Some thoughts sparked by interacting a bit with chatgpt and stable image diffusion:
One thought is that the generative process undertaken by both of these "ai" (I think they probably fall significantly short of intelligence but I might be wrong) is not too dissimilar from creativity. My experience of creativity anyway usually follows this sort of pattern- load up my mind with things I like aesthetically and thematically, which are drawn with a small amount of discrimination from a variety of sources, then remix them, shuffle them around, take the component parts and extrude them into something I can call "my own". The goal of this activity tends to be to create something which is more than the sum of its parts (on a good day). It also can produce a sort of vacuous performance of egoism (on a bad day). Its just a process and it is neutral as to the quality of the production.
This process of remixing seems to be literally not far off from what a "large language model" does. More-or-less taking the processing power we now have and coupling with the ability to recombine something until it achieves the appearance of meaningfulness. That being said, it is better seen as a tool than an agent. The reason, I think, that it can be confused as such is that, as a tool, it ventures an extra step into the operators traditional aegis by gathering the fodder automatically that will provide the components for reassemblage. So, that part of the artist's pride in their "good taste" or prudent choice of which inter-generational conversation to join can be partially off-loaded to the tool.
One might level a criticism that it feels like cheating to be able to steer the engine with a prompt like "In the style of Ernest Hemmingway" or "Caravaggio" or "archival footage of traditional Indonesian percussion" without ever having consumed the body of work in question. To this criticism I might be tempted to respond with a "Yes and No". The truth is that the producer is cheating, in the sense that they are cheating themselves of the enjoyable part of the process which is consuming the inspiration for digestion and subsequent regurgitation. That being said, I don't think there is "cheating" in the game of creativity. Contrariwise, cheating is often a very fecund activity when building a minotaur out of a bike-seat or some other sort of amelioratory cargo-cult. This leads me into another observation...
A distinct line-of-thought that these tools have sparked in me is one of Form and Content. At first the notion came to me that the ais seemed to be able to reproduce a form, say the structure of a 3 act play or a pop song, but how could they posses the faculties to populate that format with rich content? But the whole thing becomes more troubled the more one mulls it over, for what exactly is content? Is it the moral of the story, and is it maybe too delicate for a robot to handle with aplomb? Yet, we all know that "the medium is the message".
I was thinking of Eco's The Name of The Rose, its form is that of a detective mystery drawing heavily on the Holmes/Watson archetypes but its content is a thick and nutrient dense sludge of semiotics and scholastic philosophy. The overall effect of the novel is a rewarding sense of having "slummed it" in the trenches of our dubious cultural heritage, due in no small part to the interplay of the very "genre" elements with the many passages of complex highly abstract thought. All that to say, I'm torn between the idea of content as the very human touch of genius or inspiration on a essentially lifeless golem of form, and the notion that content is an emergent property of either form itself or the audience's bestowal thereof. Maybe sometimes it is one and sometimes the other, I don't need to be dogmatic about it.
Ephemerality: I'm not sure how to segue into it so I'll just start... The infinite monkeys tapping out their infinite typewriters will necessarily produce huge mountains of wasted paper. There is a point where the limited resource of human ability to digest and regurgitate cultural content becomes overwhelmed by the seemingly unending ability of the bots to emulate this process with great efficiency. By removing themself from the deep and slow research a human artist may surf the wave of brutal processing power, this may be thrilling. You can "compose" 10,000 words a day of legible prose or make 300 images of Rembrandt-level etchings (I hasten to underscore here that this glut of production is not the same thing as a physical etching or leatherbound book, quite the contrary it is only a medium resolution jpg or pdf or whatever), in a society that fetishizes productivity to the degree that we do this is indeed a thrill. I suppose my point here is to acknowledge the fleeting nature of a discreet piece of production at this stage. The production of high volumes must be seen as severely devaluing. The valuable skill may be that of sifting out the dross rather than cultivating anything new.
Ok, so I got Chat Gpt to edit the post and this is the new version:
Exploring Creativity and AI: Reflections on ChatGPT and Stable Image Diffusion
As I've interacted with ChatGPT and stable image diffusion, I've noticed intriguing similarities between their generative processes and the creative process. While I hesitate to label these AI systems as truly intelligent, their ability to remix and recombine elements resonates with how human creativity often unfolds. Personally, my own creative journey follows a similar pattern: gathering aesthetic and thematic inspirations from various sources, remixing them, and transforming them into something unique. The goal is to create something greater than the sum of its parts, although sometimes it can devolve into self-indulgence. Ultimately, this process is neutral, neither guaranteeing quality nor lacking potential.
Remixing and reassembly, key components of these AI systems, closely resemble the activities of an artist. However, it's important to view them as tools rather than autonomous agents. One reason they can be mistaken for more is that they automate the gathering of source material for reassembly. Consequently, the artist's pride in their discerning taste or ability to engage in inter-generational conversations is partially offloaded onto the tool.
Critics may argue that it feels like cheating to steer the AI engine with prompts like "In the style of Ernest Hemingway" or "Caravaggio" without having consumed their respective bodies of work. My response to this criticism would be both "yes" and "no." Admittedly, the producer misses out on the enriching part of the creative process, which involves consuming and digesting inspirational material before giving it a fresh spin. However, I don't believe there is inherent cheating in the game of creativity. In fact, cheating can often be a catalyst for novel ideas, like building a minotaur out of a bike seat or experimenting with amelioratory cargo-cult concepts. Thus, the act of cheating can both hinder and inspire creativity.
Another fascinating aspect to consider is the distinction between form and content. At first, it seemed AI could replicate the form of creative works, such as the structure of a three-act play or a pop song. However, questions arose about its ability to infuse rich content into these formats. What exactly constitutes content? Is it the moral of the story, something too delicate for a robot to handle adeptly? Yet, we all acknowledge that "the medium is the message."
Think of Umberto Eco's "The Name of the Rose." It follows the form of a detective mystery, drawing heavily on Holmes/Watson archetypes, but its content delves into semiotics and scholastic philosophy, resulting in a dense and intellectually nourishing experience. The interplay between genre elements and passages of complex, abstract thought enhances the overall impact of the novel. This leaves me torn between the idea of content as a human touch of genius or inspiration imposed upon a lifeless form, and the notion that content emerges as an inherent property of the form or through the audience's interpretation. Perhaps it varies, and there's no need to adopt a dogmatic stance.
Now, let's shift our focus to the concept of ephemerality. The metaphorical infinite monkeys, tirelessly typing away on their typewriters, will inevitably produce mountains of wasted paper. There comes a point where the limited capacity of human beings to consume and regurgitate cultural content is overwhelmed by the seemingly endless efficiency of AI in emulating this process. By stepping away from deep and slow research, human artists can ride the wave of brute processing power, which can be exhilarating. One can "compose" 10,000 words a day of legible prose or create 300 images akin to Rembrandt-level etchings (although it's important to note that this glut of production exists primarily in digital formats like medium-resolution JPGs or PDFs). In a productivity-obsessed society like ours, this thrill holds a certain allure. However, it's crucial to acknowledge the fleeting nature of such creations at this stage. High-volume production must be seen as devaluing, and the valuable skill lies in sifting through the mass to extract the gems rather than merely cultivating more output.
In conclusion, reflecting on the creative process alongside AI tools like ChatGPT and stable image diffusion opens up intriguing avenues of thought. By incorporating examples, engaging with counterarguments, and considering ongoing debates surrounding AI and creativity, we can delve deeper into the impact of these tools. As we navigate the evolving landscape of AI-assisted creativity, we should strive to strike a balance between harnessing the power of these tools and maintaining the unique human perspective that brings depth and meaning to our creative endeavors.
