EASST 2026 Panel: Synthetic Data and Representation – The Politics of AI Generated Computational Practices

The increased use of artificially generated data and media raises a multitude of concerns, including how the world is represented, and how generated data is enacted as equivalent through chains of “circulating reference” (cf. Latour, 1999). Synthetic data is increasingly used to represent real-world referents and is presented as a solution to key problems of data-driven technological development and knowledge production (Lee et al. 2025). For example, generative models are used to mimic health data that would otherwise be difficult to share or to simulate traffic to train self-driving cars (Jacobsen, 2023). As our futures become increasingly intertwined with synthetic data, this panel aims to investigate what forms of knowledge, futures, and worlds that they enact.

We aim to create a meeting space for posing questions about the increasing use of synthetic data: What are the epistemic consequences of using synthetic data? How does the relation between “reality,” “data,” and “representation” change? How is synthetic data used to address “missingness” and gaps of representativity? What are the impacts on transparency and understanding of representations, as synthetic datasets travel and are reused across projects and worlds? How can we evaluate the accuracy of synthetic datasets?

By posing these questions, this panel will expand emerging discussions in STS about the ontological politics of artificial intelligence and synthetic data (Jacobsen, 2023, Johnson & Hajisharif 2025, Lee et al. 2025). One of the goals of the panel is to connect ongoing discussions about data (Gitelman, 2013; Leonelli, 2019), algorithms (Seaver, 2017), and artificial intelligence (Suchman, 2023) in order to inquire into how synthetic data is emerging as a new set of practices and politics in knowledge production. We welcome empirical, theoretical, and artistic contributions that engage with synthetic data in knowledge making.

The panel has received 15 accepted contributions from researchers at universities in Sweden, Italy, the United Kingdom, Norway, and Denmark. Papers range across empirical domains—medicine, urban governance, facial recognition, historical archives, and social science methodology—while sharing a common concern with the epistemic and political stakes of AI-generated data. Together they ask not only what synthetic data makes possible, but what it silences, excludes, and reshapes.