Springe direkt zu Inhalt

Paper published in the Convergence journal

News from Nov 28, 2022

We are pleased to announce that our paper "From critical technical practice to reflexive data science"[1] by Simon David Hirsbrunner, Michael Tebbe and Claudia Müller-Birn was published in the »Convergence« journal.


In this article, we reconsider elements of Agre’s critical technical practice approach (Agre, 1997) for critical technical practice approach for reflexive artificial intelligence (AI) research and explore ways and expansions to make it productive for an operationalization in contemporary data science. Drawing on Jörg Niewöhner’s co-laboration approach, we show how frictions within interdisciplinary work can be made productive for reflection. We then show how software development environments can be repurposed to infrastructure reflexivities and to make co-laborative engagement with AI-related technology possible and productive. We document our own co-laborative engagement with machine learning and highlight three exemplary critical technical practices that emerged out of the co-laboration: negotiating comparabilities, shifting contextual attention and challenging similarity and difference. We finally wrap up the conceptual and empirical elements and propose Reflexive Data Science (RDS) as a methodology for co-laborative engagement and infrastructured reflexivities in contemporary AI-related research. We come back to Agre’s ways of operationalizing reflexivity and introduce the building blocks of RDS: (1) organizing encounters of social contestation, (2) infrastructuring a network of anchoring devices enabling reflection, (3) negotiating timely matters of concern and (4) designing for reflection. With our research, we aim at contributing to the methodological underpinnings of epistemological and social reflection in contemporary AI research.

For more information click here.

[1] Hirsbrunner, S. D., Tebbe, M., & Müller-Birn, C. (2022). From critical technical practice to reflexive data science. Convergence, 0(0). https://doi.org/10.1177/13548565221132243

5 / 100