Challenges and affordances of doing media research with generative AI

This series explores the challenges and affordances of doing media research with generative AI. AI is not only a subject of media studies, but deployed to collect data, interpret texts, perform mult-modal analysis, and assist writing. But what happens when prompts, models, and training data enter the methodological core of media studies? On what grounds can we cooperate with AI in research? The seminar series Synthetic Methods takes these questions as its point of departure. It explores current practices, tools, approaches and issues of synthetic methods, asking how AI participates in producing, mediating, and interpreting knowledge.

In recent years, the tools and infrastructures of generative AI—large language models, multimodal systems, and computer-vision pipelines—have begun to blur the boundaries between data collection, analysis, and interpretation. AI does not simply assist researchers in automating tasks; it brings in its own epistemic logics, biases, and inscriptions. Text and image generation models suggest categories, segment data, or simulate field interactions. They act as synthetic interlocutors in ethnographic work, as co-coders in qualitative analysis, or as analytical lenses in cultural analytics. The series engages with these developments hands-on and conceptually, examining what it means to “do research with AI.”

At the centre of the series lies an interest in the distributed accomplishment of discovery between humans and AI. Instead of handing analytical capacity entirely to computational systems, we will explore how reasoning, interpretation, and sense-making can emerge collaboratively across human and AI agencies. Generative models may extend perception and imagination, but they also depend on human intervention, interpretation, and evaluative judgment. The sessions thus foreground research as a shared practice of translation and negotiation, where human reflexivity and AI inference together shape what counts as evidence, relevance, and insight. This distributed perspective opens the space to examine the affordances, issues, and evaluative criteria that govern scholarship when AI becomes part of the epistemic process: How can we maintain reflexive, critical and ethical orientations while experimenting with new, mixed agencies of knowing?

Over the course of the semester, the series will address a range of perspectives and practices. An initial session on infrastructures and AI ethics situates large-scale models within the political economies of cloud computing, highlighting questions of privacy, transparency, and data provenance. Subsequent meetings explore how AI reshapes established methodological domains: as an assistant in qualitative analysis and ethnography, as a writing companion and reflective mirror in academic text production, and as a tool for analyzing visual and multimodal materials. Participants will experiment with both commercial and locally hosted models, comparing their capacities and constraints.

Specific attention will be given to the question if and how synthetic methods require specific modes and practices of methodological reflexivity. The series does not treat models as neutral instruments but as infrastructures with their own histories, biases, and aesthetics. Engaging with generative systems thus becomes an exercise in distributed reflexivity: models, prompts, and humans co-produce insight. This distributed agency raises fundamental questions of authorship, responsibility, and transparency that reach beyond technical documentation. To “work synthetically” is to navigate this entanglement without surrendering critical distance—to cultivate a mode of inquiry that remains aware of its own mediations.

Schedule
22.10. Infrastructure and synthetic reasoning – Understanding beyond AI with Hinau Firdaus and Amita Kapoor

05.11. AI for Ethnographic Analysis

19.11. Writing with AI with Sergei Pashakhin

3.12. System Prompts with Marcus Burkhardt & Hendrik Bender

17.12. Voice of Machine Theft with Rosa Menkman und David Gauthier

14.01. Metabolic Images and Method Maps with Elena Pilipets

Location

All lectures take place on-site in Siegen with a hybrid setting. You can register to get the Webex-link and join sessions online.

University Siegen
room: AH-A 217/18
Herrengarten 3
57072 Siegen

Registration

Please register via email to info[æt]sfb1187.uni-siegen.de

Organisation

The series is organized by INF project „Infrastructures for Collaborative Sensory RDM Practices“ in collaboration with Carolin Gerlitz, Elena Pilipets, Dominik Schrey, Sara Messelaar Hammerschmidt, Sergei Pashakhin & Hina Firdaus.