noise is both the presence and absence of information
we are surrounded by extra noise, hums of traffic and electricity
noise to drown out noise - noise cancelling headphones through active noise - this is hypermediation
oversaturated with data, much data is noise
noise is the ‘success’ state of the information age
information was once scarce and is now too prevalent
information age started with the transistor - directing and controlling voltage according to various inputs
punch holes remediated to electrical on off remediated into binary remediated into assembly code and every code since and then GUIs
the more we feed the machine, the more decisions the machine can make
but decision-making is power.
decisions encroach on our own, we no longer have to make them
it may seem as though mathematically, scientifically made decisions are more worthy but our entire societies function on irrationality as well as rationality, and science/maths aren’t inherently neutral but embedded with societies’ problems
we always find ways to recreate dichotomy, good and evil, yes and no, swipe left and swipe right
with barebones circuitry the computer is designed to ‘reveal’ information and noise is any ‘disturbance’ to the flow of energy through the wires
the ‘noise’ of the information age eg tv and phone static is generally cosmic background radiation
radiation retained from the big bang -> omnipresent force -> godlike -> divine machinery, ghost in the machine, folklore
generative AI essentially also works on an accept/reject signal
every eg caption becomes information, when does the human consent for their expressions to be quantified in this way
“The entirety of human visual culture has a new name. We used to call these collections archives, or museum holdings, or libraries. Today, we call them datasets.”
AI image generators are a super-combination, quantum superposition of the visual media offered (etc for nonvisual this is just more clear)
but decisions are made on the curation without a human hand, without input source consent and with undue credit to the puppeteer
these works in libraries etc are treated with care, require citation, there was some credence given to the wishes of the author - no ‘do not mutilate’ for AI source data
our collective visual heritage in digital form is remixed with less care, without provenance
personal works with personal contexts and relationships to their maker and necessary deliberate mediations are clumped together as a pot of humanity’s work decontextualised
use of the word ‘diffusion’ for AI is interesting - aspects of aesthetic and content and value diffused throughout the new pieces
Server farms mined data with guiding principles of surveillance and financialisation
the goal of social media has always been to have us sharing more
generative AI is just another term for surveillance capitalism
late stage capitalism looks like uncanny images of machines with no experience of the world, unsettling as they don’t resemble what they represent, as automated surveillance never does
AI is abstraction and prediction. (Often the wrong abstractions, and the moneymaking predictions)
what survives history is what we choose to preserve through a series of institutionalised decisions
archives more clearly show their gaps, and the gaps can be challenged. AI conglomerates into a soup.
when humans are in the loop, humans can intervene in the loop
a dangerous promise of a paradigm shift
each combined image is actually degraded images
AI navigates averages. Averages are not always useful
the original context is irretrievable
abandoned meanings - meanings occur from relationships
echoes of colonialism - to reduce history to a singular flat ground from which to build
nothing holds more weight than anything else - victims and perpetrators fused together
a generated image is not new, it is wrong - the hypothesis of an image that might exist, but doesn’t
we are all fused together as one and sorted by the weight of the most common
we can’t study the geanologies of these images
the idea of the remix is that we CHOOSE elements to rearrange, arrange them to the moment
a haunting
Mark Fisher sonic hauntology, music that references past visions of the future
data is haunted by what it didn’t include, even the census eg children not born because of who could not marry
data contains only the trajectories of history that have been allowed to us
feeding only past data means the ghosts become police
the living are sorted by the dead
we refer to datasets in AI rather than archives as archive proposes q greater degree of responsibility. Reliance on humans and curation and contextualisation
datasets strip context
archives group and sort by meaning but datasets only by quantified things
the avant-garde as liminal space between archive and noise
residue, telling the machine it doesn’t know anything
‘erotic footage’, the machine leering, tracking and tracing a body as just as invasive when a machine does it
following a bodies contours is incredibly intimate, why do we let machines do it
but in being translated into media the image appears to enjoy itself. Literally submerging herself in the data
there is such a thing as Art that involves AI, but art is not just what AI generates.
creativity is in agency
contorting the machines towards us
fear with AI is more than we will fall into historic grooves with no concern for building a deliberate future
Nam June Paik “I use technology in order to hate it properly”
archives as collective memory
the challenge of the age of noise is to find and preserve meaning