Project timeline

A chronological overview of all project outputs, important milestones, and the strands of research connecting them.

Publication Chapter

The Resource Debate in Machine Translation and Large Language Models

Paolo Caffoni contributes to a reference entry on the language resource debate in LLMs to the Handbuch Soziale Praktiken und Digitale Alltagswelten

Pedagogy of Machines
LLM Translation Resource
v
Publication Article

The Cultural Politics of Artificial Intelligence in China

Qiaoyu Cai examines the cultural politics of AI in China through the lens of postsocialism, proposing the concept of a ‘postsocialist AI’ that goes beyond the dominant paradigm of neoliberal informationalism

China Postsocialism Neoliberalism Informationalism
v
Publication Conference poster

Stitching Sparse Autoencoders of Different Sizes

Patrick Leask and Noura Al Moubayed introduce SAE stitching, a new method for mechanistic intepretability, in a poster at NeurIPS 2024.

Latent Mechanistic Interpretability
SAE Stitching Stitching SAE Sparsity Autoencoders Latents Mechanistic Interpretability
v
Publication Conference poster

BatchTopK Sparse Autoencoders

Patrick Leask contributes to BatchTopK, a new SAE architecture introduced in a NeurIPS'24 poster.

Latent Mechanistic Interpretability
BatchTopK SAE Sparsity Autoencoders Mechanistic Interpretability Architecture
v
Event Conference talk

Carceral Diffusion: From Galton's Criminal Composites to Horse-riding Astronauts and Beyond

Eleanor Dare gives a talk in the frame of the “Eugenics, Race, and Colonialism in the AI Boom” panel at the University of Virginia’s conference “Rethinking the Inevitability of AI, Part 2: Assimilation and Refusal in The History of AI”

Exploratory machine learning Interfaces
Eugenics Generative AI
v
Event Keynote

Response-ability in Sociotechnical Systems Design

Claude Draude gave a keynote on “Response-ability in Sociotechnical Systems Design” at the 10th conference of the Gender Studies Association Austria (ÖGGF).

AI Design Interventions for Social Diversity
Response-ability Design
v
Event Workshop

Participatory and socially responsible technology development

Claude Draude takes part in a workshop discussing user participation in technology development

AI Design Interventions for Social Diversity
Social responsibility Participatory design HCI
v
Publication Journal Article

Theories of Automation from the Industrial Factory to AI Platforms: An Overview of Political Economy and History of Science and Technology

Matteo Pasquinelli evaluates the state-of-the-art in the debate on automation, advocating for a renewed dialogue between political economy and the history of science and technology (HST)

Automation Political economy STS Platforms HST
v
Event Keynote

Spatial Ideology and Speculative Spaces for Critical Education and Performance

Eleanor Dare delivers a keynote at the 4th International Conference on Possibility Studies

Exploratory machine learning Interfaces
Spatiality Arts-based approaches Materiality Digital space Virtuality Deleuze
v
Event Installation

Worldbuilding

Eleanor Dare creates an intra-active arts installation for the the 4th International Conference on Possibility Studies

Exploratory machine learning Interfaces
Worldbuilding Technology Design Intelligence Mechanistic interpretability
v
Event Workshop talk

The Resource Debate in Machine Translation and Large Language Models

Paolo Caffoni presents at and chairs a workshop organised as a collaboration between the KIM research group and the AIModels project at Ca’ Foscari University of Venice.

Pedagogy of Machines
LLMs Machine translation Linguistics
v
Announcement Research Residency

Algorithms + Slimes

Goda Klumbytė and Ren Loren Britton collaborate on artistic-academic research on slime moulds

AI Design Interventions for Social Diversity
Slime mould Algorithms
v
Event Workshop talk

Diagrams, Transpositions, Implicated Futures. Some Strategies for Critical Technical Practice

Goda Klumbytė gives a talk as part of the workshop ‘Critical computing. Calculating with relations’ at HBK Braunschweig

AI Design Interventions for Social Diversity
Diagrams Transpositions Critical Computing
v
Event Workshop talk

Experiential Heuristics

Goda Klumbytė and Dominik Schindler present a talk at the After Explainability workshop

AI Design Interventions for Social Diversity
XAI Experiential heuristics
v
Event Workshop talk

XAI as Science

Fabian Offert, Paul Kim, and Qiaoyu Cai present a talk at the After Explainability workshop

Interpretability and Accountability of AI Systems in the Sciences
XAI Science
v
Event Workshop

After Explainability: AI Metaphors and Materialisations Beyond Transparency

Goda Klumbytė In collaboration with the “Gender, Medien und Affekt” research network

AI Design Interventions for Social Diversity
XAI Explainability Transparency
v
Event Workshop talk

Accountability

Arif Kornweitz presents a talk at the After Explainability workshop

AI Interpretability and Accountability in the Humanitarian Sector
Accountability XAI
v
Event Workshop talk

Tokens of Translation: The Political Economy of the Sign

Paolo Caffoni contributes to a workshop on AI and linguistic turn in political economy

Pedagogy of Machines
Machine translation Language
v
Event Talk

Artificial Intelligence Design and the Logic of Social Cooperation

A conversation with Matteo Pasquinelli on AI and the intelligence of social relations

Automation Intelligence Labour Social relations
v
Event Workshop

Labour, Intelligence, Automation: Tracing AI’s Coming of Age

A workshop with Matteo Pasquinelli on the theoretical and political challenges that surround AI at the time of its coming of age

Labour Intelligence Automation
v
Publication Preprint

Synthesizing Proteins on the Graphics Card. Protein Folding and the Limits of Critical AI Studies

Fabian Offert, Paul Kim, and Qiaoyu Cai look at Meta’s ESM-2 protein folding ’language model’, asking what kind of knowledge the transformer architecture produces?

Interpretability and Accountability of AI Systems in the Sciences
Transformer architecture Protein folding Language models LLMs
v
Event Workshop

AI Forensics IRL Meeting

The AI Forensics team assembled in Berlin for an internal workshop

AI Forensics Interpretability
v
Publication Book chapter

Projektbericht: Maschineninterpretation mit Interpretationsmaschinen. Explainable Artificial Intelligence als bildgebendes Verfahren und bildwissenschaftliches Problem

Fabian Offert reports on the project for a volume on pictorial forensics.

Interpretability and Accountability of AI Systems in the Sciences
XAI Interpretability
v
Event Conference talk

Organism, Machine, Language, Model: The Making of the Artificial Mind between Animism and Formalism

Matteo Pasquinelli lays out a combined intellectual genealogy of AI

Organism Machine Model Genealogy Language Linguistics LLMs Cybernetics
v
Publication Conference paper

CXR-IRGen: An Integrated Vision and Language Model for the Generation of Clinically Accurate Chest X-Ray Image-Report Pairs

At WACV 2024, Junjie Shentu and Noura Al Moubayed propose a model designed specifically for generating chest X-Ray image-report pairs.

Healthcare Synthetic data CXR Multimodality
v
Publication Journal article

Reproducibility and explainability in digital humanities

Fabian Offert joins Thorsten Ries and Karina van Dalen-Oskam to edit a special issue of the International Journal of Digital Humanities, introduced in this editorial

Digital humanities Reproducibility Explainability Methodology
v
event Book presentation

Pro qm: The Eye of the Master

Book Presentation with Matteo Pasquinelli and Henning Schmidgen

Automation Intelligence Labour Social relations
v
Publication Conference paper

Length is a Curse and a Blessing for Document-level Semantics

At EMNLP 2023, Chenghao Xiao, Yizhi Li, G Hudson, Chenghua Lin, and Noura Al Moubayed question the length generalizability of contrastive learning-based models and introduce LA(SER)3 document representation learning framework.

Contrastive learning Semantic shift Language models Isotropy
v
Event Round-table talk

Algorithmic Cultures

Arif Kornweitz presented a paper in response to the invitation to discuss ‘how algorithms actively shape and are shaped by cultural norms, values, institutions, and human behaviors’.

AI Interpretability and Accountability in the Humanitarian Sector
Algorithmic cultures Algorithms Sociotechnical Behaviour
v
Publication Book chapter

KI-basierte Verfahren in der bildenden Kunst

Fabian Offert contributes a chapter on AI in the visual arts to the German-language Handbook of Artificial Intelligence and the Arts

Creativity Art Computer art
v
Event Conference panel

Artificial governance

Fabian Offert speaks on a panel at the ‘Algorithmic Societies’ conference

v
Event Panel

The politics and aesthetics of synthetic media

Fabian Offert co-organises a panel on synthetic media at 4S 2023

Interpretability and Accountability of AI Systems in the Sciences
Synthetic media Synthetic data STS
v
event panel

Theories of Automation Before and After Artificial Intelligence

Members of the AI Forensics project take part in a panel investigating the spectrum of automation theories and technological transformation, focussing on AI as the last stage of the automation of manual, mental, and care labour.

panel conference historical materialism automation
v
event Conference talk

The Labour Theory of Automation from Babbage to AI

Matteo Pasquinelli grounds the rise of automation and AI in the development of metrics of labour and social behaviours

Historical Materialism psychometrics history of science and technology
v
event Conference talk

The Industrial Automation of Aesthetics in Machine Vision

Leonardo Impett demonstrated how current AI visual systems for the recognition and generation of images are grounded on specific and arbitrary measurement of beauty and how old AI visual systems keep on reproducing these standards and conventions of ‘beauty’ through their implementation in many applications.

Scopic regimes of neural wiring
measurement beauty
v
event Conference talk

Machine Translation as a Technology of Bordering

Paolo Caffoni discusses the indexing of language’s value in Machine Translation

Pedagogy of Machines
Historical Materialism panel conference language machine translation
v
Event Workshop

Bayesian Knowledge: Situated and Pluriversal Perspectives

Goda Klumbytė co-organises a workshop on Bayesian epistemology

AI Design Interventions for Social Diversity
Bayesian statistics Pluriversal knowledges
v
event Book presentation

The Eye of the Master: A Social History of Artificial Intelligence (Verso, 2023)

A book presentation with Matteo Pasquinelli at the University of Cambridge

Automation Labour
v
Publication Magazine article

AI en de regels van de kunst

Arif Kornweitz discusses generative AI in an art magazine

Art Aesthetics Visual culture Cultural production
v
Publication Preprint

CoinRun: Solving Goal Misgeneralisation

Goal generalisation is shown to be a solvable by way of the ACE concept extrapolation algorithm on CoinRun

Latent mechanistic interpretability
Goal generalisation Misgeneralisation ACE CoinRun Alignment
v
Event Conference talk

Embedded Knowledge as Computational Technique and Epistemic Paradigm

Fabian Offert speaks at the “Alien Knowledges” panel at the 2023 Society for Literature, Science & the Arts conference.

Computation
v
Publication Workshop paper

Feminist epistemology for machine learning systems design

A series of feminist epistemological concepts for developing critical, more accountable, and contextualised approaches to ML systems design.

AI Design Interventions for Social Diversity
ML systems design Feminist epistemologies Situated knowledges Critical fabulation
v
publication book

The Eye of the Master: A Social History of Artificial Intelligence

Matteo Pasquinelli’s new book is published on Verso.

book ai
v
Publication Essay

The Sound of Multidimensional Space: How Avant-Garde Music Foreshadowed Artificial Intelligence

Matteo Pasquinelli and Arif Kornweitz contribute an essay to the catalogue of the Music Biennale in Venice.

Music Multidimensional space Avant garde music Aesthetics Cultural production Latent space
v
Publication Book chapter

The Meaning Trap [Protokoll 13]

Fabian Offert and Wolf Kittler engage with ChatGPT.

Meaning LLMs ChatGPT
v
Event Lecture

What is a language model? The contribution of semiotics to the studies of AI and automation

Matteo Pasquinelli mobilises interpretative semiotics to illuminate the invention of algorithmic models.

Semiotics LLMs Automation
v
Event Conference talk

From Mechanical Thinking to Models of Mind: Notes Towards a Historical Epistemology of Artificial Intelligence

What model of knowledge is the current form of AI (i.e. machine learning) representing? In the history of human civilizations tools have always emerged together with a system of (technical) knowledge associated to them, but this aspect seems very confused in an artefact that is said to directly automate human „intelligence“ and „learning.“ This epistemological dimension, that is the distinction between knowledge and tool of course exist also in machine learning as the distinction, for example, between programming languages and application, but it seems to be continuously removed from the debate on AI that is fixated on an equation unique in the history of epistemology: machine = intelligence. To criticize this assumption this keynote reads the idea of „machine intelligence“ not as a novelty but as the latest stage of the history of algorithmic thinking, and this as the confluence of the longer history of mechanical thinking with statistical thinking. Whereas the epistemology of mechanical thinking and statistical thinking has been flourishing, the epistemology on machine learning is still fragmentary and the keynote attempts an overview of the different epistemological schools and methods that could help consolidate this field.

Historical Epistemology
v
Publication Journal article

Predicting acute clinical deterioration with interpretable machine learning to support emergency care decision making

Noura Al Moubayed joins Stelios Boulitsakis Logothetis et al. systematically compare the performance of ML algorithms for predicting patients’ imminent clinical deterioration.

Healthcare Emergency care Interpretable ML
v
Publication Preprint

There Is a Digital Art History

Leonardo Impett and Fabian Offert revisit Johanna Drucker’s question in the light of transformer-based vision models.

Art history Transformers Digital art
v
Publication Conference paper

On Isotropy, Contextualization and Learning Dynamics of Contrastive-based Sentence Representation Learning

Chenghao Xiao, Yang Long, and Noura Al Moubayed look at contrastive SRL through the lens of isotropy, contextualization and learning dynamics at ACL 2023.

Contrastive learning NLP SRL Semantics Language models Isotropy Contextualization
v
Event Panel

Extracting Humanness, Exploiting Labour: The Inhumane Face of Artificial Intelligence

Matteo Pasquinelli chairs a panel at the STS Italia Conference

STS Conference Panel
v
Publication Conference paper

Natural Language Explanations for Machine Learning Classification Decisions

At IJCNN 2023, James Burton, Noura Al Moubayed, and Amir Enshaei address the challenge of providing understandable explanations for machine learning classification decisions

Explanation Natural language explanation Classification
v
Publication Journal article

A Computational Approach to Hand Pose Recognition in Early Modern Paintings

Valentine Bernasconi, Eva Cetinić, and Leonardo Impett present a computational approach to hand pose recognition in paintings.

Scopic regimes of neural wiring
Hand pose estimation Hand gestures Digital art history Digital humanities Visual culture
v
event talk

The Labour of Energy and Information: Notes for a Historical Metrology of the Anthropocene

Matteo Pasquinelli gives a talk at the Historical Epistemologies of Planetary Modelling workshop at the Max Planck Institute for the History of Science in Berlin

Epistemology Modelling
v
Event Workshop

Workshop in political epistemology: ecosocialism and the general intellect

The KIM Group takes part in a workshop at the University of Venice discussing the relevance of the general intellect and, more specifically, the Spinozist idea of the social cortex to contemporary AI.

General intellect Social brain Socialist cortex Extended cognition
v
Event Colloquium

Semiotics and Large Language Models

The KIM convened in a special colloquium to open a new chapter on interpretability research mobilising semiotics in light of large language models

Semiotics LLMs interpretability structuralism
v
Publication Journal article

Can We Read Neural Networks? Epistemic Implications of Two Historical Computer Science Papers

Fabian Offert reviews two historical papers from the machine learning literature

Epistemology Humanities Computation
v
Event Workshop

Zürich Digital Visual Studies Workshop

An informal workshop with scholars from Cambridge and Zürich

Scopic regimes of neural wiring
Art history Visual studies
v
Publication Journal article

On the Concept of History (in Foundation Models)

Fabian Offert interrogates models such as CLIP and DALL-E 2 for their concept of history

History Foundation models Modelling Visual culture
v
Publication Workshop paper

Explaining the ghosts: Feminist intersectional XAI and cartography as methods to account for invisible labour

Klumbytė, Piehl, and Draude propose cartography as an XAI method to sensitise users to the role of invisible labour.

AI Design Interventions for Social Diversity
HCI XAI Feminism Intersectionality Cartography Labour Invisible labour
v
Publication Conference paper

Towards Feminist Intersectional XAI: From Explainability to Response-Ability

Goda Klumbytė, Hannah Piehl, and Claude Draude present a paper at CHI ‘23

AI Design Interventions for Social Diversity
Feminism Intersectionality XAI
v
Class Seminar

The work of translation after ChatGPT and Large Language Models

Paolo Caffoni and Matteo Pasquinelli co-teach a seminar on machine translation in the KIM at HfG Karlsruhe

Pedagogy of Machines
translation machine translation LLMs
v
event Conference talk

AI and the Concept of the Work of Art

Arif Kornweitz contributes to the symposium From Hype to Reality: Artificial Intelligence in the Study of Art and Culture

Art Culture Aesthetics Cultural production Epistemology
v
event Talk

What Are Large Visual Models Models Of?

How is (visual) culture modelled?

Visual models Modelling history Models of_
v
event Talk

A History of Machine Visuality

Leonardo Impett introduces a new history of machine visuality

Scopic regimes of neural wiring
Visuality Machine vision history
v
Event Colloquium

KIM Colloquium Summer 2023 Symposium

  • Arif Kornweitz

  • Mariana Silva

  • Paolo Caffoni

  • Alan Diaz

  • Patrick Riechert

  • George MacBeth

AI Interpretability and Accountability in the Humanitarian Sector
KIM Humanitarian
v
Event Lecture & Workshop

AI, biometrics, and making art in the age of surveillance

Adam Harvey gives a lecture and runs a workshop at Carnegie Mellon University

Exposing the production pipeline of facial recognition systems
AI Art Surveillance Biometrics
v
Event Lecture

Computer vision, surveillance, and privacy

Adam Harvey gives a lecture at the University of Michigan Ann Arbor

Exposing the production pipeline of facial recognition systems
Privacy Surveillance Computer vision
v
PR Podcast

Latent Deep Space

Fabian Offert discusses Generative Models, Synthetic Data, the role of Latent Spaces and the future of General Computation from Artificial Intelligence.

Interpretability and Accountability of AI Systems in the Sciences
Generative models Synthetic data Latent space Computation General computation Protein folding
v
Publication Conference paper

Learning How to MIMIC: Using Model Explanations To Guide Deep Learning Training

Matthew Watson, Bashar Awwad Shiekh Hasan, and Noura Al Moubayed propose a novel model architecture that utilises model explanations during training.

Healthcare Explanations Training
v
Publication Conference paper

A Priority Map for Vision-and-Language Navigation with Trajectory Plans and Feature-Location Cues

Jason Armitage, Leonardo Impett, and Rico Sennrich take inspiration from neuropsychological studies and attain state-of-the-art performance in VLN

Scopic regimes of neural wiring
VLN Priority maps Trajectory planning Cross-modal alignment
v
Post Blog

On the Emergence of General Computation from Artificial Intelligence

Fabian Offert identifies a paradigm change towards AI-based general computation

General computation Computation
v
event lecture

Beyond Quantity: Research with Subsymbolic AI

Prof. Matteo Pasquinelli delivered a keynote address on the impact of AI on scientific research.

lecture keynote conference
v
event presentation

Public presentation of the AI Forensics project

The KIM research group inaugurated and presented the project AI Forensics.

presentation panel
v
Cultural production

The use of AI in cultural production, its aesthetics, and its epistemology.

General computation

What is the relationship between AI and general computation? Will the latter emerge out of the former?

Mechanistic interpretability

Interpreting neural networks on the basis of their parameters.

Metrics and Metrology

Measuring and the study of measurement.

Models of_

What are foundation models in fact models of? Do LLMs really model language? What is the object of a model of visual culture?

Pedagogy

What kind of learning comes after machine learning?

Pluriversal epistemologies

Feminist epistemologies of situated knowledge and Bayesian statistics present avenues for a non-totalising approach to interpretability of automation vis-à-vis knowledge/s.

Political economy

… of datasets, training labour, automation…

Semiotic interpretability

Contemporary LLMs and their effective “solving” of NLP in fact highlights the need to develop a deeper formulation of ‘interpretability’ on the basis of semiotics.

Structuralism

Contemporary AI systems and their embedding spaces seem to revitalise structuralism in a new, significantly more complex form.

Synthetic bootstrapping

How does training algorithms on synthetic data affect their interpretability?

Vectors

AI Forensics © 2025

Contact | Imprint