Fabian Offert
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Fabian Offert has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Jennifer Jacobs
University of California Santa Barbara, Santa Barbara, USA
1 shared publication - Kai Massey Thaler
University of California, Santa Barbara, Santa Barbara, California, USA
1 shared publication Paul Kim
University of California, Santa Barbara, Santa Barbara, USA
1 shared publication- Payton Croskey
University of California Santa Barbara, Santa Barbara, USA
1 shared publication - Peter Bell
Institut für Kunstgeschichte Philipps-Universität Marburg Germany
1 shared publication Qiaoyu Cai
University of California, Santa Barbara, Santa Barbara, USA
1 shared publication
Latest Publications
- Open Access
- 19 February 2026
- Published by Springer Nature in AI & SOCIETY
- Vol. 41 (6) , 5467-5479
Synthesizing proteins on the graphics card: protein folding and the limits of critical AI studies
This paper investigates the application of the transformer architecture in protein folding, as exemplified by DeepMind’s AlphaFold project, and its implications for the understanding of so-called large language models. The prevailing discourse often assumes a ready-made analogy between proteins—encoded as sequences of amino acids—and natural language, which we term the language paradigm of computational (structural) biology. Instead of assuming this analogy as given, we critically evaluate it to assess the kind of knowledge-making afforded by the transformer architecture. We first trace the analogy’s emergence and historical development, carving out the influence of structural linguistics on structural biology beginning in the mid-twentieth century. We then examine three often overlooked preprocessing steps essential to the transformer architecture, including subword tokenization, word embedding, and positional encoding, to demonstrate its regime of representation based on continuous, high-dimensional vector spaces, which departs from the discrete nature of language. The successful deployment of transformers in protein folding, we argue, discloses what we consider a non-linguistic approach to token processing intrinsic to the architecture. We contend that through this non-linguistic processing, the transformer architecture carves out unique epistemological territory and produces a new class of knowledge, distinct from established domains. Our search for intelligent machines thus has to begin with the shape , rather than the place , of intelligence. Consequently, the emerging field of critical AI studies should take methodological inspiration from the history of science in its quest to conceptualize the contributions of artificial intelligence to knowledge-making, within and beyond the domain-specific sciences.
- 19 February 2026
- Published by Springer Nature in AI & SOCIETY
- Vol. 41 (6) , 5467-5479
Open AccessSynthesizing proteins on the graphics card: protein folding and the limits of critical AI studies
This paper investigates the application of the transformer architecture in protein folding, as exemplified by DeepMind’s AlphaFold project, and its implications for the understanding of so-called large language models. The prevailing discourse often assumes a ready-made analogy between proteins—encoded as sequences of amino acids—and natural language, which we term the language paradigm of computational (structural) biology. Instead of assuming this analogy as given, we critically evaluate it to assess the kind of knowledge-making afforded by the transformer architecture. We first trace the analogy’s emergence and historical development, carving out the influence of structural linguistics on structural biology beginning in the mid-twentieth century. We then examine three often overlooked preprocessing steps essential to the transformer architecture, including subword tokenization, word embedding, and positional encoding, to demonstrate its regime of representation based on continuous, high-dimensional vector spaces, which departs from the discrete nature of language. The successful deployment of transformers in protein folding, we argue, discloses what we consider a non-linguistic approach to token processing intrinsic to the architecture. We contend that through this non-linguistic processing, the transformer architecture carves out unique epistemological territory and produces a new class of knowledge, distinct from established domains. Our search for intelligent machines thus has to begin with the shape , rather than the place , of intelligence. Consequently, the emerging field of critical AI studies should take methodological inspiration from the history of science in its quest to conceptualize the contributions of artificial intelligence to knowledge-making, within and beyond the domain-specific sciences.
- Open Access
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 900-913
Liberatory Collections and Ethical AI: Reimagining AI Development from Black Community Archives and Datasets
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 900-913
Open AccessLiberatory Collections and Ethical AI: Reimagining AI Development from Black Community Archives and Datasets
- Open Access
- 31 December 2024
- Published by DIGSUM (Centre for Digital Social Research) in Journal of Digital Social Research
- Vol. 6 (4) , 49-59
A sign that spells
In this paper, we examine how generative artificial intelligence produces a new politics of visual culture. We focus on DALL·E and related machine learning models as an emergent approach to image-making that operates through the cultural technique of semantic compression. Semantic compression, we argue, is an inhuman and invisual technique, yet it is still caught in a paradox that is ironically all too human: the consistent reproduction of whiteness as a latent feature of dominant visual culture. We use Open AI’s failed efforts to “debias” their system as a critical opening to interrogate how DALL·E dissolves and reconstitutes politically and economically salient human concepts like race. This example vividly illustrates the stakes of the current moment of transformation, when so-called foundation models reconfigure the boundaries of visual culture and when “doing” anti-racism means deploying quick technical fixes to mitigate personal discomfort, or more importantly, potential commercial loss. We conclude by arguing that it simply does not suffice anymore to point out a lack – of data, of representation, of subjectivity – in machine learning systems when these systems are designed and understood to be complete representations of reality. The current shift towards foundation models, then, at the very least presents an opportunity to reflect on what is next, even if it is just a “new and better” kind of complicity.
- 31 December 2024
- Published by DIGSUM (Centre for Digital Social Research) in Journal of Digital Social Research
- Vol. 6 (4) , 49-59
Open AccessA sign that spells
In this paper, we examine how generative artificial intelligence produces a new politics of visual culture. We focus on DALL·E and related machine learning models as an emergent approach to image-making that operates through the cultural technique of semantic compression. Semantic compression, we argue, is an inhuman and invisual technique, yet it is still caught in a paradox that is ironically all too human: the consistent reproduction of whiteness as a latent feature of dominant visual culture. We use Open AI’s failed efforts to “debias” their system as a critical opening to interrogate how DALL·E dissolves and reconstitutes politically and economically salient human concepts like race. This example vividly illustrates the stakes of the current moment of transformation, when so-called foundation models reconfigure the boundaries of visual culture and when “doing” anti-racism means deploying quick technical fixes to mitigate personal discomfort, or more importantly, potential commercial loss. We conclude by arguing that it simply does not suffice anymore to point out a lack – of data, of representation, of subjectivity – in machine learning systems when these systems are designed and understood to be complete representations of reality. The current shift towards foundation models, then, at the very least presents an opportunity to reflect on what is next, even if it is just a “new and better” kind of complicity.
- Open Access
A Sign That Spells: DALL-E 2, Invisual Images and The Racial Politics of Feature Space
In this paper, we examine how generative machine learning systems produce a new politics of visual culture. We focus on DALL-E 2 and related models as an emergent approach to image-making that operates through the cultural techniques of feature extraction and semantic compression. These techniques, we argue, are inhuman, invisual, and opaque, yet are still caught in a paradox that is ironically all too human: the consistent reproduction of whiteness as a latent feature of dominant visual culture. We use Open AI's failed efforts to 'debias' their system as a critical opening to interrogate how systems like DALL-E 2 dissolve and reconstitute politically salient human concepts like race. This example vividly illustrates the stakes of this moment of transformation, when so-called foundation models reconfigure the boundaries of visual culture and when 'doing' anti-racism means deploying quick technical fixes to mitigate personal discomfort, or more importantly, potential commercial loss.
Open AccessA Sign That Spells: DALL-E 2, Invisual Images and The Racial Politics of Feature Space
In this paper, we examine how generative machine learning systems produce a new politics of visual culture. We focus on DALL-E 2 and related models as an emergent approach to image-making that operates through the cultural techniques of feature extraction and semantic compression. These techniques, we argue, are inhuman, invisual, and opaque, yet are still caught in a paradox that is ironically all too human: the consistent reproduction of whiteness as a latent feature of dominant visual culture. We use Open AI's failed efforts to 'debias' their system as a critical opening to interrogate how systems like DALL-E 2 dissolve and reconstitute politically salient human concepts like race. This example vividly illustrates the stakes of this moment of transformation, when so-called foundation models reconfigure the boundaries of visual culture and when 'doing' anti-racism means deploying quick technical fixes to mitigate personal discomfort, or more importantly, potential commercial loss.
- Open Access
- 12 October 2020
- Published by Springer Nature in AI & SOCIETY
- Vol. 36 (4) , 1133-1144
Perceptual bias and technical metapictures: critical machine vision as a humanities challenge
In many critical investigations of machine vision, the focus lies almost exclusively on dataset bias and on fixing datasets by introducing more and more diverse sets of images. We propose that machine vision systems are inherently biased not only because they rely on biased datasets but also because their perceptual topology, their specific way of representing the visual world, gives rise to a new class of bias that we call perceptual bias. Concretely, we define perceptual topology as the set of those inductive biases in machine vision systems that determine its capability to represent the visual world. Perceptual bias, then, describes the difference between the assumed “ways of seeing” of a machine vision system, our reasonable expectations regarding its way of representing the visual world, and its actual perceptual topology. We show how perceptual bias affects the interpretability of machine vision systems in particular, by means of a close reading of a visualization technique called “feature visualization”. We conclude that dataset bias and perceptual bias both need to be considered in the critical analysis of machine vision systems and propose to understand critical machine vision as an important transdisciplinary challenge, situated at the interface of computer science and visual studies/Bildwissenschaft.
- 12 October 2020
- Published by Springer Nature in AI & SOCIETY
- Vol. 36 (4) , 1133-1144
Open AccessPerceptual bias and technical metapictures: critical machine vision as a humanities challenge
In many critical investigations of machine vision, the focus lies almost exclusively on dataset bias and on fixing datasets by introducing more and more diverse sets of images. We propose that machine vision systems are inherently biased not only because they rely on biased datasets but also because their perceptual topology, their specific way of representing the visual world, gives rise to a new class of bias that we call perceptual bias. Concretely, we define perceptual topology as the set of those inductive biases in machine vision systems that determine its capability to represent the visual world. Perceptual bias, then, describes the difference between the assumed “ways of seeing” of a machine vision system, our reasonable expectations regarding its way of representing the visual world, and its actual perceptual topology. We show how perceptual bias affects the interpretability of machine vision systems in particular, by means of a close reading of a visualization technique called “feature visualization”. We conclude that dataset bias and perceptual bias both need to be considered in the critical analysis of machine vision systems and propose to understand critical machine vision as an important transdisciplinary challenge, situated at the interface of computer science and visual studies/Bildwissenschaft.
- 29 January 2019
- Published by Springer Nature
- p. 710-715
Images of Image Machines. Visual Interpretability in Computer Vision for Art
- 29 January 2019
- Published by Springer Nature
- p. 710-715
Images of Image Machines. Visual Interpretability in Computer Vision for Art
- Open Access
"I know it when I see it". Visualization and Intuitive Interpretability
Most research on the interpretability of machine learning systems focuses on the development of a more rigorous notion of interpretability. I suggest that a better understanding of the deficiencies of the intuitive notion of interpretability is needed as well. I show that visualization enables but also impedes intuitive interpretability, as it presupposes two levels of technical pre-interpretation: dimensionality reduction and regularization. Furthermore, I argue that the use of positive concepts to emulate the distributed semantic structure of machine learning models introduces a significant human bias into the model. As a consequence, I suggest that, if intuitive interpretability is needed, singular representations of internal model states should be avoided.
Open Access"I know it when I see it". Visualization and Intuitive Interpretability
Most research on the interpretability of machine learning systems focuses on the development of a more rigorous notion of interpretability. I suggest that a better understanding of the deficiencies of the intuitive notion of interpretability is needed as well. I show that visualization enables but also impedes intuitive interpretability, as it presupposes two levels of technical pre-interpretation: dimensionality reduction and regularization. Furthermore, I argue that the use of positive concepts to emulate the distributed semantic structure of machine learning models introduces a significant human bias into the model. As a consequence, I suggest that, if intuitive interpretability is needed, singular representations of internal model states should be avoided.