[Air-L] [Flashlab Seminar] Dieuwertje Luitse (UvA) - June 2, 2025 (4-5:30pm CET) - "Rethinking Data Bodies: On the Technopolitical Production of ‘Otherness’ through ML Modelling and Evaluation in Medicine"

[Assia WIRTH]

Dear colleagues,



The Flashlab seminar’s quarterly cycle "Contemporary Challenges of
Artificial Intelligence" third and final session will take place on*
Monday, June 2 (4pm-5:30pm CET),* online and at Sciences Po Paris (1 place
St-Thomas d'Aquin, Paris). We are pleased to welcome *Dieuwertje Luitse*,
PhD student at the University of Amsterdam and co-organizer of the "*Critical
AI Seminar*
<https://www.create.humanities.uva.nl/seminar-series/the-critical-ai-seminar-series-returns/>",
for a presentation entitled:



*Rethinking Data Bodies: On the Technopolitical Production of ‘Otherness’
through ML Modelling and Evaluation in Medicine *



*Abstract: *Following the rapid development of AI technologies in the
medical domain, the amplification of bias and the potential reproduction of
existing social or health inequities have become a central object of
concern in the field (Seyyed-Kalantari et al., 2021; Lee et al., 2023). To
mitigate these issues, critical work on the ethics and politics of medical
AI has primarily focused on scrutinising the construction of (very) large
datasets used to train and evaluate machine-learning systems for medical AI
research and development or clinical deployment. Even though this type of
critical work in AI ethics is vital, I stress that this particular focus on
data and data practices risks obscuring the many additional factors at play
here. Instead, AI-inscribed bias and health inequalities may emerge through
‘small differences’ in a series of entangled processes that underpin
medical AI decision-making: machine-learning modelling and specific modes
of evaluation. It is critical to further investigate these processes, and
the underlying practices that feed into them, as they allow to better
understand and deepen the ethico-political scrutiny of technical operations
by machine-learning systems beyond just data production.

Following these observations, this talk takes a combined Critical AI and
STS research approach to demonstrate how data, machine-learning modelling
and evaluation processes are continuously composing and recomposing
categories of ‘others’ on which AI-based medical decisions can be
made—including those that may lead to the (re)productions of social biases
and health inequities. To do so, I propose to conceptually rethink the
construction of ‘data bodies’ as technopolitical enactments of constituted
medical datasets, machine-learning models and their underlying
infrastructures for AI-based medical decision-making. Drawing on the notion
of the ‘body multiple’ (Mol, 2002) data bodies are constituted and always
‘oscillate between multiplicity and singularity’ (Bucher, 2018). Exploring
their enactment, I address the configurations of the lines between self and
various others from within the entangled technical arrangements and
sociotechnical practices underlying machine-learning operations in the
medical realm and consider their implications for decision-making in
practice.

*Bio: **Dieuwertje Luitse is PhD Candidate in the Department of Media
Studies at the University of Amsterdam (UvA). Her research explores the
ethics, politics and power of data, artificial intelligence (AI) system and
application development in healthcare. This project is part of the UvA’s
interdisciplinary research priority area on AI for Health decision-making
<https://www.uva.nl/en/shared-content/zwaartepunten/en/artificial-intelligence-for-health-decision-making/artificial-intelligence-for-health-decision-making.html>
that brings together researchers from Computer Science, Medicine, Law and
the Humanities. In addition, her research interest focuses on the study of
computational infrastructures and the (historical) development of AI
systems in relation to their socio-economic and political implications.In
line with her research activities, Dieuwertje is the co-organiser of the
Critical AI Seminar Series
<https://www.create.humanities.uva.nl/seminar-series/the-critical-ai-seminar-series-returns/>
hosted by the Critical Data and AI research group at UvA’s Faculty of
Humanities and one of the co-editors of the topical collection on the
Politics of Machine Learning Evaluation
<https://link.springer.com/collections/bbbehaibcj> in *Digital Society.

---

To get this session’s link, please sign up at this link
<https://framalistes.org/sympa/subscribe/flashlab> or get in touch via
email.



The full program is available on the seminar’s *site*
<https://flashlabinfo.wordpress.com/>.

---



For those interested, a session of the medialab seminar will also be held
on the following day, *Tuesday 3 June* *(2-4pm)* – during which* Dieuwertje
Luitse *et *Anna Schjøtt Hansen *will introduce a special issue they
co-edited for *Digital Society* on the *"Politics of Machine Learning
Evaluation"*, with a collective discussion with some of the contributors.
All information is available on the* medialab *
<https://medialab.sciencespo.fr/en/news/the-politics-of-machine-learning-evaluation-from-present-to-future/>
website.



Kind wishes,

Valentin Goujon & Assia Wirth

-----------

*Assia Wirth*

PhD candidate - ENS Paris Saclay (IDHES)