LLM Coding and the Phenomenological Construction of Art
Phiosophical idea, when is something software?
I would make the argument that a good definition for software is "A sequence of instructions in order to accomplish some given task"
Say for example that we had a random word generator that contained all words to do with food and cooking. Similar to a million monkeys on a million typewriters, we could eventually end up with a "good" recipie
But - My argument would be that the recipie does not exist, it only exists as a collection of words until it is perceived at which point it becomes a recipie
This is because we are actually doing the work by evaluating and constructing those concepts through the work of perceiving them. It is not that we have discovered hidden gems, it is that we have created gems using those signals in combination with our expertiese and understanding.
Institutional Theory of Art (Arthur Danto / George Dickie) Something becomes art because it is perceived, accepted, or treated as art within a cultural or conceptual framework. The viewer’s recognition plays a key role in “creating” the artwork. Closely related (and often a better fit for your phrasing): Reader-Response Theory (applied beyond literature) Meaning and artistic value are produced by the observer, not embedded solely in the object. Phenomenology of Art Art exists in the experience of the perceiver. Without perception, the “art” does not fully exist as art.
The core idea Danto is arguing that seeing something as art is not the same as simply seeing it with your eyes. Two people can look at the same physical object and experience something entirely different: one sees an ordinary object, the other sees a work of art. What makes the difference is context, not appearance. “Something the eye cannot descry” This means: No visual feature alone makes something art. You cannot identify art purely by shape, color, skill, or beauty. Example: A supermarket Brillo box and Andy Warhol’s Brillo Box sculpture look the same. Visually, there is no difference. Yet one is art, and the other is not. The eye alone cannot explain why. “An atmosphere of artistic theory” This refers to the ideas, concepts, and debates surrounding art at a given time: What counts as art now? What questions artists are exploring? What conventions are being challenged? In Warhol’s case: The theory of art had reached a point where questioning mass production, authorship, and consumerism was meaningful. Without those theories, the object would remain a box. “A knowledge of the history of art” Art is understood relationally, not in isolation. A work often responds to earlier movements, styles, or assumptions. Understanding it requires knowing what came before. For example: Duchamp’s Fountain (a urinal) only makes sense against centuries of handcrafted, “beautiful” art. Without that history, it’s just plumbing.
The core idea you’re expressing Art is not fully contained in the object or the institution, but is actively constructed by the perceiver’s expectations, intentions, and mode of attention. In other words: The object affords art The viewer enacts it Art happens in the act of seeing, guided by what the viewer brings.
Phenomenological Aesthetics (strongest fit) Especially in Merleau-Ponty and Roman Ingarden. Art exists as a phenomenon in experience Meaning is not fixed The viewer “completes” the work through perception Ingarden explicitly argued that artworks are schematic: They contain gaps The observer fills them in This applies even to mundane objects.
I think that ultimately this strongly relates to the current state of play in the LLM programming space. Similar to the pre-llm space we initially used it to operate at higher levels of abstraction, so instead of thinking "how do i move these bytes into this buffer?" we moved up to "how do I parse this file?" and then further again up to thinking in systems instead.
I think that as it stands we're still in the space where we are creating software, which is why the guidance and supervision is so critical. The LLM isn't writing the software, you are by guiding it to your target destination.
I think the problem comes when we delegate the thinking away to a probabilistic black box.
We can see this in the fact that for a long period of time, there has been a misunderstanding in terms of what a software engineer does. The real job is the thinking not the typing. You'll see this every time that someone overenthusiastically suggests a rewrite of your entire companies systems, the reason this is so painful is that within that running system are all of the decisions and thoughts and specifications that created it - even if the code is a bit ugly
This delegation of "the work" to a probability box is why the results kinda suck
So in summary, the tools might change but the job isn't going anywhere, because software is just organized thoughts.