technobiogenesis is 1. my bachelor in visual arts' thesis, developed mostly throughout 2022; 2. a broad concept i apply to other projects as well. i like using the same name for multiple things, because i like being confusing and mixing metaphors. at its core, it simply means life created from technology.
this page specifically is an introduction to the executable i made for my thesis. for multiple reasons i'm not gonna post the whole pdf, and focus on this simulated fictional space. if you'd like to view it first and ask questions later (as it were), here's a download link (available on linux, windows and mac).
simply put, technobiogenesis (the executable) is set on an earth where lifeforms are created from data and metadata. see also: dataplankton, which is still in progress but set against the same background. as the scale of the obsession with datahoarding grows, so does the size and complexity of this hyperreal world and the life it spawns. the search for the perfect universal dataset, a map the size of the territory, creates a world so large it collapses under its own weight, merging earth-as-we-know-it and the hyperreal-data-earth.
technobiogenesis (the executable) is a simulation of this post-apocalyptic world. the word simulation is important, because i never intended it to be a faithful representation of what this new earth would look like within the fiction. so it functions more or less as a gallery of itself. the synthetic images used throughout are snapshots of the data world, arranged in a crescendo of complexity to represent the passage of time, until the final, largest image-structure pair, which represents the collapse.
this is, of course, a purely aesthetic piece of software. whatever you'd call "mechanics" or "interactivity" is just for that end. so here's a little breakdown of the visual elements that make up the environment.
the executable uses a 480 x 270 resolution, stretched to 1280 x 720, to give it some pixelated crunch. the draw distance is also pretty short, so elements often abruptly appear or get cut off in the peripheral vision, and there's of course the fog. all of this was born out of my personal interest on retro games (i am not immune to the psx revival era), the themes of unrestrained technical advancement as a destructive force (and subsequently a desire for the work to be lightweight enough to run on "bad" hardware, although i have no real way to test that out), and a desire to keep this environment clearly simulated (as in, no pretense of grounding it in reality or getting people to suspend their disbelief).
for starters, some terminology: i've been using the phrase synthetic images in lieu of "AI art", for multiple reasons. first, while "artificial intelligence" can be useful shorthand, it has too many sci-fi connotations, and doesn't really mean anything. and i'd much rather avoid a discussion on the "art" part at all. i'm not sure if "synthetic image" is that much better (it's definitely not as clear), but i prefer it regardless. in particular, i like the generic "image" as well, since these are only ever bundles of pixels. it annoys me to overuse it, but you gotta do what you gotta do.
so! synthetic images are the output of neural network image generators. i've used four to create the images in here:
i didn't really generate anything specifically for this project, choosing instead to use images i'd already generated since i started playing with these tools in 2020. i talk a little more about it on my april artlog. i curated with a focus on locations and creatures, to convey the idea of synthetic images as windows into the hyperreal data world.
while the executable is technically nonlinear (since it's simply an environment), the layout is more or less chronological within the fiction. so the images grow in size and resolution, and become billboards as life begins to sprout. i love the somewhat confrontational nature of billboards, images that reverse the way we think of viewing art, images that turn to face you. i used the same thing for a little synthetic gallery, an assignment in early 2021.
the paintings (ie the floating creatures), are the fruit of pareidolia, from a single artbreeder image. i just drew on top of it, looking for marine animal looking shapes. like most of the assets, these predate the project by a fair bit.
while the synthetic images are representations of the hyperreal, the paintings are the new lifeforms themselves. i don't think that's necessarily clear, but i don't mind that too much.
all meshes in the executable are simple primitives, grabbed directly in godot. i was mostly interested in combining them in ways that can only exist digitally, and adding complexity as the data world grows. so each of the four "hubs" is a microcosm of a single image, with the "final" hub, the tower, representing the collapse itself.
i had a great time with the interlocking polyhedra. much like the billboard images, they are all about the physical impossibilities that can be achieved through computer graphics.
this covers my aesthetic choices and so on, which means it's now time to tackle neural networks, the mess that currently is "artificial intelligence", and the fiction of technobiogenesis.
the fiction of technobiogenesis lies in a odd spot of mixed metaphors, given my love for apocalypses (see also: transgenderize the eschaton, and my plans for dataplankton).
in any case, i avoided a rogue AI narrative, with the collapse being caused by obsessive datahoarding whereas the artificial intelligences are just neutral lifeforms. the apocalypse is a manufactured cosmic horror taking the shape not of a singularity, but of an environmental catastrophe.
i wouldn't call this view realistic, obviously, but it aligns more closely with the current state of neural networks and the capitalist approach to technology in general. the whole internet is considered free for the taking, there's the illusion of creating an AI that'll make perfect images of anything, write whatever you ask. even replace search engines with pure fabrication now (a natural evolution of seo soup that plagues search engines for years now)! there's an obsession with compressing the world into a model.
all very uninteresting and potentially harmful uses, concocted by uninteresting and definitely harmful minds in the silicon valley that want to automate everything. of course, all this relies on extremely cheap (or free) unseen human labor (whether that's crowdsourcing people labeling images a la mechanical turk or internet scraping, and also all the cutting edge hardware used for training), and handily (for the dominant classes) tech-washes biases.
for further reading, here's a handful of articles and books i referenced in my thesis.
so, with all that said (or if you just skipped directly here through the index), here's the download link, available on linux, windows and mac ^_^