Clustering: Automated Order in the Social Sciences
International Workshop, organized by Anna Echterhölter and Markus Ramsauer in cooperation with the working group “How is AI Changing Science”, University of Vienna, and the International Research Center for Cultural Studies (ifk).
Date: November 28-29, 2024
Venue: International Research Center for Cultural Studies (Internationales Forschungszentrum Kulturwissenschaften), Reichsratsstraße 17, 1010 Vienna (ground floor)
Online audience: Please use the following Zoom links:
Donnerstag, 28.11.
us06web.zoom.us/j/87831389185
Meeting-ID: 878 3138 9185
Kenncode: mcX7T5
Freitag, 29.11.
us06web.zoom.us/j/87613835276
Meeting-ID: 876 1383 5276
Kenncode: ZwhF8y
Download the poster: here
The workshop is part of the research project HiACS, funded by one of Europes largest research funding institutions, the Volkswagen Foundation Hannover. Members are Jens Schröter and Andreas Sudmann, University of Bonn, Alexander Waibel and Fabian Retkowski KIT/Carnegie Mellon, as well as Anna Echterhölter and Markus Elias Ramsauer, University of Vienna: howisaichangingscience.eu/projektbeschreibung/
Programme
Thursday, November 28, 2024
13.00 CET / Opening remarks / 7 am EST /
Julia Boog-Kaminski and Andreas Gehrlach (IFK): Welcome to the IFK
Anna Echterhölter and Markus Ramsauer (University of Vienna): Introduction
13.20 CET / Keynote I / 7.20 am EST /
Evangelos Pournaras (University of Leeds):
Privacy as a Collective Value and how to Protect it in the Era of AI
14.20 CET coffee break
14.30 CET / Panel I – Perspectives from Media Ethnography and Archaeology / 8.30 am EST /
Chair: Jens Schröter (University of Bonn)
14.30-15.00 Fabian Retkowski (Karlsruhe Institute of Technology) and Andreas Sudmann (University of Bonn): Automated Coding: Multilabel-Classification in Ethnography
15.00-15.30 CET Andreas Sudmann and Jens Schröter (University of Bonn):
AI in Science and Epistemic Media
15.30 CET coffee break
16.00 CET / Panel II – Clustering and Detecting Tensions within the Social Order / 10 am EST /
Chair: Clemens Apprich (University of Applied Arts, Vienna)
16.00-16.30 CET Fenwick McKelvey (Concordia University, Montreal):
US Elections and the Electric Cluster Making Machine
16.30-17.00 CET Orit Halpern (Dresden University of Technology):
Mirror Worlds: Clustering, AI, and the Management of Catastrophe
17.00-17.30 CET Aaron Gluck-Thaler (Harvard University):
Identification through Pattern Recognition in Cold War America
17.30 CET coffee break
18.00 CET / Keynote II / 12.00 am EST /
Rebecca Lemov (Harvard University): History of Sentiment Analysis
Chair: Markus Ramsauer (University of Vienna)
20.00 CET dinner (ASPIC, Garnisongasse 10)
Friday, November 29, 2024
09.00 CET / Panel III – Automated Social Order / 3 am EST /
Chair: Sarah Davies (University of Vienna)
09.00-09.30 CET Dinah Pfau (Deutsches Museum, Munich):
Epistemology of a Matrix, or How Communication Technology Invented the Human
09.30-10.00 CET Eva-Maria Gillich (Bielefeld University): Norms out of Patterns
10.00-10.30 CET Tobias Matzner (Paderborn University):
On some Similarities between Clustering and Embeddings
10.30 CET coffee break
11.00 CET / Panel IV – Clustering: Work and Health / 5 EST /
Chair: Christian Dayé (Graz University of Technology)
11.00-11.30 CET Phoebe Moore: (University of Essex):
Affective Computing at Work: Policy Provocations and Rights for the Left
11.30-12.00 CET Rudolf Seising (Deutsches Museum, Munich): Fuzzy Sets and Systems: An Alternative Approach to Blur and Inaccuracy for AI in the 20th Century
12.00-12.30 CET Wrap Up
13.00 CET lunch (Gastwirtschaft Blauensteiner, Lenaugasse 1)
The research project and workshop are made possible by the Volkswagen Foundation.
Visit MiniSeg, the new benchmark video segmentation model, or a database of European research projects using AI. Also more on the working group HIACS: https://howisaichangingscience.eu/
The venue is provided by ifk: https://www.ifk.ac.at/
For more in formation on the history of science working group in Vienna: https://fsp-wissenschaftsgeschichte.univie.ac.at/
Conference Abstract
Clustering -- Automated Order in the Social Sciences
ai\research\explorations workshop IV
Organized by the Viennese working group of the project “How is AI Changing Science” in cooperation with the International Research Center for Cultural Studies (IFK) Vienna
28.-29. November 2024
One of the key elements in unsupervised learning is clustering. Thus, this particular data practice sits at the core of modern Artificial Intelligence, which is based on artificial neuronal networks. Whereas classification operates by organizing labeled data into specific categories, clustering relies on cheaper, unlabeled data for deciphering similarities inside a given set.
While many scientific disciplines might be interested in this new element of technical progress, the social sciences should be. The workshop poses the open question if unsupervised data clustering has the potential of identifying and generating new patterns of the social. This idea is not new. As Orit Halpern has remarked, attempts to break free from stable categories like race, identity, territory, or ethnicity with the help of pattern recognition can be found e.g. in the works of political scientist Karl Deutsch already in the 1960s (Halpern 2014, p. 191). Can clustering come up with entirely new orders of the social, such as tribes of movements identifiable from telephone data, do they detect political party affiliation, friendship or kinship-patterns that are not blood-related, and thus resemble totemistic orders? Or does automatization in the analysis of social data reproduce older hierarchies and familiar stratifications with necessity? While it is crucial not to fall prey to techno-utopian fantasies of non-situated (AI) technologies ‘overcoming’ race, class or gender, the transformative potential of clustering practices for analysis and reorganization of society and resource management in crisis should not be dismissed entirely.
While the history of quantification has made great strides to trace centers of calculation (Didier 2021; Wiggins and Jones 2023), while the cold war genealogy of AI is being established (Seising 2018; Dick 2021; Babintseva 2023), the history of data has developed additional perspectives. The focus lies with the practical handling of digital information as element of scientific or bureaucratic practices (Suchman 2006; Aronova 2017; Rheinberger 2018; De Chadarevian and Porter 2018; Dommann and Stadler 2020; Schlicht et al. 2021). The workshop singles out one particular episode from such a data journey (Leonelli and Tempini 2020). There are general discussions of classification and clustering (Arabie, Lawrence and de Soete 1996; Brunton and Kutz 2019; Bowker 2001), some already with respect to data practices within specific methods, and even fewer in the social sciences and humanities (Boumans and Leonelli 2020).
Clustering practices are typical for automated learning across the disciplines, which relies on large amounts of data, thereby introducing an element of noise and ambiguity. They do seemingly replace human cognition as source of social order. Thus, automated ordering can be said to introduce a novel element of ambiguity to the representation of society. From this position we ask, if clustering algorithms like K-means introduce elements of irrationality into planning processes or sociological methods. The automated social order comes with a new level of imprecision. The history of rationality is thus faced with a new ambiguity, after the probabilistic revolution of the 19th century did already alter the accepted forms of evidence (Krüger et al. 1987). When during the cold war, “reason almost lost its mind” (Erickson et al. 2013), the mind now seems to prevail over rational forms of conclusion.
We are particularly interested in this new mechanism that seemingly replaces conscious order with an automated process of matchmaking. As Sabina Leonelli has emphasized for the case of biology, clustering is the very pre-condition for data to become a representation of the world (Leonelli 2023, p. 317-318). Yet, in unsupervised learning, the outcomes seem to be uncontrollable. It stands to reason, which kind of tradeoff between rational planning and seemingly irrational ordering the new branch of computational sociology may soon come up with. Are we close to accepting a new age of similarities and epistemologies of similitudes which abandons factual evidence in exchange for patterns and noisy data clouds?
Following the suggestions of the history of data this conference approaches AI in social sciences from the perspective of one key practice. The workshop invites media archaeologists, historians of science and quantification as well as computational sociologists and data curators to reflect on the history, promise a potential of clustering. For this we want to establish both, the analogue and digital histories of clustering.
The Project: How is AI Changing science
The workshop is funded by the Volkswagen Foundation and part of a larger research project. We ask how artificial intelligence (AI) technologies do affect research and science? By following this perspective, the project is less concerned with research on AI per se than with how different disciplines use AI as a tool and as an epistemic entity within larger (post)digital infrastructures. The central focus lies on how heterogeneous concepts and operations of the social sciences and humanities, on the one hand, and the natural and technical sciences, on the other, are integrated into applications of AI. Research on the latter will also explore the extent to which critical perspectives inform and accompany the use of AI. The project concentrates on artificial neural networks (ANN) because of their dominant status among current AI approaches. Hence, the project not only explores the similarities and differences between the various areas of application of AI, but also sheds light on the cultural and national specificities inherent to these processes in an international context, particularly in Europe and the USA.
howisaichangingscience.eu/projektbeschreibung/
Members: Anna Echterhölter, Markus Elias Ramsauer University of Vienna, Jens Schröter and Andreas Sudmann, University of Bonn and Alexander Waibel and Fabian Retkowski KIT/Carnegie Mellon
Literature:
Arabie, Phipps, Lawrence J. Hubert, and Geert De Soete. Clustering and Classification. Singapore: World Scientific 1996.
Aronova, Elena, von Oertzen, Christine und Sepkoski, David: Introduction: Historicizing Big Data. In: Osiris 32/2017, S. 1–17.
Babintseva, Ekaterina: Rules of Creative Thinking: Algorithms, Heuristics, and Soviet Cybernetic Psychology. In: BJHS Themes, Volume 8: Histories of Artificial Intelligence: A Geneology of Power 2023, S. 81-95.
Boumans, Marcel, and Sabina Leonelli: From Dirty Data to Tidy Facts: Clustering Practices in Plant Phenomics and Business Cycle Analysis. In: Data Journeys in the Sciences, edited by Sabina Leonelli and Niccolò Tempini. Heidelberg: Springer 2020, S. 79–101.
Bowker, Geoffrey C.: Biodiversity Datadiversity. In: Social Studies of Science 30:5/2001, S. 643–684.
Brunton, Steven L., and J. Nathan Kutz. ‘Classification and Clustering’. In: Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control. Cambridge, MA: Cambridge University Press 2019, S. 154–194.
De Chadarevian, Soraya and Porter, Theodore: Introduction: Scrutinizing the Data World. In: Historical Studies in the Natural Sciences 48:5/2018, S. 549–556.
Dick, Stephanie: Of Models and Machines: Implementing Bounded Rationality. In: ISIS 106: 3/ 2015, S. 623-634.
Didier, Emmanuel: Quantitative Marbling, New Conceptual Tools for the Socio-history of Quantification (=Anton Wilhelm Amo Lectures n°7) Martin-Luther-Universitat, Halle-Wittenberg press 2021.
Dommann, Monika, and Max Stadler: Introduction. In: Data Centers: Edges of a Wired Nation, hg. von Monika Dommann, Hannes Rickli und Max Stadler. Zürich: Lars Müller Publishers 2020, S. 8–29.
Erickson, Paul, Judy L Klein, Lorraine Daston, Rebecca M. Lemov, Thomas Sturm and Michael D. Gordin: How Reason Almost Lost Its Mind : The Strange Career of Cold War Rationality. Chicago: University of Chicago Press 2013.
Halpern, Orit: Beautiful Data: A History of Vision and Reason since 1945. Durham, N.C.: Duke University Press 2014.
Krüger, Lorenz et al. (Hg.): The Probabilistic Revolution. 2 Bde. Cambridge: MIT 1987.
Leonelli, Sabina: Classificatory Theory in Data-Intensive Science: The Case of Open Biomedical Ontologies. In: International Studies in the Philosophy of Science 26:1/2012, S. 47–65.
Laubichler, Manfred D.; Maienschein, Jane; Renn, Jürgen: Computational Perspectives in the History of Science: To the Memory of Peter Damerow. In: Isis 104/2013, S. 119–130.
Rheinberger, Hans-Jörg: Infra-Experimentality: From Traces to Data, from Data to Patterning Facts. In: History of Science 49:3/2011, S. 337–348.
Schlicht, Laurens; Ledebur, Sophie; Echterhölter, Anna (Hg.): Data at the Doorstep. Intimacy and Formalization. In: Science in Context 34:4/2021 (topical issue).
Seising, Rudolph: The Emergence of Fuzzy Sets in the Decade of the Perceptron—Lotfi A. Zadeh’s and Frank Rosenblatt’s Research Work on Pattern Classification. In: Mathematics/Fuzzy Mathematics 6:7/2018, S. 227–246.sei
Suchman, Lucy: Human-Machine Reconfigurations: Plans and Situated Actions. Cambridge: Cambridge University Press 2006.
Wiggins, Chris; Jones, Matthew L.: How Data Happened. A History from the Age of Reason to the Age of Algorithms. New York: Norton 2023.
