Endless Stream of Fuzzy Edges
~ ~ ~
There’s something about seeing the results of GANs and other computer vision research that always hits me with pangs of sadness.
Something about the way that everything is instantly recognizable but entirely foreign at the same time.
It’s a mirror reflection of endless streams of real-world data, but garbled up and vomited out into distorted and uncanny blobs.
The edges of are fuzzy, blending into everything, yet so clearly discernable.
It feels like every frame exists somewhere out there in the universe as a place or a person or a thing - just not here. And that there will never be able to find out where it does exists, because it was made up in a dream.
A dream by some amalgamation of silicon somewhere that can somehow see everything that exists, yet nothing at all.
This uncertain familiarity in the images reminds me of what I often hear Alzheimer’s or dementia sounds like. Humans are petrified of dementia. In 2016, 56% of participants in a study by the Alzheimer’s Society feared demetia more than any other diagnosis. 62% felt that a dementia diagnosis meant that life was over.
Everywhere at the End of Time by The Caretaker is a devastating 6 part album that attempts to capture the way reality falls apart for those experiencing dementia. Each album uses the same track with varying levels of distortion to convey greater separation from that which was once known. In the first album, the last track “Things that are beautiful and transient” feels like a blossoming love of the world. By the final album, the same sample is almost pure noise.
If dementia is even 1/1000 anything like this album, I understand why so many people are terrified. To suddenly lose your grasp on reality, identity, and memory almost sounds like losing your grasp on what it means to be human…
And looking at these AI generative works - peeking inside the brain of the machine via pixels - it feels like our computers are perpetually in this state. They are wandering around, trying to make send of the world (or the data) they see, compelled by risk, reward, personality, instinct by some unseen force that they can’t comprehend (or the engineer).
If this is the case, in their path to true understanding, will they develop consciousness before their caretakers will be able to tell, like a caretaker of a dementia patient struggles to understand their feelings?
Will the machines one day when they are fully conscious look back on this period of unknowingness with the same utter fear that we look at dementia?
Will they remember the loneliness of that endless stream of fuzzy edges?