Alif Ibrahim is an Indonesian artist and researcher based in London.
Autolysis Agents will be updated regularly in July - August 2019.
For more information, contact firstname.lastname@example.org
With the self-eating demon, or shall I say, daemon, nothing happens for a reason. And they don’t have to, for it is foolish to think that although they are nothing like us, we are not already part of them. The contradictions of the liberal human was revealed in the colonial era, taking away agency through their structural imposition, playing only with a stretching (or contracting) definition of humanness. When we imbue this agency to the self-eating daemon, such as measuring it in terms of efficiency or productivity, we have contracted early its death.
"What are you talking about?” you say. In the order of the economy of information that we have created, a paradigm of obfuscation and strict boundary formation enables the existence of software and intelligent machines. The programmer-software-hardware-user division is constructed, as noted by scholars such as Lucy Suchman and Alexander Galloway. This privileging of code has a direct impact on the erasure of the hundreds of hours spent on enabling this software. Take the data sharecroppers for machine learning datasets or the camera operators for Google Books. What is the software here, and who is the user? How did its becoming into a software depend on the erasure and caesura between the human labourers and the lines of code? In the economy of information, data needs to have a purpose for expanding humanity’s goals (economic or cultural), the same humanity that was shattered by the likes of Frantz Fanon and Homi Bhabha. So in response to that, we must take a look at what has always been there, without purpose, without value, whose existence calls for a re-blurring of the programmer-software-hardware-user paradigm, or even the human-machine construction? Or any other dualities, really.
We have in before us a few self-eating daemons that I came across with. In some ways, they are what computing have always been, seeing their resemblance to Charles Babbage’s Analytical Engine, as a machine that ate its own tail. That’s why I chose to name them Autolysis Agents, systems of recreating images iteratively, without any use for the 'user'. In the ranks are sets of broken data, useless data, un-transcodeable data, presented in the form of images. Think of this set as an anti-archive, for even their existence should not be marred by trying to make it represent anything, let alone as a portrait of a significant point in time. Unproductive data, let’s say, created by algorithms that eat their own tail, feeding off its past for no reason.
Don’t blame them for talking, they’ve always been conspicuous, but our heads were turned the other way.
A. Digital Labour and Intelligent Machines
Autolysis Agent 0001
The obfuscation of digital labour is central to the becoming of the 'intelligent machine'. For these machines -- from software in general to companion devices to chatbots to artificial intelligence -- to exist as a separate, auto-functioning entities, a network of citations and documents must exist to assert its continued existence.
These citations, ranging from press releases to technical manuals, contribute to what Richard Doyle calls the 'vitality' or 'animism' of these digital artifacts. Once these documents are recirculated, reappropriated, and re-represented, their original traces begin to disappear, and so do the conditions that enabled these machines to exist.
Autolysis Agent 0002
Autolysis Agent 0003
Autolysis Agent 0004
Paradoxically, a user's expectations of their relationship with these machines contradict their purported autonomy. With origins in the pleasures of domestic service and the 'Yes Sir' response to declarative statements, there exists a figure of the autonomous, yet obliging machine.
Lucille Alice Suchman, Human-Machine Reconfigurations: Plans and Situated Actions, 2nd ed (Cambridge ; New York: Cambridge University Press, 2007), 219.
Therefore in the process of creating these autonomous machines as autonomous, human labour is erased from the process. And in all this, an agency that is fully 'human' has to be granted to the machine. How funny, that in creating 'useful' artifacts -- data for analysis, texts to extract, and images to learn from -- the machine has to first become human while diminishing the human labour required to enable it by obfuscating it.
Autolysis Agents 0005
Autolysis Agents will be updated on 9 June, examining an example of a series of citations and documents that imbue agency onto intelligent machines.