Training the Archive – Conference on Art and Algorithms

 

On 17 & 18 November 2022, a conference will be held on the topic of Art & Algorithms as part of the research project Training the Archive at the Ludwig Forum Aachen. On this occasion, renowned scholars will provide an overview of the current state of research in the application of algorithms in the context of art and museums.

Thursday, 17 November 2022 18:00 – 20:15 | Friday, 18 November 2022 09:00 – 16:15

Free admission. No registration required.
Moderation: Mads Pankow

The conference will be bilingual on Friday. All presentations will be simultaneously translated German/English and English/German.

--- Programme on Thursday, 17 November 2022 ---

PROLOG REFLECT
Introduction to the History and Present State of Algorithms and Culture

RECEPTION – 18:00 (in German)
Eva Birkenstock, Ludwig Forum for International Art Aachen
Julia Mai, Kulturstiftung des Bundes, Halle (Saale)
Prof. Dr. Leif Kobbelt, Visual Computing Institute, RWTH Aachen University
Dr. Inke Arns, HMKV Hartware MedienKunstVerein, Dortmund

SYBILLE KRÄMER – 18:15 (in German)
»Digital Culture. Fundamentals and Ambivalences of Algorithmisation«

We associate the algorithmic with computers. But is this true? In fact, there is an embryonic digitality before electronic networking! Is the transition from computational rules to recommendation algorithms an inversion of empowerment into uncontrollability? The keynote seeks answers to the question of what a ‘cultural technique of digital literacy’ consists of.

INKE ARNS, SYBILLE KRÄMER & YVONNE ZINDEL – 18:45 (in German)
»Culture between Humans and Machines: Who Curates Whom?«

How do algorithms influence artistic and curatorial work? What does the data reveal about us and our biases? And who programs whom in the end, the human the Algorithm or vice versa? A panel discussion.

HANNES BAJOHR – 19:30 (in German)
»Sequential and Connectionist: Digital Literature with and without AI«

Digital generative literature exists as a sequential and connectionist process – the result of classical, linear algorithms and Machine Learning. In his lecture performance, the author Hannes Bajohr argues for the respective aesthetics of these approaches and reads examples from his poetic work.

Get-together with food & drinks afterwards.

 

--- Programme on Friday, 18 November 2022 ---

PLENARY – 9:00
Summary of the Previous Day

PANEL SELECT
Selection as a gesture in the possibility space of current AI models

FABIAN OFFERT – 9:15
»On the Concept of History (in Foundation Models)«

This talk discusses the concept of history inherent in so-called ‘foundation models,’ focusing on OpenAI’s CLIP model, a large-scale multimodal model that projects text and image data into a common embedding space. How does the idiosyncratic historical perspective that emerges from such models impact their potential as art historical tools?

ROLAND MEYER – 9:50 (in German)
»Images on Command. Text-Image Relations in DALL·E 2«

Text-to-image generators like DALL·E 2 promise to produce any possible image on command. The virtual archive of textually annotated and digitally mobilised images of the past serves them as a seemingly inexhaustible data resource – with consequences for our concept of images that are just beginning to become apparent.

TILLMANN OHM – 10:25
»Curatorial Algorithms and Collection Spaces«

Cultural heritage such as art collections can be quantified and embedded in shared computational spaces including vector embeddings and networks. Interfaces and algorithms help researchers and curators to explore these possibility spaces. This talk provides an overview of such techniques and introduces examples of algorithmic curation.

PANEL RETRIEVE
Multimodal access to museum collections and iconography.

EVA CETINIĆ – 11:30
»Multimodal Models as Cultural Snapshots«

Trained on millions of image-text pairs sampled from the internet at a certain point in time, multimodal models learn not only literal similarities between images and words, but also various cultural references. As synchronic cultural snapshots embedded in a specific technological framework, they become not just tools but also new objects of study.

DOMINIK BÖNISCH – 12:05 (in German)
»How to Train the Curator’s Machine«

The ‘Curator’s Machine’ is a Prototype that enables curators to approach museum collections intuitively and free of fixed search terms via a multimodal retrieval system. An explorative process of human-machine interaction also includes the experts’ contextual knowledge in the machine learning.

VINCENT CHRISTLEIN – 12:40
»Pose Estimation for Artworks Retrieval«

Human pose detection represents an important factor in scene analysis of artworks. Additionally, ‘pose estimation’ can be used for an interpretable image retrieval. This talk will discuss pose estimation and its role in image retrieval in various fields of application such as classical archaeology and art history.

PANEL COMBINE
Curation as a Link between Humans and Machines

GAIA TEDONE – 14:15
»Human-Algorithmic Curation: Between the Visible and Invisible«

The presentation discusses human-algorithmic curation as a networked practice that is both reflexive and playful and that can engender new forms of cooperation and value co-creation between humans and technical agents. It speculates on its potential to expose the invisible power structures and technical asymmetries of the networks.

KATRIN GLINKA – 14:50 (in German)
»Old Canon, New Bias«

In her lecture, Katrin Glinka emphasises the necessity of designing algorithmic systems in such a way that they support a critically reflective use. She contrasts technology-inherent potentials of algorithmic procedures with regard to accessibility and representation of objects with established structures of museum knowledge representation.

GEOFF COX – 15:25
»Towards Human-Machine Visual Literacy«

Referring to “Ways of Seeing” (John Berger, 1972) the presentation argues for an expansion of visual literacy to examine how machine vision further unsettles received humanist notions. When images are made by machines for other machines, and part of vast annotated datasets, how are worldviews reinforced differently, and what kind of literacy applies, if at all?

PLENUM – 16:00
Summary of the Conference

 

----

Training the Archive (2020 – 2023) is a research project that explores the possibilities and risks of AI in relation to the automated structuring of museum collection data to support curatorial practice and artistic production. Further information can be found here.