This is an essay about how I see art, what I think computational creativity is and some things I would like to express about assisted creativity. Before I begin, here is a disclaimer. I am not an artist, neither I am a computer scientist or an Artificial intelligence practitioner. I am just a software engineer, who keeps hearing about these things from various people around me and have thus formed an opinion on these things. I don’t have any authority to influence what the definitions of these things actually are, and also don’t expect most people to agree on my views. All I want from this essay is to express my views and hope to stir some conversations among some groups of people, and I would be lucky to get some criticisms from smarter people.
Having said that, I must begin with defining what I mean by a work of art. And also some thoughts about things that can and cannot be considered art based on my definition.
What Art is
In my definition, a work of art is a essentially a way to express some emotion or feeling a person has. This emotion or feeling is usually a result of a large number of influences, which would include his/her past life experiences — both through personal encounters and through learning from other people’s encounters served in some distilled form. It would also include a large array of subjective thoughts and ideas acquired by means which may or may not be traceable to a known event or idea. Dreams are one example of such influences as well.
Does it have to be something beautiful? Preferably, but not necessarily. Besides, the definition of beauty itself changes from time to time and from context to context, so I’d prefer not to rely on that as a measure of a successful art work.
What Art is not
In my definition, art is not a problem to be solved. It is not a combination of things that serve some utility. A piece of art might be solving some problem given some circumstances, but it doesn’t have to.
For example, if I come up with a way to cure a known disease, I can justify that it is a useful thing, but I can’t necessarily claim that as a piece of art. The reason is that I would be creating that solution to solve a problem, not to express my feelings. The words “state of the art” thus continue to confuse me when they are used to describe an excellent utilitarian object.
Art also is not a selected combination from a sea of valid combinations. At least that’s not how I’d define it — although mathematically speaking we can express everything as one of many combinations of things that constitute a larger object. But the point is — mathematics does not provide us tools to select that one particular combination which distinguishes itself as a work of art while rejecting others — at least not without help from human subjectivity.
Properties of Art
Perhaps more important than the definition of art is what properties a work of art has. Or what value does it provide, if not a utilitarian one? Here’s an example of a piece of art.
It tells us quite a bit about the life and times of the painter (Leonardo). It tells us about the society in which this work was produced, what values were considered virtues of the society and to some extent the general level of importance people would give to religious guidelines. It also tells us quite a bit about the color palette and its relation to the things which the painter would have wanted to express. Or is it just one of the combinations of random colors, and all the above things are what my mind created while I saw this painting? There is certainly a high amount of subjectivity involved both in the mind of the painter and the observer. Perhaps, the people of 23rd century will interpret this painting as completely different than most people today would. But I think that is the vital point — subjectivity. In absence of subjectivity, it truly is just a combination of colors and canvas behind it.
This raises an important point towards my above definitions of art. It essentially puts subjectivity in the center, and almost promotes it as the only thing that differentiates a piece of art from other things. One important thing to note is that both parties — the painter and the observer must involve subjectivity in the process of producing art. A piece of art is produced infinite number of times, once when it is produced by the creater, and then once everytime when an observer looks at it.
For moving to the next section, we need to have some relation between art and creativity. For the purpose of this essay, I’d like to put creativity in the driver’s seat by saying that any piece of art is only possible when creativity is used in making progress towards producing it. In absence of creativity, there is no subjectivity. And in absence of subjectivity, there is no art. How is creativity necessary for subjectivity? Well, a subjective object doesn’t really exist — its essentially created from thin air by a creative being based on other objective observations or yet other subjective experiences. For example, the emotion called “jealousy” doesn’t really exist as a physical object in our universe. It is a subjective experience humans develop based on some objective fact like a colleague getting promoted, and mixing that with some personal experiences or memories or emotions. We’re essentially creating subjective experiences that we then use to experience art. So creativity must be in drivers seat, or somewhere close.
Computational Creativity and Art
Building on the set of opinions and definitions thus far would completely throw the idea of computational creativity out of the window. At least if we restrict ourselves to the idea of machines creating art works. It seems absurd to consider anything which a machine produces to be a work of art if we know that we haven’t yet developed machines that will first produce subjective experiences.
But why is there a hype about computational creativity to produce art? I’d say the idea of computational creativity works its way by having a different definition of creativity and art. In my understanding, the idea we fancy is somewhere on the lines of producing a large array of valid combinations of things based on varying levels of following rules or guidelines, and finding some way to reject some combinations and promote others until the remaining set is small enough to confuse most people whether it was produced by a human or a machine. Note that we don’t really bring subjectivity into the equation until it goes out of the gate to human observers. Its not quite convincing to me, but that’s just my opinion, so I’ll play along for a while.
(I would again like to apologize if I somehow misrepresented the concept of computational creativity for producing art work. But based on my learning so far, this seems to be a good simplification to work with. I seek corrections if you are an expert in this field and would like to provide me more insights)
There is always a middle ground, they say. How about creating algorithms which don’t claim to be creative themselves, but are only generating a lot of interesting combinations based on either statistical models built on existing art works, or some heuristics which are built based on our knowledge of how art pieces can be generated, or some combination of both, perhaps putting some randomness into the mix? Then, we can use these algorithms to support artists while they are working on creating pieces of art so they don’t have to work hard collecting past experiences and rejecting inferior ideas.
Well, it barely works in theory (if at all), and I wonder where we are going with this anyways. The basic assumption of this field of work is to produce workable ideas which are supposed to assist the process of creativity while producing art works. But how much do we know about the process itself? Is this supposed to provide continuous flow of ideas to an artists so that he/she is not “stuck” during the process? Is it in some way supposed to improve the productivity of an artist? Or is the claim bolder — saying that its not productivity but rather the quality of art which they seek to improve? Or is it even more bolder, which is to say its both?
Is an artist ever stuck?
In my observations, an artist reports to have experienced varying productivity levels during the process of creating a piece of art. For instance, there is a term called “writer’s block” which represents the state of mind a writer has when there are just no interesting ideas coming to mind, at least not interesting from the writer’s own personal perspective. Does this mean, it is harmless to eliminate these seemingly unproductive periods and enable the artist to have a continuous uninterrupted period of time when there will always be a lot of ideas available to be used? Or is this emptiness somehow necessary to stimulate more refined ideas, and to process already existing “good” ideas to really find out they are “not that good really”?
What would a writer do when he/she experiences a writer’s block. I suppose he/she would go out and meet people, do some chores, maybe read other people’s work or just struggle and accumulate an emotional state of disappointment. In fact, none of these activities look like useless to me. I wouldn’t expect a writer to write a meaningful story if everything in his/her personal life went as planned and he/she never had to interact with other people or do things which were frustrating or seemed wasteful. I truly believe all these experience shape the writer’s personality and in one way or another would contribute towards building a subjective experience, which is essential to create art. They may appear to be unproductive periods of time to an engineer, or a manager, but we’re not talking about digging holes in the ground, are we?
Does the idea of dips in productivity present in all fields of art so as to make sure we’re not just talking about writer’s here? I suppose it is. Based on my experience, I can’t imagine someone producing pieces of art flawlessly without getting “stuck” multiple times during the process.
The illusion of productivity
What assisted creativity essentially promises to provide is a never ending stream of good ideas. It doesn’t claim any of these ideas to be real pieces of art, but they do say these things are pretty close and only needs guidance from an artists to quickly convert into real works of art. I am imagining an algorithm which never shuts up while I am “stuck” and reflecting on my experiences while working on a piece of art. It would lure me into considering one of its suggested ideas to continue with my work rather than wasting time exploring my own original ideas. Here I am with hundreds of potential ideas which are screaming for attention, none of which are based on my personal subjectivity. How long can I resist this “goldmine” of ideas, given I am already starting to feel the need to move ahead with my work. Its so convenient that it looks like impossible to resist. But have I considered what I have to sacrifice to accept this way of working? I will be sacrificing on the amount of time I would have otherwise given to reflect more on my original ideas, which could be none at the time, but they could come if I was willing to work my way through the frustration. Instead, I am tempted to pick one of the suggested ideas. I have a completely different problem now — to narrow down on these generated ideas, which are far more than what I can handle.
Using algorithms to generate ideas does not increase the capacity of human brain to process ideas. The capacity stays the same, only now the work the brain needs to do is never ending before it can get back to being “stuck”. This looks like productivity, but I don’t think many of us would agree that it really is.
So, should we abandon the project?
No, we shouldn’t. I might have given an impression that I see no hope in this direction, but this is far from what I think. Let me explain.
To me computational creativity and many other AI fields are like space exploration. We need to do it because we don’t know what’s out there. By exploring the unknown, we give ourselves greater chances of encountering things which we never can imagine on our own. It widens the horizons, it prepares us to be humble and look at the world differently. In that way, simply exploring computational creativity on its own can enable us to create new art, which is simply impossible until we experience it. To me, experiencing what was earlier unknown is a far more important contribution to art than building any tool.
And its not just about art. Although this essay is about art, I really hope computational creativity is by no means limited to just producing art. It could be used, for example in places where humans can’t reach, like outer space or deep oceans. It could be used in completely new scenarios like bringing peace between nations when human minds give up on the amount of data to process. It could be used in unimaginable scenarios, which we would only imagine once we widen our horizons. Even if it was only used for generating “imitation” art, there could still be places to use it, for example in therapy or medicine or economics. Who knows? We need to give it a chance so we’ll know more.
I am only concerned about us creating stories around things that don’t have any. Our current economic and political structure forces us to think of reasons to justify pursuing research in any field. So, we might sometimes come up with stories to apply computational creativity, just convincing enough to justify an immediate financial gain or some military advantage or something similar. Otherwise, we won’t get funds to sustain the research. And I think we’re very good at creating such stories — we’re good storytellers. My concern is we getting fooled by our own stories and going too far without realizing how we started and where we ought to go. As owners of a mature mind, we need to invest in the idea of not trying too hard to create meaning from things that have none. And from what I see so far, there is none.