Animated Intelligence – Reformulating the Problem of Pattern Discrimination

Clemens Apprich, University of Applied Arts Vienna

Based on current research on ‘pattern discrimination’ (Apprich et al. 2019), I want to look at the creative potential of machine learning algorithms. While pattern discrimination describes the filtering of data on the basis of algorithmically calculated and, therefore, pre-defined patterns, it also points to the fact that with each iteration the outcome of the filtering process is slightly different (Apprich 2020). To deal with this creative, but also highly political issue of algorithmic filters, artistic approaches provide incentives to better understand data filtering and the related phenomenon of machine learning (see, for example, Crawford/Paglen 2019). Underlying is the somehow paradoxical question of how to filter, that is to discriminate information from data without being discriminatory? How can we make use of filtering algorithms without reinserting retrograde identity politics, based on categories such as gender, class or race?

In seeking to answer this question, I will pursue an alternative idea to today’s predominant conception of machine learning and artificial intelligence: instead of trying to explain what intelligence is, I will focus on the creative processes needed to generate machine intelligence. Intelligence, in this perspective, is less individual but rather collective. It is not something that sits in our head, a uniform thing that can be reproduced in form of artificial neural networks. Rather, intelligence is external, a social ensemble of human and non-human elements that animates intelligent behavior and, as a consequence, also messes with pre-defined social categories. By shifting the focus from simulation to animation, I seek to reformulate the problem of pattern discrimination: not only with regard to its ethical implications, but also, and in particular, to aesthetic considerations needed to pose a critique of artificial intelligence and machine learning.


Apprich, C., W. H. K. Chun, F. Cramer, and H. Steyerl. Pattern Discrimination. Minneapolis/Lüneburg: University of Minnesota Press/meson press, 2019.

Apprich, C., “The Never-ending Network: A Repetitive and (thus) Differentiating Concept of Our Time.” In The Eternal Network, edited by K. Gansing and I. Luchs, 24-32. Amsterdam: INC, 2020.

Crawford, K. and T. Paglen, “Excavating AI: The Politics of Training Sets for Machine Learning,` The AI Now Institute (2019). Available Online:

BIO | Clemens Apprich is full professor in media theory and history at the University of Applied Arts in Vienna, as well as guest researcher at the Centre for Digital Cultures at Leuphana University of Lüneburg. He is an affiliated member of the Research Centre for Media Studies and Journalism at the University of Groningen, of the Digital Democracies Institute at Simon Fraser University, and of the Global Emergent Media Lab at Concordia University. His current research deals with filter algorithms and their application in data analysis as well as machine learning methods. Apprich is the author of Technotopia: A Media Genealogy of Net Cultures (Rowman & Littlefield International, 2017), and, together with Wendy Chun, Hito Steyerl, and Florian Cramer, co-authored Pattern Discrimination (University of Minnesota Press/meson press, 2019). He is founding co-editor of spheres – Journal for Digital Cultures (


Variations on a Glance

Nicolas Malevé, Centre for the Study of the Networked Image

Today’s digital platforms increasingly rely on computer vision algorithms to classify, filter, label, censor, augment and organize their visual content. The recent breed of algorithms performing these tasks are often based on a deep learning framework and their efficacy depends on the quality of their training. In this context, the training consists of feeding a program with huge curated sets of data from which it “learns” regularities. The production of these datasets requires an infrastructure at web scale. A large population of precarious workers, recruited on crowdsourcing platforms, annotate billions of images to describe their contents to machines. In this economy of looking, a certain way of seeing is privileged: the glance. As the cost of gathering annotations is bound to the production rate of the annotators, they are working at a pace that barely allows them to see the images. For computer vision, the glance has become infrastructural.

Crucial in the annotation environment, the glance is also a privileged object of experimental research in the computer vision lab. The glance is studied as a model of vision through various psychological experiments with the objective of speeding up and optimising the annotation process. An early experiment conducted in 2007 at Caltech by Fei Fei Li, initiator of ImageNet, one of the most popular visual datasets, offers a case in point. In a laboratory, the subjects were asked to describe photographs shown for a few milliseconds and to filter them through a taxonomy.

This presentation introduces a practice-based research engaging performatively with the Caltech experiment. The project Variations on a Glance, has developed in collaboration with The Photographers’ Gallery, London. It consists of a series of re-enactments of the Caltech experiment. The experiment is used to engage with the model of vision and the micro-temporal rhythm that subtend the modes of production of labelled data and the labour behind it. The original experimental protocol is submitted to several variations, called re-experiments, exploring its potential to produce a time-critical model of vision and collective visual interpretations. The experimental protocol is re-designed iteratively to explore specific configurations of micro-temporal vision and different collectives of human and non-human participants. The presentation will examine the dynamics of these collectives, in particular how they reach consensual interpretation, and how the taxonomic practices of the lab interfere in this process. To conclude, the presentation will reflect on the potential for conducting collaborative research in cultural institutions that challenges the norms and assumptions of computer science.

Bio | Nicolas Malevé is an artist, programmer and data activist living in Brussels. Nicolas has recently completed his PhD at London South Bank University, as part of a collaboration with The Photographers Gallery. In this context, he initiated the project Variations on a Glance (2015-2018), a series of workshops on the photographic elaboration of computer vision. He is currently a postdoc researcher at Centre for the Study of the Networked Image, London, and a research assistant in the Research Group Visual Narrative at Lucerne University.


Terra Analytica

FRAUD (Francisco Gallardo & Audrey Samson)

“[T]he garden is a rug onto which the whole world comes to enact its symbolic perfection, and the rug is a sort of garden that can move across space” (Foucault 1986:6).

Composed of two garden carpets suspended as hammocks, Finis-terra (2018) by artist duo FRAUD (Audrey Samson & Francisco Gallardo) calls attention to the practices and politics of datafication. Among the key purposes of satellite technologies today is the translation of terrestrial phenomena – weather patterns, agricultural yields, resource reserves, etc. – into computable data. More than a benign means of representing the earth as image or model, this data often facilitates a “bird’s eye” sense of authority and ownership over the spaces captured, in turn enabling practices of financial, surveillant and political speculation and action. The densely woven rugs visualize the technical dimensions and applications of satellite data. Their presentation references Columbus’ appropriation of the Mayan hammock as a perch from which to survey the world.  

FEATURE IMAGE: FRAUD, Finis-Terra, 2018 — Image Credit: Carolina Vasquez-Lazo



Related Projects