Tobias Blanke
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Tobias Blanke has a total of 96 co-authors. Here are the top co-authors by number of shared publications:
- Mark Hedges
King's College London, London, UK
25 shared publications - Claudia Aradau
King's College London
9 shared publications - Michael Bryant
King’s College London
7 shared publications - Stuart Dunn
Department of Digital Humanities, King’s College London, UK
6 shared publications Andreas Aschenbrenner
State and University Library Göttingen, Göttingen, Germany
4 shared publicationsAdil Hasan
Liverpool University
3 shared publications
Latest Publications
- Open Access
- 22 September 2026
- Published by SAGE Publications in Big Data & Society
- Vol. 13 (3)
An archive of contestation: Digital surveillance and human–machine recompositions of resistance
This article proposes to make sense of the ‘atmosphere of failure’ diagnosed by key players in the mobilisation against mass surveillance in the United Kingdom. Firstly, by entangling oral history interviews and computational analyses of NGO websites, we recompose an archive of contestation that is not temporally limited to the Snowden ‘moment’. We advance the notion of a ‘discordant machine’ to render the generativity of discordances and surprises in entangling qualitative and digital methods. Secondly, we argue that the campaigns against surveillance did not lead to a shared vocabulary of critique and a collective practice of contestation. We show that contesting digital state surveillance was mostly expressed through the conduits of privacy and transparency, while more collective democratic claims were connected instead to ‘elsewheres’, such as democratisation in Global South states. Although the Snowden disclosures were a key moment for academic scholarship, our analysis shows that they played little role in mobilisations against mass surveillance. References to oversight were limited to the immediate post-Snowden moment. Finally, corporate surveillance remained a feature of NGO advocacy but was represented mainly at a distance from state surveillance. Our analysis suggests that the challenge for mobilisations against digital surveillance is not only to hold together its complexities but also to articulate its antidemocratic effects.
- 22 September 2026
- Published by SAGE Publications in Big Data & Society
- Vol. 13 (3)
Open AccessAn archive of contestation: Digital surveillance and human–machine recompositions of resistance
This article proposes to make sense of the ‘atmosphere of failure’ diagnosed by key players in the mobilisation against mass surveillance in the United Kingdom. Firstly, by entangling oral history interviews and computational analyses of NGO websites, we recompose an archive of contestation that is not temporally limited to the Snowden ‘moment’. We advance the notion of a ‘discordant machine’ to render the generativity of discordances and surprises in entangling qualitative and digital methods. Secondly, we argue that the campaigns against surveillance did not lead to a shared vocabulary of critique and a collective practice of contestation. We show that contesting digital state surveillance was mostly expressed through the conduits of privacy and transparency, while more collective democratic claims were connected instead to ‘elsewheres’, such as democratisation in Global South states. Although the Snowden disclosures were a key moment for academic scholarship, our analysis shows that they played little role in mobilisations against mass surveillance. References to oversight were limited to the immediate post-Snowden moment. Finally, corporate surveillance remained a feature of NGO advocacy but was represented mainly at a distance from state surveillance. Our analysis suggests that the challenge for mobilisations against digital surveillance is not only to hold together its complexities but also to articulate its antidemocratic effects.
- Open Access
- 16 September 2026
- Published by Springer Nature in International Journal of Digital Humanities
- p. 1-26
Countering silences in the EHRI collection: a graph neural network approach for archival justice
Searching for archival voids is a central concept in critical archival research, yet traditional approaches remain largely restricted by what is present and legible: Natural Language Processing and text-based methods tend to operate at the level of the document, struggling to engage with what is structurally absent. Operationalizing the archival principle of respect des fonds, instead, demands methods that preserve the structure and hierarchy of a collection as a whole. In archival collections of the Holocaust, silences take many forms: evidence destroyed by perpetrators, undocumented stories, records lost to dispersal or rendered invisible by digital archiving practices. Though computational approaches cannot tend to all voids, we aim to support the interpretive work of archival justice research by exploring ways for reading absence itself as a form of evidence and recovering fragmented information that might escape conventional approaches with the help of provenance and relational context. To this end, we introduce a Graph Neural Network pipeline for the critical engagement with collections from the European Holocaust Research Infrastructure. Our methodology treats documents not as isolated nodes but as entities embedded in relational structures of institutional provenance, hierarchy, and semantic content. We present a series of investigations demonstrating how our GNN approach can detect anomalies, discontinuities, and structural gaps that signal archival silences, offering a transferable framework for a critical interrogation of archival voids and the reconstruction of dispersed records in historical collections.
- 16 September 2026
- Published by Springer Nature in International Journal of Digital Humanities
- p. 1-26
Open AccessCountering silences in the EHRI collection: a graph neural network approach for archival justice
Searching for archival voids is a central concept in critical archival research, yet traditional approaches remain largely restricted by what is present and legible: Natural Language Processing and text-based methods tend to operate at the level of the document, struggling to engage with what is structurally absent. Operationalizing the archival principle of respect des fonds, instead, demands methods that preserve the structure and hierarchy of a collection as a whole. In archival collections of the Holocaust, silences take many forms: evidence destroyed by perpetrators, undocumented stories, records lost to dispersal or rendered invisible by digital archiving practices. Though computational approaches cannot tend to all voids, we aim to support the interpretive work of archival justice research by exploring ways for reading absence itself as a form of evidence and recovering fragmented information that might escape conventional approaches with the help of provenance and relational context. To this end, we introduce a Graph Neural Network pipeline for the critical engagement with collections from the European Holocaust Research Infrastructure. Our methodology treats documents not as isolated nodes but as entities embedded in relational structures of institutional provenance, hierarchy, and semantic content. We present a series of investigations demonstrating how our GNN approach can detect anomalies, discontinuities, and structural gaps that signal archival silences, offering a transferable framework for a critical interrogation of archival voids and the reconstruction of dispersed records in historical collections.
- Open Access
- 7 August 2026
- Published by Springer Nature in AI & SOCIETY
- Vol. 41 (10) , 10191-10204
Zones of (un)certainty: studying the temporal stabilization of ground truths for medical AI
Medical data annotation is a critical yet contingent process that (re)configures ambiguous medical decisions and data representations into ground truths for artificial intelligence (AI). Extending critical data, AI, and STS research on uncertainty in data annotation, this article develops the concept of zones of (un)certainty to capture the epistemic time-spaces through which annotators transform uncertainty and dissonance into stabilized ground-truth data for medical AI. Drawing on interviews with AI and medical researchers at two Dutch University Medical Centres and complemented by observations at an interdisciplinary workshop on annotation practices, we empirically unfold three such zones : first, we identify an epistemic time–space where ground-truthing is approached spatially as practitioners decompose and recompose data layers, views, and metadata to reach “good enough” certainty. Second, we analyze temporal zones of drift where micro-judgments are routed through negotiations with colleagues, experts, AI suggestions, and automated evaluations. Third, we examine thresholding practices in which uncertain cases are gradually compressed into singular labels. Taken together, this article provides insights into the temporal–spatial dimensions and thresholds that annotators use to settle on ground-truth decisions. Consequently, we open a broader analytical vocabulary for studying how uncertainty and dissonance are operationalized for AI-based medicine and beyond.
- 7 August 2026
- Published by Springer Nature in AI & SOCIETY
- Vol. 41 (10) , 10191-10204
Open AccessZones of (un)certainty: studying the temporal stabilization of ground truths for medical AI
Medical data annotation is a critical yet contingent process that (re)configures ambiguous medical decisions and data representations into ground truths for artificial intelligence (AI). Extending critical data, AI, and STS research on uncertainty in data annotation, this article develops the concept of zones of (un)certainty to capture the epistemic time-spaces through which annotators transform uncertainty and dissonance into stabilized ground-truth data for medical AI. Drawing on interviews with AI and medical researchers at two Dutch University Medical Centres and complemented by observations at an interdisciplinary workshop on annotation practices, we empirically unfold three such zones : first, we identify an epistemic time–space where ground-truthing is approached spatially as practitioners decompose and recompose data layers, views, and metadata to reach “good enough” certainty. Second, we analyze temporal zones of drift where micro-judgments are routed through negotiations with colleagues, experts, AI suggestions, and automated evaluations. Third, we examine thresholding practices in which uncertain cases are gradually compressed into singular labels. Taken together, this article provides insights into the temporal–spatial dimensions and thresholds that annotators use to settle on ground-truth decisions. Consequently, we open a broader analytical vocabulary for studying how uncertainty and dissonance are operationalized for AI-based medicine and beyond.
- Open Access
- 20 May 2026
- Published by SAGE Publications in Big Data & Society
- Vol. 13 (2)
Making GenAI valuable: Benchmarks, singularities, and the enrichment economy
As investment in Generative AI (GenAI) has reached tens and hundreds of billions, anxieties about an ‘AI bubble’ have been on the rise. This article offers an alternative perspective on how GenAI is made valuable. Drawing on Luc Boltanski and Arnaud Esquerre's economy of enrichment, it argues that benchmarks have become devices of GenAI valuation, which make Large Language Models (LLMs) into singular, exceptional and non-standard objects. Benchmarks enrich LLMs by mobilising epistemic cultures of science and narratives of a future perfect of Artificial General Intelligence (AGI). To develop this argument, we first show that the perspective of enrichment can supplement the frameworks of surveillance capitalism, platform capitalism, and assetisation by accounting for the centrality of benchmarks as devices that integrate models within a collection of elite models and simultaneously differentiate them as singularities. Second, instead of cultures of the past, GenAI narratives draw on epistemic cultures of science. With GenAI, the creation of new benchmarks has become a commercial pursuit, going beyond computer science and relying on various benchmarks from the ‘psy’ sciences to address the demands of hyper-scale commercial GenAI. Third, we unpack how narratives of saturation, surpassing, and emergence singularise models by situating them in proximity to a future perfect of AGI. In the conclusion, we offer reflections on the potential of the enrichment economy for critical engagement with GenAI.
- 20 May 2026
- Published by SAGE Publications in Big Data & Society
- Vol. 13 (2)
Open AccessMaking GenAI valuable: Benchmarks, singularities, and the enrichment economy
As investment in Generative AI (GenAI) has reached tens and hundreds of billions, anxieties about an ‘AI bubble’ have been on the rise. This article offers an alternative perspective on how GenAI is made valuable. Drawing on Luc Boltanski and Arnaud Esquerre's economy of enrichment, it argues that benchmarks have become devices of GenAI valuation, which make Large Language Models (LLMs) into singular, exceptional and non-standard objects. Benchmarks enrich LLMs by mobilising epistemic cultures of science and narratives of a future perfect of Artificial General Intelligence (AGI). To develop this argument, we first show that the perspective of enrichment can supplement the frameworks of surveillance capitalism, platform capitalism, and assetisation by accounting for the centrality of benchmarks as devices that integrate models within a collection of elite models and simultaneously differentiate them as singularities. Second, instead of cultures of the past, GenAI narratives draw on epistemic cultures of science. With GenAI, the creation of new benchmarks has become a commercial pursuit, going beyond computer science and relying on various benchmarks from the ‘psy’ sciences to address the demands of hyper-scale commercial GenAI. Third, we unpack how narratives of saturation, surpassing, and emergence singularise models by situating them in proximity to a future perfect of AGI. In the conclusion, we offer reflections on the potential of the enrichment economy for critical engagement with GenAI.
- Open Access
- 15 April 2026
- Published by SAGE Publications in Big Data & Society
- Vol. 13 (2)
Making machine learning good enough – studying the political endeavour of finding ‘right’ metrics and thresholds
As machine learning (ML) technologies move from their discrete existence in research to being highly applied technologies across society, critical scholars have begun to address the epistemological conditions that shape the emergence of such systems and their societal implications. In this paper, we investigate a specific epistemological condition of ML, namely, how ML systems rely on ongoing negotiations and agreements of ‘good enough’ to be deployed. We do so by drawing on ethnographic fieldwork with the British Broadcasting Corporation (BBC) – a large data- and value-driven organisation. In studying the epistemological function and politics of ‘good enough’, we take an AI lab studies approach, following the Recommendations Team's efforts to materialise ‘good enoughness’ and make it negotiable as they develop and modify recommender systems that aim to better serve the BBC's audiences. Through our ethnographic account, we demonstrate how the team relies on various metrics and qualitative evaluations to inform provisional performance thresholds before submitting the ML systems to A/B testing to establish whether one of them is ‘good enough’ to deploy. By following these processes of establishing ‘good enough’, we see how these negotiations are entangled in various, sometimes competing organisational objectives, as well as particular data and technical infrastructures. By extension, we show how the metricised performance scores of AB testing are negotiated in practice by readjusting performance thresholds to manoeuvre different values and constraints. Ultimately, our paper shows that establishing ‘good enough’ is a political endeavour of adjusting seemingly objective evaluation criteria to find the best-fitting metrics and ‘right’ thresholds.
- 15 April 2026
- Published by SAGE Publications in Big Data & Society
- Vol. 13 (2)
Open AccessMaking machine learning good enough – studying the political endeavour of finding ‘right’ metrics and thresholds
As machine learning (ML) technologies move from their discrete existence in research to being highly applied technologies across society, critical scholars have begun to address the epistemological conditions that shape the emergence of such systems and their societal implications. In this paper, we investigate a specific epistemological condition of ML, namely, how ML systems rely on ongoing negotiations and agreements of ‘good enough’ to be deployed. We do so by drawing on ethnographic fieldwork with the British Broadcasting Corporation (BBC) – a large data- and value-driven organisation. In studying the epistemological function and politics of ‘good enough’, we take an AI lab studies approach, following the Recommendations Team's efforts to materialise ‘good enoughness’ and make it negotiable as they develop and modify recommender systems that aim to better serve the BBC's audiences. Through our ethnographic account, we demonstrate how the team relies on various metrics and qualitative evaluations to inform provisional performance thresholds before submitting the ML systems to A/B testing to establish whether one of them is ‘good enough’ to deploy. By following these processes of establishing ‘good enough’, we see how these negotiations are entangled in various, sometimes competing organisational objectives, as well as particular data and technical infrastructures. By extension, we show how the metricised performance scores of AB testing are negotiated in practice by readjusting performance thresholds to manoeuvre different values and constraints. Ultimately, our paper shows that establishing ‘good enough’ is a political endeavour of adjusting seemingly objective evaluation criteria to find the best-fitting metrics and ‘right’ thresholds.
- Open Access
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2470-2479
Modeling Strategic Risk in School Choice: A Case for Transparent Design
School choice algorithms, which assign students to schools based on preferences, are often designed to balance matching efficiency with strategyproofness. In Amsterdam, policymakers introduced a more efficient assignment mechanism in 2019 that permitted preference manipulation. As strategic behavior increased, transparency around the algorithm was reduced—motivated by prior studies showing that additional information about matching methods tends to increase manipulation. Yet transparency is also essential for institutional trust and informed decision-making. We test whether transparency about the consequences of strategic behavior deters or encourages manipulation. We conduct a behavioral experiment embedded in a survey of 140 parents participating in Amsterdam’s 2025 school choice process. Parents selected among truthful, mildly strategic, and aggressively strategic rankings across three scenarios with varying degrees of probability disclosure. While transparency increased the rate of strategic behavior, truthful reporting remained the most common choice under all conditions. To explain these patterns, we estimate a discrete-choice model grounded in Security-Potential/Aspiration (SP/A) theory, which models how parents weigh aspirational goals against the risk of poor outcomes. The model fits observed behavior better than expected utility and prospect theory benchmarks, even though individual covariates are not statistically significant. We find that parents whose children applied to competitive school tracks (e.g., pre-university) were more risk-averse, selecting safer strategies even when aggressive options offered high chances of success. These results challenge the assumption that transparency inherently worsens inequality, and suggest that communicating risk may serve as a viable policy lever for fairness in algorithmic systems.
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2470-2479
Open AccessModeling Strategic Risk in School Choice: A Case for Transparent Design
School choice algorithms, which assign students to schools based on preferences, are often designed to balance matching efficiency with strategyproofness. In Amsterdam, policymakers introduced a more efficient assignment mechanism in 2019 that permitted preference manipulation. As strategic behavior increased, transparency around the algorithm was reduced—motivated by prior studies showing that additional information about matching methods tends to increase manipulation. Yet transparency is also essential for institutional trust and informed decision-making. We test whether transparency about the consequences of strategic behavior deters or encourages manipulation. We conduct a behavioral experiment embedded in a survey of 140 parents participating in Amsterdam’s 2025 school choice process. Parents selected among truthful, mildly strategic, and aggressively strategic rankings across three scenarios with varying degrees of probability disclosure. While transparency increased the rate of strategic behavior, truthful reporting remained the most common choice under all conditions. To explain these patterns, we estimate a discrete-choice model grounded in Security-Potential/Aspiration (SP/A) theory, which models how parents weigh aspirational goals against the risk of poor outcomes. The model fits observed behavior better than expected utility and prospect theory benchmarks, even though individual covariates are not statistically significant. We find that parents whose children applied to competitive school tracks (e.g., pre-university) were more risk-averse, selecting safer strategies even when aggressive options offered high chances of success. These results challenge the assumption that transparency inherently worsens inequality, and suggest that communicating risk may serve as a viable policy lever for fairness in algorithmic systems.
- 4 May 2024
- Published by Elsevier in Political Geography
- Vol. 113
Review forum
- Open Access
- 29 March 2024
- Published by Taylor & Francis in Information, Communication & Society
- Vol. 28 (10) , 1735-1756
AI competitions as infrastructures of power in medical imaging
Medical imaging is a vital research field for artificial intelligence (AI1), particularly since CTs or MRIs provide vast amounts of valuable health data (Prevedello et al., 2019). To stimulate deve...
- 29 March 2024
- Published by Taylor & Francis in Information, Communication & Society
- Vol. 28 (10) , 1735-1756
Open AccessAI competitions as infrastructures of power in medical imaging
Medical imaging is a vital research field for artificial intelligence (AI1), particularly since CTs or MRIs provide vast amounts of valuable health data (Prevedello et al., 2019). To stimulate deve...
- Open Access
- 28 February 2024
- Published by Oxford University Press (OUP) in Digital Scholarship in the Humanities
- Vol. 39 (2) , 485-499
Using deep learning to analyse the times of the UN Security Council
This article analyses how digital humanities scholarship can make use of recent advances in deep learning to analyse the temporal relations in an online textual archive. We use transfer learning as well as data augmentation techniques to investigate changes in United Nations Security Council resolutions. Instead of pre-defined periods, as it is common, we target the years directly. Such a text regression task is novel in the digital humanities as far as we can see and has the advantage of speaking directly to historical relations. We present not only very good experimental results but also demonstrate how such text regressions can be interpreted directly and with surrogate topic models.
- 28 February 2024
- Published by Oxford University Press (OUP) in Digital Scholarship in the Humanities
- Vol. 39 (2) , 485-499
Open AccessUsing deep learning to analyse the times of the UN Security Council
This article analyses how digital humanities scholarship can make use of recent advances in deep learning to analyse the temporal relations in an online textual archive. We use transfer learning as well as data augmentation techniques to investigate changes in United Nations Security Council resolutions. Instead of pre-defined periods, as it is common, we target the years directly. Such a text regression task is novel in the digital humanities as far as we can see and has the advantage of speaking directly to historical relations. We present not only very good experimental results but also demonstrate how such text regressions can be interpreted directly and with surrogate topic models.
- Open Access
- 2 February 2024
- Published by Springer Nature in Humanities and Social Sciences Communications
- Vol. 11 (1) , 1-12
Reassembling digital archives—strategies for counter-archiving
Archives have long been a key concern of academic debates about truth, memory, recording and power and are important sites for social sciences and humanities research. This has been the case for traditional archives, but these debates have accelerated with the digital transformation of archives. The proliferation of digital tools and the fast-growing increase in digital materials have created very large digitised and born-digital archives. This article investigates how new digital archives continue existing archival practices while at the same time discontinuing them. We present novel methodologies and tools for changing memory and power relations in digital archives through new ways of reassembling marginalised, non-canonical entities in digital archives. Reassembling digital archives can take advantage of the materiality and the algorithmic processuality of digital collections and reshape them to inscribe lost voices and previously ignored differences. Digital archives are not fixed and are changed with new research and political questions and are only identified through new questions. The article presents six distinct techniques and strategies to reassemble digital archives and renders these according to three different types of new digital archives. We consider both the extension of archives towards evidence that is otherwise thrown away as well as the provision of new intensive, non-discriminatory viewpoints on existing collections.
- 2 February 2024
- Published by Springer Nature in Humanities and Social Sciences Communications
- Vol. 11 (1) , 1-12
Open AccessReassembling digital archives—strategies for counter-archiving
Archives have long been a key concern of academic debates about truth, memory, recording and power and are important sites for social sciences and humanities research. This has been the case for traditional archives, but these debates have accelerated with the digital transformation of archives. The proliferation of digital tools and the fast-growing increase in digital materials have created very large digitised and born-digital archives. This article investigates how new digital archives continue existing archival practices while at the same time discontinuing them. We present novel methodologies and tools for changing memory and power relations in digital archives through new ways of reassembling marginalised, non-canonical entities in digital archives. Reassembling digital archives can take advantage of the materiality and the algorithmic processuality of digital collections and reshape them to inscribe lost voices and previously ignored differences. Digital archives are not fixed and are changed with new research and political questions and are only identified through new questions. The article presents six distinct techniques and strategies to reassemble digital archives and renders these according to three different types of new digital archives. We consider both the extension of archives towards evidence that is otherwise thrown away as well as the provision of new intensive, non-discriminatory viewpoints on existing collections.