Synthetic data: provocations and frictions of emerging data regimes

Synthetic data are defined as information created by computer simulations or algorithms reproducing structural and statistical properties of so-called “real-world” data. In recent years, synthetic data have emerged as a promise to fix such problems of the AI industry as more and better-quality data for training datasets, protecting personal data via anonymization, and mitigating data bias. A growing field of scholarship uncovers issues regarding the emergence of new data economies and labour practices in synthetic data industries. Furthermore, synthetic data impacts the level of policy, data regulation and notions of personal data. Although the idea of synthetic data has been known and well-used since the 1940s, its potential applications as training data for AI present new challenges and require thorough investigation from social sciences and the humanities.

By surfacing frictions and provocations of synthetic data, the seminar aims to create a fruitful dialogue within a nascent field. We invite contributions on synthetic data and the synthetic turn from the social sciences and humanities, such as e.g. critical data studies, cultural studies, cultural theory, philosophy, political economy, sociology and STS.

Attendance is free but requires registration. If you want to present a paper, please send a short abstract (around 150-200 words) to kpas@hum.ku.dk by 10 October.