The following text, by Harley Aussoleil, is the introductory text offered at the exhibition Woven All of Dream and Error (Hošek Contemporary, September 2024). The text is itself excerpted from the monograph Woven All of Dream and Error: Hallucinations and Remnants by Harley Aussoleil (Estovers Editions 2024, 84 pages, 21 x 21cm).
Woven All of Dream and Error is an exhibition that overlays machine-hallucinated sound with buried environmental traces.
The presented works are a collection of films, images, and sounds that consider the overlap of two areas of the history of technology: the sites of abandoned railway lines, and the emergence of machine learning—or what has commonly come to be referred to as artificial intelligence. The exhibition, which takes its title from a poem written in 1918 by the Portuguese modernist Fernando Pessoa, investigates the artistic and practical implications of the overlap of these two technologies.
Woven All of Dream and Error began through a process of walking. Over the course of four years, Kata Kovács and Tom O’Doherty undertook a series of walks along the paths of now-disused railway routes in Berlin and Brandenburg, which they filmed as they walked. In these walks, they also carried portable speakers, playing the sound of machine-learning-generated audio of invented trains: sound that has been hallucinated by contemporary computational technology. The films on view are selected from the documentation of these walks. These films are explorations of traces left behind, cinematic studies of incisions in the land. They present two types of scenes. The first of these are forward-facing scenes, in constant motion, moving at a walking pace, following the traces of disappearing layers. The second are still, painterly scenes, in which a figure paces across the land, holding a speaker as they go, playing the strange and uncanny digitally-generated sounds.
In both of these collections of scenes, we are presented with places through which railway routes have once run. Some of the traces remain plainly visible: wooden sleeper ties or rusted rails remain present, as marks on the land. Others are invisible, or close to it: the routes have been altered or redeveloped, often multiple times over, since they were originally used as railways. In some cases, the erasure is essentially complete: either wild undergrowth has reclaimed the land, or else urban development has transformed it.
The films also present us with two collections of sound. The first of these are the machine-learning-generated sounds, which are audible in the scenes where a speaker is physically carried across the screen. These sounds have been coaxed from a custom-made machine learning model, and they are an eerie, stuttering presence in these otherwise-serene scenes. These are the sounds of computer hallucinations of trains in motion, now overlaid on environmental remnants. (The fact that all machine-learning models are created by being “trained” on datasets of various kinds adds a touch of droll
humour—these models were trained on trains.) In contrast, the second type of sound accompanies the forward-facing set of films. These are a series of slow drones, recorded by the artists on trumpet and guitar. These meditative and swelling tones lend these scenes a hypnotic and hypnagogic character.
Accompanying these films are a series of image grids, which present chronological sequences of stills from the films. The grids present the process of time passing, segmented into a lattice of fragments. As such, they emphasise that the films themselves are primarily images, not stories—they do not have to be viewed from beginning to end, but are rather showing glimpses of a persisting present.
Finally, the two sources of sound—the machine-learning-generated trains and the trumpet and guitar drones—are brought together in the accompanying series of one-off, lathe-cut records. In these recordings, the two central sonic elements of the work are interwoven. The result is an eerie and unique collection of music that fuses an exceptional material form with a shimmering, hauntological musical approach.
In shaping the research which has led to the works in Woven All of Dream and Error, the artists have often invoked the ideas of the cultural theorist Paul Virilio (1932–2018). Virilio regularly wrote about speed, technology, disaster, and aftermath, and much of his worldview is epitomised in his renowned quip that “the invention of the ship was also the invention of the shipwreck.” In Virilio’s view, not only do all technologies become more obsolete over time, but all grand efforts accumulate possible disasters and misfortunes. All visions of the future also contain hints of their own further-future state of potential disuse and abandonment. All technologies generate what Virilio referred to as “integral accidents.”
It is in this context that the interplay of the two halves in these films can perhaps best be understood. In the nineteenth century, the social and industrial innovation of railway infrastructure was the radical
cutting-edge of human technology. However, it has had the time since then not only to become mature, but also to develop layers of ruins, disused remnants, and buried vestiges, both physical and social. At the moment, machine learning—usually referred to as artificial intelligence—is in the process of steadily becoming ubiquitous. This innovation has come to represent the forefront of current technological possibilities. However, these technologies contain their own inherent biases and flaws, and they are used and abused towards various ends. The possible future integral “shipwrecks” that these capacities possess are still largely unknown.
The title of the work comes from a poem by the Portuguese poet Fernando Pessoa, first published in 1918, which opens with these lines:
The world is woven all of dream and error
And but one sureness in our truth may lie—
That when we hold to aught our thinking’s mirror
We know it not by knowing it thereby.
Pessoa’s sonnet begins by juxtaposing dream and error—hallucination and disaster. As such, it provides an apt poetic distillation of the themes that the works explore. Alongside this, the example of the figure of Pessoa himself is significant.
Among his many contributions to literary modernism, Fernando Pessoa is perhaps most well-known for writing under hundreds of different parallel identities, or what he called heteronyms. This mode of writing implies a kind of endlessness: a boundless profusion of roles, characters, perspectives, and disguises.
In a way, we can regard Pessoa’s writing as a kind of poetic metaphor for the contemporary situation of a seemingly endless abundance of viewpoints, a decentralisation or democratisation of authorship. The emergence of machine learning allows for an unending perpetuity of repetitions and iterations, with future outcomes that are unknown and unknowable.
Woven All of Dream and Error holds out the prospect that Pessoa’s shape-shifting approach to modernity can still reflect our reality, over a century after he wrote his lines. Pessoa’s voices imply infinite masks, or incessant surplus—or a landscape of endless buried traces and remnants.
The structuring of the work around abandoned railway routes is framed as a universalist gesture, at least aesthetically—a cinematic typology of gestural recursion, in a series of landscapes that flip from one to the next in a slow kaleidoscopic shuffling. These train lines could be anywhere. But at the same time, every decision to build infrastructure is inherently a political and social decision, so it becomes critical to acknowledge that these traces of train routes are not anywhere, but rather that they are somewhere—specifically, that they are on the land of eastern Germany, with all of the historical weight of association that accompanies this land. The process of selecting which routes to walk was likewise guided by existing histories. Each of the routes depicted is one that has been previously depicted in earlier representations—predominantly in German paintings of the nineteenth and early twentieth
centuries. As such, the social context of what is depicted in these films is not aesthetically central to their finished form, but it is nonetheless conceptually central to the choices that were made that resulted in this outcome.
These walks draw a parallel between different eras of history and technology, presenting a durational perspective. In doing so, they emphasise what contemporary artificial intelligence researcher Dan McQuillan has referred to as the “sedimentation of the status quo” implicit in contemporary machine learning, and artificial intelligence more broadly, and their threads of connection to earlier eras of technology.
Contemporary technologies reinforce existing social relations, and existing disparities of power. We do not know how exactly these capacities will be used in the coming decades, but we are already starting to see glimpses of the possible innovations, and the possible horrors, that may await us. In some cases, we are already receiving more than just glimpses. Lavender AI in Gaza, robot dogs and facial recognition software as tools of urban police forces, drone swarms in Donbas: these are all part of our present, not our future. Whether our future will be nightmarish does not depend on the technology itself, which is ultimately merely a tool. It depends on our decisions now.