Trevor Paglen - Bloom & Further Research.
“This is one of the philosophical dangers of using widespread automation, which is, that it fixes meaning.” - Trevor Paglen
Trevor Paglen is an American artist, geographer and author, born in Maryland, USA in 1974 and currently lives and works in Berlin, Germany. His diverse education spans a BA in religious studies, an M.F.A, and a Ph.D. in Geography. Paglen tackles topics of mass surveillance and data collection, collaborating with scientists and human rights activists in his ambitious multimedia projects. Paglen’s work studies the contrast between secrecy and revelation, evidence and abstraction. The artist aims to bring awareness rather than “evidence” into public consciousness.
ImageNet - “Training Humans” a project with Kate Crawford:
Kate Crawford is a leading academic on the social and political implications of artificial intelligence, her work focuses on understanding large-scale data systems and AI in the wider contexts of history, politics, labor, and the environment. The project “Training Humans” was a piece that exposed facial recognition biases that led to a leading database to remove more than half a million images. The piece was a internet based program that allowed people to upload images of themselves to a website where an AI trained on the most widely used image recognition database, analysed what it saw. The results were deeply problematic. People of colour were labelled “wrongdoer, offender”, an Asian woman as a “Jihadist”. A child wearing sunglasses is labelled as a “failure, loser, non-starter, unsuccessful person.”
This was precisely the aim of Paglen and Crawford. To expose how systemic biases have been passed onto machines through the humans who trained their algorithms. The project uses the expansive database ‘ImageNet,’ which was first compiled by researchers at Stanford University in 2009 to develop algorithms applied in deep learning (a process by which machines are trained to recognise images, to thereby understand the world) and us one of the most widely used training sets for machine learning. This database by which an algorithm is trained, contained thousands of photographs of people, sorted into descriptive categories. These categories range from “cheerleaders, flower girls” to more loaded terms like “slut, slovenly woman, trollop, alcoholic, failure”. This is extremely problematic. These labels were assigned purely by humans in labs or strangers paid to label images through crowdsourcing tools.
Both Paglen and Crawford stress that their project calls attention to the problem of categorizing people in this way - especially given the rapid uptake of algorithmic systems within institutions from education, to healthcare to law enforcement. “The whole endeavour of collecting images, categorizing them, and labelling them is itself a form of politics, filled with questions about who gets to decide what images mean and what kinds of social and political work those representations perform.” (Paglen, 2020).
“Training Humans” explores two fundamental issues - how humans are represented, interpreted and codified through training datasets, and how technology systems harvest, label and use this material. Within computer vision and AI systems, forms of measurement easily turn into moral judgements. Paglen expresses the cultural link between these images and colonial past, by capturing people’s images without consent in order to classify, segment, and often stereotype them in ways that evokes similar practices to the colonial past. Whereby Crawford adds that, “there is a stark power asymmetry at the heart of these tools”, the tools of AI must be critically analysed and understood before or if these systems are implemented globally, as with them may bring much false, inaccurate, meaningless, but nonetheless, dangerous categorization.
Bloom @ PACE Gallery - London.
Pace Gallery, founded in 1960, is a leading contemporary art gallery representing many of the most significant international artists of the 20th and 21st centuries. PACE currently has 9 locations, one of which being Burlington Gardens, London. The galleries play a critical role in the shaping of history, creation, and engagement with modern and contemporary art globally.
Trevor Paglen - Bloom / Octopus (Sep 10th - Nov 4th, 2020)
The exhibition titled Octopus allowed visitors globally to virtually experience the London exhibition through a live web portal connected to cameras placed inside the gallery. Online viewers see both the work, as well as visitors experiencing the work, and if they so wish, they are invited to be “present” in the space by streaming their webcam feed to monitors displayed within the exhibition. This work responds to the post-COVID era, allowing options for both digital and physical viewing entwined, making either no longer exclusive, offering a brand new perspective on virtual engagement with a gallery space.
Paglen describes the exhibition to be about mourning, working, and being alive - specifically in a moment in time where fragility is all around us, and a part of our own lives during COVID. These themes of fragility entwine with his infatuation with technological layers. The exhibition Bloom is built on a series of photographs of flowers and plants in bloom. The photographs are taken with a very high resolution camera and are then subjected to artificial intelligence algorithms that he has developed in the studio that attempt to dissect those images into their component parts, i.e., textural differences, objects and regions. These differences in the images are then represented by assigning them different colours.
Paglen reflects on his thoughts about flowers being a constant reminder through art history and art images of the fragility of life, and recounts that at this time, “the sense of fragility is so present”. Reverting back to technological aspects of the work, Paglen insists, “we’re living in a moment where we’re seeing things like a rise in AI, to which a large extent is about automating the interpretations of images.” Here Paglen reminds us that whilst we live in such fragility and the blooming of flowers is fleeting, that AI and institutions of surveillance still leer, a hard contrast to the previous reflections on flowers.
Here we see Paglen’s true motivations for the exhibition, Paglen, known for his research and exploration into the nature of AI and systems of automatic recognition (i.e., facial recognition, computer vision, AI). We learn that through the rise of AI, we must become more aware of the nature of such technology. For as Paglen believes “technologies are never neutral” and they never could be, they are taught by humans with conscious and unconscious opinions of their own, and therefore should not be the basis for a technology making widespread judgements on people via categorization.
Paglen encourages us to think about this dislocatedness within media through the dozens of cameras throughout the gallery, those cameras streaming to live a live web platform, and allowing the webcam viewer to be seen by the person experiencing the exhibition inside the gallery itself. Posing further questioning and thinking toward remoteness, presence, and the different kinds of mediations and abstractions in these relationships, as well as being conscious of the interactions with each other as human beings, that have become a part of a machine that is designed to extract as much valuable information from us as possible for benefit of those in power, to the expense of those who become objects of observation.
