theme of ‘machine creativity’. Automating UV packing or retopology is a technical task to increase productivity, no different than a sharper or more extensive set of chisels to sculpt stone. I agree 100% that if learning machine algorithms can help us in the more mundane tasks of 3D art, then great, but it has nothing to do with the creativity aspects of 3D art.
Yes, this.
In these threads in general people most adamant about creativity and imagination being human-only qualities always seems so certain that machines can’t do it, and are therefore no threat, but I’m not at all certain because I don’t see those terms defined in a way that excludes non-human ability per se. Unless, of course, one inserts “human” into the definition, but sorry, I don’t find tautological arguments convincing or even useful. Humans much smarter and more knowledgeable than me can’t agree in any detail on what art, creativity, and consciousness are, so how could I be so arrogant to be certain?
I see machines do things now that we could only dream of 40+ years ago when I was really into AI for a short time, and we were excited about rules-based expert systems (ha!). Deep learning and neural networks have changed the playing field profoundly. So no, this isn’t the same old same old. Back then any outcries about AI replacing humans were IMO uninformed fear-mongering. I no longer feel that way. (Well, there’s still a lot of uninformed fear-mongering. but there are quite well-informed fears now too, and I wouldn’t throw them out with the bathwater.)
And of course this isn’t a one-size-fits-all scenario either. I doubt that even a much more evolved AI than DALL-E will replace all human art. But regardless of whether a machine is ever declared conscious or creative, it will be able to take over a lot of functions, because if we’re honest, many are utterly repetitive and mundane. What sort of art hangs in your average hotel room?
Handcrafting objects has survived the Industrial Revolution, and more recent mass production. Photography didn’t vanquish painting and illustration. Painters still paint even though we now have prints of Old Masters made by machines. Most people never ever acquire any original art, they have cheap copies. Some will continue to appreciate the personal touch of a human artist (and I’m not so sure that originality and creativity are always deciding factors there either). And while humans continue to try and one-up each other, some will always want something “special”, unique, that nobody else has, just for status. So there will be human artists even if AI were to take over a lot of the domain. But whenever there is a technical revolution, a lot of jobs disappear. And I don’t think that this time there will be a lot of new jobs that artistic people could slide into. For a while, sure, with AI as a helpmate. In the long term? I’m thinking “prompt crafting” might become a thing, but will most artists be excited by that?
I don’t think we’re there yet in general, but we’re awfully close. And of course in some areas it’s already happening. So I feel we owe it to the next generations to not get stuck in mere denials. These discussions aren’t worthless even if we can’t define the terms all that well, because hopefully they make us think a bit more and learn something new. But maybe I’m just spiritually underdeveloped, * snort *.
I am not talking about logic-based creations.
To clarify my view. I am not trying to claim that machine learning systems today are creative. However, I believe that we are going to get to a point where a machine is going to be able to create something (even if it is just a digital images in the sense of DALL-E) which people would rate creative, if they were not told whether it was created by a human or a machine.
According to you:
When I talk to orchestral composers, more often than not, they are telling me, you can’t wait for ‘Eureka’ moments. Even though they apparently sometimes have those. They usually seem to just start with something and go with the flow, revise things, … . And according to them, they couldn’t generally tell whether they preferred the pieces where they had a ‘Eureka’ moment.
Even though that’s just anecdotal, it still shows me that the creative process of artists can be quite broad and not requiring a ‘Eureka’ moment.
I had this conversation with them because I was very interested in how they came up with their ideas which were and still are very creative to me.
I want to make clear that this is not meant disrespectful. I just want to make my position very clear without having to write a whole essay.
This is not relevant to me.
I don’t care about that.
From my point of view, I am not doing that. The criterion whether something is creative depends on the person who perceives it in a certain way (watching, listening, …). That’s why (with that criterion in mind), it doesn’t matter whether something is created by a human or a machine. And more important, I am not attributing any human traits to a machine.
Here’s something to consider: anything philosophical aside, no AI will ever replace me as an artist, because it doesn’t have my ideas, my style, my methods, my fictional worlds and characters. Can AI be creative? Possibly, but in terms of replacing artists, it’s irrelevant. Van Gogh didn’t replace Rembrandt. Kabalevsky and Rachmaninov both are revered composers. Can AI make art? Maybe. Can AI make my art? No, and neither can any human living, dead, or yet unborn. As an artist, I choose not to be replaced by AI 
If I was a professional artist, I’d be more worried about humans replacing me, since there’s always someone cheaper who can do the same thing you can. Since I make art for my own passion and enjoyment, however, I am completely unbothered by any other artist- human, machine, elephant, dolphin, or chimpanzee
If you are not interested in said aspects, then you cannot discuss art in general.
Labels like ‘creativity’ (whatever it means) are malleable and therefore are not the best tool to express complex emotions and state of being in general. If you are only interested in attaching whatever label to whatever the newest state-of-the-art machine can do - you can already do it.
This however will not change how the creative process works. This will change only your perception and semantics which you are using to describe surrounding world.
Well concerning AI and art I agree with what @Musashidan said about it. It is compareable and also related to these many discussions about crafting vs art. If its about art we directly and indirectly imply a mind, consciousness, perhaps even a soul.
And usage of the term AI in science is really not identical or congruent with its usage in science fiction literature. Science fiction can simply break any existing barrier and expand a subjects scope into something far more sophisticated.
Thats why there is a flux compensator or star treks heisenberg compensator. Perhaps just talking about learning patterns would imply less and lead to less abstruse assumptions, but Danny is simply right if he states that we can stop thinking about it until there’s a “ghost in the shell” that breaks any existing barrier for AIs.
I think its already problematic to presume imputability. Is an AI an autonomous entity or rather a tool? One could argue an AI is just a programmers tool - as what if an artist puts paint into a ballon and makes it explode above a piece of paper. Is that the artists or the balloons product. And opposing to that, what do we think if we teach a child to paint and it goes on with that knowledge and transforms it into something new.
Is it art if a bottle with color is hanging from a cliff and the wind smashs the bottle against the cliff? And whose art is it, the winds, the bottles or do we judge the guy on top to be the artist? Does it have any deeper meaning if the color drying on the cliff forms a recognizable shape?
Sure the outputs/decisions of AIs can be pretty amazing, judged in their context. But we are far from having AIs that have consciousness, a mind and because of that real artistic creativity.
So yes the artistic process needs ingredients that still keep lifeless entities from being an artist, even if the result looks pretty artistic.
That’s true. I don’t think it is necessary to consider what philosophers have to say about art to be honest.
The emotions or whatever an artists wants to express does not always align with what people are seeing in their art. An artist who doesn’t have empathy or lacks certain emotions may still create art trying to mimic that emotion. To me, this can still be art, even though the artist lacks those emotions. Art can be about subjects the artist hasn’t experienced.
I haven’t seen any machine learning project yet where I would say that the results have an artistic value for me. However, I don’t see a reason why this couldn’t change in the future.
I personally don’t understand why art would need to be created by a conscious being. In other places I have read, art could only have a soul if it was created by humans.
It would be interesting to get a better understanding of how often people perceive something as very valuable art with lots of interpretations around it, but in fact the artist just needed one more piece and just had to get it done. Would this not anymore be art?