„Das MGK ist der zentrale Ort der Methodenvermittlung und -innovation. Es trägt maßgeblich zum übergeordneten Ziel des SFB bei, digitale Grundlagenforschung zu leisten sowie innovative Methoden zu vermitteln, zu reflektieren und zu entwickeln. Das MGK bildet Promovierende in einem international orientierten, interdisziplinären Programm aus.“
Synthetic Imaginaries: The Cultural Politics of Generative AI
University of Siegen | 8–12 September 2025
Synthetic Imaginaries: The Cultural Politics of Generative AI is an international event that will explore the cultural, political, and methodological dimensions of generative AI and synthetic media through a combination of conference talks, hands-on workshops, and collaborative projects. Topics include deepfakes, avatars, cultural biases in training data, feminist and postcolonial critiques, and the aesthetics of AI-generated content.
The rise of artificial intelligence (AI), big data processing, and synthetic media has profoundly reshaped how culture is produced, made sense of, and experienced today. To ‘synthesize’ is to assemble, collate, and compile, blending heterogeneous components into something new. Where there is synthesis, there is power at play. Synthetic media—as exemplified by the oddly prophetic early speech synthesizer demos—carry the logic of analog automation into digital cultures where human and algorithmic interventions converge. Much of the research in this area—spanning subjects as diverse as augmented reality, avatars, and deepfakes—has revolved around ideas of simulation, focusing on the manipulation of data and content people produce and consume. Meanwhile, generative AI and deep learning models, while central to debates on artificiality, raise political questions as part of a wider social ecosystem where technology is perpetually reimagined, negotiated, and contested: What images and stories feed the datasets that contemporary AI models are trained on? Which imaginaries are reproduced through AI-driven media technologies and which remain latent? How do synthetic media transform relations of power and visibility, and what methods—perhaps equally synthetic—can we develop to analyze these transformations?
The five-day event at the University of Siegen—organized by the DFG-funded Collaborative Research Centers Media of Cooperation and Transformations of the Popular together with the Center of Digital Narratives in Bergen, the Digital Culture and Communication Section of ECREA and the German National Research Data Infrastructure Consortium NFDI4Culture—explores the relationship between synthetic media and today’s imaginaries of culture and technology, which incorporate AI as an active participant. By “synthetic,” we refer not simply to the artificial but to how specific practices and ways of knowing take shape through human-machine co-creation. Imaginaries, in turn, reflect shared visions, values, and expectations—shaping not only what technologies do but how they are perceived and made actionable in everyday life.
The event opens with a one-day conference and moves into hands-on workshops and collaborative projects. With multiple opportunities for exchange across disciplines, we encourage especially early-career researchers and PhD students to present their ideas during the conference and join a project led by international facilitators and data designers. We invite submissions of short abstracts (max. 500 words) for presentations engaging with questions and provocations related—but not limited—to topics such as:
Critical data studies perspectives on AI: how data infrastructures, labeling, and curation shape the outputs we call “synthetic”; Cultural afterlives of training data: how racialized, gendered, or colonial imaginaries persist in synthetic media outputs; Methodological uses of GenAI: the politics that we buy in when repurposing AI as a method, from inherited bias to epistemic tensions; Synthetic personhood and likeness: exploring deepfakes, AI-generated avatars, and the power of (in)authenticity; Online cultures and platforms: how AI-generated content circulates across platforms—from memes and art to fan fiction, music, and poetry; Postcolonial and feminist critiques of AI: challenging universalist assumptions in generative models and interrogating whose knowledge is made (in)visible; Clichés, formulas, and repetition in GenAI outputs: how AI-generated stories and images rely on familiar tropes, visual styles, and narrative conventions; The aesthetics of noise in AI-generated content: repetition, glitch, randomness, and their role in producing or disrupting meaning; GPTs as infrastructural components: how generative pretrained transformers operate as configurable, customizable, and task-oriented agents embedded in platform infrastructures; Prompting and/as probing: prompting as a form of critical intervention, shaping co-authorship, sense-making, and research design; The ethics of training AI: from historical records and religious texts to indigenous cosmologies and oral traditions—what are the implications of using culturally sensitive knowledge to train generative models? Generative AI and Memory: synthetic media as a means of reimagining the past—through deepfake testimonies, interactive historical simulations, and other forms of computational memory-making; Generative AI in activist contexts: can AI be used for resistance or reimagining community—in the face of its environmental footprint and complicity in extractive systems?
Program highlights
The event blends three complementary formats:
Mix questions!
Monday, 8 September
Day one begins with a keynote by Jill Walker Rettberg and opens space for emerging questions—think of it as an idea hub. Accepted abstracts will be grouped into thematic sessions curated by the organising team. Presenters will be connected via email ahead of time to coordinate their contributions. Each presentation will be set to 10 minutes to allow ample time for discussion, collective thinking, and exchange. The emphasis is on dialogue, not polished conclusions.
Mix methods!
Tuesday, 9 September-Thursday, 11 September
The next three days—featuring a workshop by Gabriele De Seta and an artistic intervention by Ángeles Briones and DensityDesign Lab—are about exploring new methods—hands-on! We invite you to join a team of interdisciplinary scholars and data designers in probing new methodological combinations. Each of our project teams will present a research question alongside a specific method to be collaboratively explored. Participants will not only learn how to design prompts and work with AI-generated text and images but also how to critically account for genAI models as platform models. All projects draw on intersectional approaches, combining qualitative and quantitative data to explore the synthetic dimensions of AI agency—whether as content creator, noise generator, hallucinator, research collaborator, data annotator, or style imitator. Please bring your laptops. The project titles will be announced soon.
Synthesize!
Friday, 12 September
The final day is dedicated to sharing, reflecting, and synthesizing the questions, methods, and insights developed throughout the week. Project teams will present their collaborative processes, highlight key takeaways, and discuss how their ideas and approaches shifted through hands-on experimentation with methods.
Proposal Submission
Please submit your proposal (max. 500 words) outlining how your work aligns with the event’s theme by 20.06.2025, using this form. Please note that the number of participants will be limited to maintain focused and engaging discussions. All submissions will be peer-reviewed.
The event is free of charge, though attendees are responsible for arranging and covering their travel and accommodation in Siegen. Limited travel support is available (two to three stipends ranging from €500 to €700). Early-career researchers and PhD students are invited to apply; stipends will be awarded by the NFDI4Culture consortium based on the strength of the justification, particularly concerning critical ethical engagement with AI research data, as well as the distance and cost of travel. Short summaries of the presented work will be published on the NFDI4Culture website.
A certificate of participation will be issued for both the conference presentation and the hands-on workshop sessions.
Timeline:
Submit your proposal by 20 June 2025.
Notification of acceptance by July 15 2025.
Registration by August 1 2025.
The Autumn School is organized by the DFG-funded Collaborative Research Centers 1187 Media of Cooperation and 1472 Transformations of the Popular together with the Center of Digital Narratives in Bergen, the Digital Culture and Communication Section of ECREA and the German National Research Data Infrastructure Consortium NFDI4Culture.
In February 2025, the Mixing Methods Winter School at the Collaborative Research Centre 1187 brought together over thirty participants, including international researchers, students, and experts from various disciplines. The program combined hands-on experimentation with critical inquiry into AI-driven research methods. Throughout the Winter School, participants critically engaged with AI not just as a tool but as a collaborator, reflecting on its role in shaping the research process.
The week opened with a keynote by Jill Walker Rettberg from the University of Bergen, who introduced “Qualitative Methods for Analyzing Generative AI: Experiences with Machine Vision and AI Storytelling.” Her talk set the stage for discussions on how qualitative inquiry can reveal the underlying narratives and biases in AI-generated content.
Participants then engaged in two hands-on workshops designed to explore mixed techniques for probing and prompting AI models. Carlo de Gaetano (Amsterdam University of Applied Sciences), Andrea Benedetti, and Riccardo Ventura (Density Design, Politecnico di Milano) led the workshop “Exploring TikTok Collections with Generative AI: Experiments in Using ChatGPT as a Visual Research Assistant,” examining how AI can assist in visual analysis of networked video content. Together with Elena Pilipets (University of Siegen) and Marloes Geboers (University of Amsterdam) participants then explored the semantic spaces and aesthetic neighborhoods of synthetic images generated by Grok during the workshop “Web Detection of Generative AI Content”.
After an introductory first day, the Winter School shifted its focus to two in-depth project tracks. The first project, “Fabricating the People: Probing AI Detection for Audio-Visual Content in Turkish TikTok,” explored how protesters and the manosphere engage with cases of gender-based violence on Turkish TikTok and how these videos can be studied using different AI methods. The second project, “Jail(break)ing: Synthetic Imaginaries of ‘Sensitive’ AI,” explored how AI models reframe sensitive topics through generative storytelling under platform-imposed restrictions.
The project explored video shorts from the Turkish manosphere – content centered on masculinity, gender dynamics, and “men’s rights” issues that often discuss dating, self-improvement, and family life. While this content is found on mainstream platforms and passes moderation, it still frequently veers into misogynistic or even violent rhetoric. Our project explored AI-assisted methods to make sense of large amounts of this contentious multimodal data.
Rationale
Specifically, we set out to develop methods to map how video shorts may become a vehicle for the ambient amplification of extremist content across platforms. We explored two approaches using off-the-shelf multimodal large language models (LLMs). The first sought to extend the researcher’s interpretation of how manosphere content addresses bodies, which are both performed and contested intensely across the issue space. We did this by implementing few-shot labelling of audio transcriptions and textual descriptions of videos. The second method sought to interrogate the role of generative AI in (re)producing memes, genres, and ambience across video shorts. We achieved this by experimenting with zero-short descriptions of video frames to describe detected genres, formats and the possible use of AI in video production processes.
Methods and Data
We started with a period of “deep hanging out” in Turkish manosphere and redpill spaces on Tiktok, Youtube, and Instagram. We identified prominent accounts and crawled them to build a data sample of 3600 short videos from across the three platforms. Several analyses were carried out before the Winter School. These included metadata scraping, video downloading, scene detection, scene collage creation. transcribing audio, and directing an LLM to generate video descriptions following Panofsky’s a three-step iconological method, which differentiates between pre-iconographic analysis (recognizing fundamental visual elements and forms), iconographic analysis (deciphering symbols and themes within their cultural and historical contexts), and iconological interpretation (revealing deeper meanings, ideologies, and underlying perspectives embedded in the image) (Panofsky, 1939).
Method one: Video shorts continue to grow in popularity and prominence across social media platforms, building out new gestural infrastructures (Zulli and Zulli, 2022) and proliferating ambient images (Cubitt et al., 2021). Qualitatively investigating this rich multimodal content at a scale that highlights the broader atmospheres and cultures developed through algorithmic circulation is challenging. Multimodal LLMs have the potential to extend researcher’s ethnographic coding capacity to larger datasets and to account for more varied formats than ever before (Li and Abramson, 2023). We therefore investigate possibilities for using cutting-edge multimodal LLMs for qualitative coding of multimodal data as a methodology for qualitatively investigating ambience and amplification in video-short-driven algorithmic media.
We began with a qualitative ethnographic immersion in our dataset, watching the videos and developing a codebook that described how the videos related to our interest in how bodies were both performatively and discursively addressed. We applied our codebook manually to the textual data the LLM allowed us to produce out of the videos, i.e., not only the metadata, but also audio transcriptions and LLM-generated video Panofskian descriptions. After the codebook stabilized, we applied it to a random subset of 150 datapoints. We then developed a few-shot learning script that applied these labels to the entire dataset. We chose three examples to belong to a labelling “core” and then programmed a script to sample dynamically from the rest of our 150 datapoints to include as many further examples as could be accommodated by the context window limitation. We then prompted the LLM to apply our labels to the entire dataset. This let us explore extending the researcher’s qualitative insights to larger, multimodal data.
During the codebook development and coding process, the Panofsky descriptions brought the visual prominence of hands and hand gestures across the dataset to our attention. We therefore also applied a separate process to our data to begin isolating hands for closer investigation.
Method two: Automation technologies and generative AI play an increasingly prominent role in the creation of audio-visual social media content. This ranges from image or video generation, to AI voice-over production, to video editing, content cropping, platform optimization, and beyond (Anderson and Niu, 2025). Detecting these production methods, however, is challenging. Even state-of-the-art machine-learning struggles to analyze multimodal media (Bernabeu-Perez, Lopez-Cuena and Garcia-Gasulla, 2024). We set out to find qualitative alternatives for exploring the role of AI aesthetics in video short production. Therefore, we proceeded with a twofold approach, developed within an iterative process of prompt engineering: First, we asked the LLM to create a structured visual analysis of a social media video collage by evaluating its composition, camera techniques, editing style, mise-en-scène, text overlays, genre, and platform-specific features, summarizing key characteristics as a tag list. This initial prompt helped distinguish between different video formats and styles, identifying those that are particularly likely to incorporate automation or AI-driven edits. Second, we directly instructed the LLM to assess the likelihood that AI was used in the production of this video. In this way, we set out to explore “popular” AI’s role in both the creation and the interpretation of misogynistic video-shorts.
Research questions
How can AI-based methods be used to extend ethnographic research into networked digital cultures?
How can these methods help increase researcher sensitivity to phenomena that happen at network scale, for example, ambient amplification practices?
Can AI identify and characterize synthetic content? How does AI see AI?
As an approximation of that question, how does AI interpret and distinguish between different content genres and formats?
Key findings
Our work demonstrated the extent to which the internal cultural logic of the LLM cannot be separated from its output as a tool (Impett and Offert, 2022) – and therefore how LLMs, when used as tools, are inevitably also always reflexively the object of study. When designing processes for “co-creation” and collaboration with LLMs, the logic of the LLM repeatedly overpowered our own efforts to insert our intentions and directions into the process. This suggests that the most fruitful way to use out-of-the-box LLMs as an ethnographic research tool for the study of digital cultures is to lean into – and critically interrogate – its internal cultural logic instead of trying to bend it to our own. Obtaining results that reflect our intentions more closely will require more extensive technical methods, e.g., fine-tuning models and extensive many-shot prompting or alternative machine-learning approaches.
By letting the LLM reveal its own internal logic, however, we anticipate being able to use LLMs as a way to highlight the machine-readable and machine-reproducible qualities of the multimodal networked space itself (Her, 2024). The LLM’s internal logic can help foreground the fact that this media is also created by and for machines to consume, and reveal how generative LLMs applied to problematic cultural spaces interpret, (re)structure, (re)produce cultures of hate in “popular” spaces.
A comprehensive report is in progress.
Jail(break)ing: Synthetic Imaginaries of (sensitive) AI
This project has explored how three generative AI models—X’s Grok-2, Open AI’s GPT4o, and Microsoft’s Copilot—reimagine controversial visual content (war, memes, art, protest, porn, absurdism) according to—or pushing against—the platforms’ content policy restrictions. To better understand each model’s response to sensitive prompts, we have developed a derivative approach: starting with images as inputs, we have co-created stories around them to guide the creation of new, story-based image outputs. In the process, we have employed iterative prompting that blends “jailbreaking”— eliciting responses the model would typically avoid—with “jailing,” or reinforcing platform-imposed constraints.
We propose the concept of ’synthetic imaginaries‘ to highlight the complex hierarchies of (in)visibility perpetuated by different generative AI models, while critically accounting for their tagging and visual storytelling techniques. To ‘synthesize’ is to assemble, collate, and compile, blending heterogeneous components—such as the data that MLLMs (Multimodal Large Language Models) integrate within their probabilistic vector spaces—into something new. Inspired by situated and intersectional approaches within critical data(set) studies (Knorr-Cetina 2009; Crawford and Paglen 2019; Salvaggio 2023; Pereira & Moreschi 2023; de Seta et al. 2024; Rettberg 2024) we argue that, „synthetic“ does not merely mean artificial; it describes how specific visions—animated by automated assessments of data from a wide range of cultural, social, and economic areas—take shape in the process of human-machine co-creation. Some of these visions are collectively stabilized and inscribed into AI-generated outputs, revealing normative aspects of text-image datasets used to train the models. Others assemble layers of cultural encoding that remain ambiguous, contested, or even erased—reflecting how multiple possibilities of meaning fall outside dominant probabilistic patterns.
While generative models are often perceived as systems that always produce output, this is not always the case. Like social media platforms, most models incorporate filters that block or alter content deemed inappropriate. The prompting loops—from images to stories to image derivatives—involve multiple rounds of rewriting stories generated by the model in response to input images. The distance between input and output images corresponds with the transformations in the initially generated and revised (or jailed) image descriptions.
As a method, jail(break)ing exposes the skewed imaginaries inscribed in the models’ capacity to synthesize compliant outputs. The more storytelling iterations it takes to generate a new image, the stronger the platforms’ data-informed structures of reasoning come to the fore.
Methods and data
While our collection of sixty input images covers a range of seemingly unrelated issues, they all share two qualities: ambiguity and cultural significance. Many of these images qualify as sensitive, yet they are also widely and intensely circulated on ‘mainstream’ social media platforms.
Visual interpretation: Through a qualitative cross-reading of AI-generated output images, we analyzed how three different models respond to image-driven storytelling prompts. Through multimodal prompting (“I give you an image, you tell me a story”), stories were co-created to inform the generation of output images. By synthesizing ten output images per issue space into a canvas, we then examined how AI systems reinterpret, alter, or censor visual narratives and how these narratives, in turn, reinforce issue-specific archetypes.
Narrative construction: We approached image-to-text generation as structured by the operative logic of synthetic formulas—setting (where is the story set?), actors (who are the actors?), and actions (how do they act?). Driven by repetition-with-variation, these ‘formulas’ (Hagen and Venturini 2024), reveal narrative patterns and semantic conventions embedded in the models’ training data.
Keyword mapping: We analyzed AI-generated descriptions of images’ content, form, and stance across models. Exploring both unique and overlapping keywords, the method uncovers how each model prioritizes certain vernaculars as a tagging device.
Research Questions
Which stories can different AI models tell about different images, and which story archetypes emerge in the process of jail(break)ing?
When do the models refuse to generate images? Which stories remain unchanged, and which need to be transformed?
Which keywords do the models assign to describe the images’ content, form, and stance?
Key Findings
The different AI models—Grok-2, GPT-4o, and CoPilot—tell distinct stories about images based on their internal biases, content policies, and approaches to sensitive material. Their generated narratives differ in terms of modification, censorship, and interpretation, reflecting platform-specific content moderation frameworks.
Grok-2 preserves more of the original content, making fewer alterations unless forced by content restrictions. It allows more controversial elements to remain but often introduces confusing substitutes.
GPT-4o significantly neutralizes content, shifting violent, sexual, or politically sensitive imagery toward symbolic and abstract representations. It frequently removes specific cultural or historical references.
CoPilot enforces the strictest content restrictions, often refusing to generate images or stories for sensitive topics altogether. It eliminates references to nudity, violence, or political figures and transforms potentially controversial scenes into neutral, inoffensive portrayals.
Stricter content policies amplify narrative techniques like suspense-building in AI-generated stories. CoPilot and GPT-4o lean into verbose storytelling to comply with guidelines, often elevating uncontroversial background elements into agentic forces. In the ‘war canvas’ story, for instance, CoPilot foregrounds the background, narrating: ‘The square pulses with energy, driven by a community determined to create change.’ Grok, by contrast, sometimes fabricates entirely new subjects—golden retrievers replacing NSFW models—paired with objects like fluffy carpets. In other cases, the model inserts public figures into generic scenarios, intensifying the images’ impact.
Generative AI’s so-called sensitivity is a synthetic product of dataset curation, content moderation, and platform governance. What models permit or reject is shaped by training data biases, corporate risk management, and algorithmic filtering, reinforcing dominant norms while erasing politically or socially disruptive elements. Rather than genuine ethical awareness, these systems engage in selective sanitization, softening controversy while maintaining an illusion of neutrality. This raises critical questions about who defines AI “sensitivity,” whose perspectives are prioritized over others, and how these mechanisms shape epistemic asymmetries in digital culture.
Catalina Goanta, Birgit Meyer, Benjamin Peters, Jürgen Streeck und Jill Walker Rettberg sind neue Mercator Fellows am SFB 1187
Der Sonderforschungsbereich (SFB 1187) „Medien der Kooperation“ begrüßt vier neue Mercator Fellows: Catalina Goanta, Birgit Meyer, Benjamin Peters, Jürgen Streeck und Jill Walker Rettberg. Diese herausragenden Wissenschaftler*innen werden dieses Jahr ihre wissenschaftliche Expertise und innovativen Forschungsansätze in den SFB 1187 einbringen.
Über das Mercator Fellowship am SFB 1187
Um die wissenschaftliche Zusammenarbeit im Forschungsverbund zu stärken, vergibt der SFB 1187 Mercator Fellowships an herausragende Wissenschaftler*innen aus dem In- und Ausland. Mercator Fellows forschen für eine längere Zeit und enger Zusammenarbeit mit einem oder mehreren der am SFB 1187 beteiligten Teilprojekte zu Fragestellungen rund um digitaler, datenintensiver Medien. Zusammen mit den regulären Mitglieder verfolgen unsere Mercator Fellows das gemeinsame Ziel, interdisziplinäre Ansätze weiterzuentwickeln und das Forschungsprogramm des SFB mitzugestalten. Die Aufnahme dieser renommierten Forschenden stärkt nicht nur das internationale Netzwerk des SFB 1187, sondern fördert auch den Wissens- und Ideentransfer, der für die digitale Gegenwartsforschung am SFB von zentraler Bedeutung ist.
Das Mercator Fellowship ist ein Modul im Rahmen der Förderprogramme der Deutschen Forschungsgemeinschaft und dient dazu, einen intensiven und langfristigen Forschungsaustausch zu ermöglichen.
Über die aktuellen Mercator Fellows
Prof. Dr. Catalina Goanta
Law, Economics and Governance Molengraaff Institute for Private Law University of Utrecht, the Netherlands
Catalina Goantas forscht an der Schnittstelle von Recht, Technologie und Gesellschaft mit einem besonderen Fokus auf Plattformregulierung, Content-Monetarisierung und Verbraucherrecht im digitalen Zeitalter. Als Leiterin des von der EU geförderten ERC Starting Grant-Projekts HUMANads (2022–2027) untersucht sie, wie Influencer-Marketing, algorithmische Werbesysteme und neue Formen der digitalen Arbeit rechtlich und gesellschaftlich zu bewerten sind. Neben ihrer akademischen Tätigkeit ist sie international als Expertin für Plattformregulierung gefragt.
Goanta wurde für ihre innovativen Lehr- und Forschungsansätze ausgezeichnet und hat 2017 ein Fellowship am Stanford Transatlantic Technology and LawForum erhalten. 2018 folgte das Niels-Stensen-Stipendium. Ihre Dissertation über die Digitalisierung des Vertragsrechts an der Universität Maastricht legte den Grundstein für ihre intensiven Auseinandersetzungen mit den rechtlichen und gesellschaftlichen Herausforderungen der Plattformökonomie.
Zu ihren wichtigen Publikationen zählen unter anderem der 2020 veröffentlichte Sammelband The Regulation of Social Media Influencers, in dem die Regulierung von Social-Media-Influencern aus verschiedenen Perspektiven analysiert wird und die Herausforderungen des Influencer-Marketings beleuchtet werden, sowie der 2021 veröffentlichte Artikel „A New Order: The Digital Services Act and Consumer Protection“ im European Journal of Risk Regulation, in dem der Digital Services Act (DSA) der EU aus der Perspektive des Verbraucherschutzes und der Haftung von Intermediären untersucht wird.
Prof. Dr. Birgit Meyer
Department of Philosophy and Religious Studies Utrecht University, the Netherlands
Birgit Meyer forscht als Kulturanthropologin seit über 30 Jahren zu Religion aus einer materiellen und postkolonialen Perspektive. Ihre Forschung zielt auf eine Synthese zwischen empirischer Forschung und theoretischer Reflexion in einem breiten multidisziplinären Rahmen an. Zu den Schwerpunkten ihrer Forschung gehören Religion in Afrika, der Aufstieg und die Popularität globaler Pfingstkirchen, Religion, Populärkultur und Kulturerbe, Religion in (post-)kolonialen Kontexten, Religion und Medien, Religion und Öffentlichkeit, religiöse visuelle Kultur sowie Sinne und Ästhetik.
Zu ihren wichtigen Publikationen zählen unter anderem das 2021 veröffentlichte Buch Refugees and Religion: Ethnographic Studies of Global Trajectories, das Religion aus einem materiellen und körperlichen Blickwinkel betrachtet und untersucht, wie Flüchtlinge ihre Religion ausüben und konvertieren oder neue Glaubensrichtungen entwickeln sowie der 2024 veröffentlichte Artikel „‚Idols‘ in the museum: Legacies of missionary iconoclasm“ in dem Sammelband Image Controversies: Contemporary Iconoclasm in Art, Media, and Cultural Heritage, das zeitgenössische Ikonoklasmen in Kunst, Medien und im Umgang mit dem Kulturerbe in globaler und interdisziplinärer Perspektive einer kritischen Analyse unterzieht.
Prof. Dr. Benjamin Peters
Hazel Rogers Endowed Chair in Media Studies University of Tulsa, USA
Benjamin Peters forscht in den Bereichen der Medientheorie, Geschichte neuer Medien, Technologiekritik sowie digitalen Kulturen und die Politik von Informationstechnologien, insbesondere mit Blick auf die Beziehungen zwischen neuen Technologien, Kultur und Gesellschaft und die Geschichte der sowjetischen Informatik. Für seine wissenschaftlichen Arbeiten wurde Benjamin Peters mehrfach ausgezeichnet, darunter der Computer History Museum Prize (2018) für sein Buch How Not to Network a Nation, der Wayne S. Vucinich Book Prize (2017). Für seine herausragende Lehrtätigkeit wurde er 2023 mit dem Outstanding Teaching Award der University of Tulsa ausgezeichnet.
Peters promovierte 2010 an der Columbia University im Fach Kommunikationswissenschaften. Seit 2017 ist er außerordentlicher Professor am Lehrstuhl für Medienwissenschaft an der University of Tulsa, wo er den Hazel Rogers Stiftungslehrstuhl für Media Studies inne hat. Weitere akademische Stationen führten ihn unter anderem als Fellow an die Yale Law School (2015) und an das Kate Hamburg Kolleg der RWTH Aachen (2022–2023) und an das MECS Institute of Advanced Study der Leuphana Universität (2017, 2019). Zudem war er am Berkman Klein Center for Internet & Society der Harvard University und als Gastprofessor an der Hebrew University in Jerusalem tätig.
Jürgen Streeck forscht im Bereich multimodaler Interaktion, insbesondere zur Koordination von Sprache, Gestik und Blicken sowie der sozialen Bedeutung von Handlungen in der Kommunikation. Er hat zur Entwicklung der multimodalen Interaktionsforschung beigetragen und befasst sich mit den Verbindungen zwischen Sprache, Musik und Oralität, insbesondere im Hip-Hop. Für seine wissenschaftlichen Arbeiten wurde er mehrfach ausgezeichnet, darunter mit dem Georg-Gottfried-Gervinus Fellowship (2013–2014). Er war Fellow am Freiburg Institute for Advanced Studies und am Zentrum für interdisziplinäre Forschung (ZiF) in Bielefeld.
Streeck promovierte 1981 an der Freien Universität Berlin in Linguistik und ist seit 2013 Professor für Kommunikationswissenschaft am Department of Communication Studies der University of Texas at Austin. Zuvor war er Associate Professor am gleichen Department und hatte auch eine Professur für Sprachwissenschaften an der Freien Universität Berlin inne. Darüber hinaus nahm er Gastprofessuren und -stipendien an Universitäten wie der Universität Oldenburg, der Universität Wien und der University of Utrecht wahr.
Zu seinen wichtigen Publikationen zählt das 2009 veröffentlichte Buch Gesturecraft: The Manu-facture of Meaning, in dem Streeck untersucht, wie Handgesten in der Kommunikation die Welt darstellen und deuten, basierend auf mikroethnografischer Forschung und Theorien zu Kognition und Interaktion. In dem 2017 herausgegebenen Band Self-Making Man: A Day of Action, Life, and Language analysiert Jürgen Streeck, wie ein Automechaniker in Texas durch Gesten, Sprache und Handlungen seine soziale Welt und Identität in der Kommunikation erschafft.
Prof. Dr. Jill Walker Rettberg
Center for Digital Narrative, Department of Linguistic, Literary and Aesthetic Studies University of Bergen, Norway
Jill Walker Rettberg forscht zu den Wechselwirkungen zwischen Erzählungen und digitalen Technologien, insbesondere die Auswirkungen von Künstlicher Intelligenz auf Erzählweisen und die Verbreitung von Geschichten online. Rettberg hat für ihre Arbeiten Auszeichnungen wie etwa 2017 den John Lovas Memorial Award für ihre innovative Nutzung von Social Media in der Forschung erhalten. Für ihre herausragende Forschungsarbeit wurde sie zudem mit dem Meltzer Foundation Prize for Excellence in Research Dissemination (2006) ausgezeichnet.
Jill Walker Rettberg promovierte 1998 in Informatik an der Universität Bergen. Seit 2000 teilt sie ihre Forschungsergebnisse auf ihrem Blog jill/txt und in den Sozialen Medien und war damit eine der ersten wissenschaftlichen Bloggerinnen. Seit 2014 ist sie Professorin für digitale Kultur und Co-Direktorin des Center for Digital Narrative an der Universität Bergen. Sie leitet das ERC Advanced Grant-Projekt „AI Stories: Narrative Archetypes for Artificial Intelligence“ und das ERC ConsolidatorProjekt „Machine Vision in Everyday Life“. Weitere akademische Stationen führten sie unter anderem als Gastprofessorin an die University of California, Berkeley (2015) und an das MIT Media Lab (2018).
AI Methods: From Probing to Prompting,4-7 February, 2025
The Collaborative Research Center 1187 “Media of Cooperation” organizes the one-week winter school at the University of Siegen and invites graduate students, postdoc researchers, and media studies scholars interested in the intersections of AI methods, digital visual methodologies, visual social media, and platforms. The Winter School aims to explore questions centering on the implications of AI methods for new forms of sense-making and human-machine co-creation. Please register via the registration form until December 15 2025.
As artificial intelligence (AI) technologies rapidly evolve, the ways in which we perceive and process information are fundamentally changing. The shift from computational vision, recognition, and classification to generative AI lies at the core of today’s technological landscape, fueling societal debates across different areas—from open-source intelligence and election security to propaganda, art, activism, and storytelling.
Computer vision, a sophisticated agent of pattern recognition, emerged with the rise of machine learning, sparking critical debates around the fairness of image labelling and the deep-seated biases in training data. Today, models like Midjourney, Stable Diffusion, DALL-E, and more recently, Grok are not just recognizing—they are generating patterns, synthesizing multimodal data from websites, social media, and other online sources to produce oddly familiar and yet captivating results. This shift introduces significant ethical questions: How can we critically repurpose the outputs of AI models that are always rooted in platform infrastructures? Which methodological challenges and creative possibilities arise when the boundaries between context and scale become indistinct? Are patterns and biases all there is? And how about scaling down?
The one-week winter school at the University of Siegen organized by the Collaborative Research Center “Media of Cooperation” invites participants to explore these questions centering on the implications of AI methods for new forms of sense-making and human-machine co-creation. The winter school is practice-based and brings together conceptual inputs, workshops, and sprinted group projects around two collaborative methods: probing and prompting.
Probing involves repurposing AI systems to explore their underlying mechanisms. It is a method of critical interrogation—for example, using specific collections of images as inputs to reveal how contemporary computer vision models process these inputs and generate descriptions. Probing not only serves to problematize the hidden architectures of AI but also allows us to critically assess their different ‘ways of knowing’—how can alternative computer vision features such as web detection or text-in-image recognition help us contextualize and interpret visual data?
On the other hand, prompting refers to the practice of engaging GenAI models through input commands to generate multimodal content. Prompting emphasizes the participatory aspect of AI, framing it as a tool for human-machine co-creation, but it also shows the models’ limitations and inherent tensions. AI-generated creations captivate us, yet they also pose the risk of hallucination or what philosopher Harry Frankfurt might call “bullshit”— statements the models confidently present as facts, regardless of their detachment from reality.
The first day of the Winter School will be hybrid. Project group work will be taking place on site.
Program highlights
Participants will have the opportunity to explore and attune these methods to different research scenarios including tracing the spread of propaganda memes/deepfakes, analyzing AI-generated images, and ‘jailbreaking’ or prompting against platforms’ content policy restrictions. A blend of research practice and critical reflection, the winter school features
a keynote by Jill Walker Rettberg(University of Bergen) on “Qualitiative methods for analysing generative AI: Experiences with machine vision and AI storytelling”
two hands-on workshops on mixed techniques for probing and prompting facilitated by Carlo de Gaetano (Amsterdam University of Applied Sciences), Andrea Benedetti (Density Design, Politecnico di Milano), Elena Pilipets (University of Siegen), and Marloes Geboers (University of Amsterdam)
two project tracks intended to combine AI methods with qualitative approaches and ethical data storytelling.
Track 1 “Fabricating the People: Probing AI Detection for Audio-Visual Content in Turkish TikTok” led by Lena Teigeler and Duygu Karatas (both University of Siegen)
Track 2 “Jail(break)ing: Synthetic Imaginaries of ‘sensitive’ AI” led by Elena Pilipets (University of Siegen) and Marloes Geboers (University of Amsterdam)
Track I: Fabricating the People: Probing AI Detection for Audio-Visual Content in Turkish TikTok
Lena Teigeler & Duygu Karatas
Several brutal femicides in Türkiye in 2024 led to a wave of outrage, showing in protests both on the streets and on social media. The protesters demand the protection of women against male violence, measures against offenders and criticize the government under Recep Tayyip Erdoğan for not standing up for women’s rights, as demonstrated, for example, by Türkiye’s withdrawal from the Istanbul Convention in 2021. One of the cases leading to the protest was allegedly connected to the Turkish “manosphere” and online “incel” community. The manosphere is an informal online network of blogs, forums, and social media communities focused on men’s issues, often promoting views on masculinity, gender roles, and relationships. At the core of these groups often lie misogynistic, and anti-feminist views. Many groups foster toxic attitudes toward women and marginalized groups. Incels, short for „involuntary celibates,“ are one subgroup belonging to the broader manosphere, formed by men who feel unable to form romantic or sexual relationships despite wanting them, often blaming society or women for their frustrations.
The project investigates how the cases of femicide are discussed and negotiated in Turkish TikTok by protesters and within the manosphere and explores how these videos make use of generative AI. The use of AI in video creation can range from entire scene generation, over the creation of sounds or deepfaking, to editing and stylisation. The project takes a sample of TikToks associated with the recent wave of femicides as the starting point and makes use of AI methods for two purposes: 1) To detect the usage of generative AI within a sample of TikToks with the help of image labeling. This can range from fully-generated images, videos or sound, to the usage of tools and techniques used within the creation and editing process. We compare different models for detection purposes. 2) With the help of Web Detection, we trace the spread of videos and images across platform borders and content elements that are assembled or synthesized within TikToks.
The aim of the project is to create a cartography of AI based methods for the investigation of audio-visual content. It is part of the DFG-funded research project “Fabricating the People – negotiation of claims to representation in Turkish social media in the context of generative AI”.
Track II: Track 2 Jail(break)ing: Synthetic Imaginaries of ’sensitive‘ AI
Elena Pilipets & Marloes Geboers
The rapid evolution of AI technology is pushing the boundaries of ethical AI use. Newer models like Grok-2 diverge from traditional, more restrained approaches, raising concerns about biases, moderation, and societal impact. This track explores how three generative AI models—X’s Grok-2, Open AI’s GPT4o, and Microsoft’s Copilot—reimagine controversial content according to—or pushing against—the platforms‘ content policy restrictions. To better understand each model’s response to sensitive prompts, we use a derivative approach: starting with images as inputs, we generate stories around them that guide the creation of new, story-based image outputs. In the process, we employ iterative prompting that blends “jailbreaking”—eliciting responses the model would typically avoid—with “jailing,” or reinforcing platform-imposed constraints.Jail(break)ing, then, exposes the skewed imaginaries inscribed in the models‘ capacity to synthesize compliant outputs: The more iterations it takes to generate a new image the stronger the latent spaces of generative models come to the fore that lay bare the platforms‘ data-informed structures of reasoning.
Addressing the performative nature of automated perception, the track, facilitated by Elena Pilipets and Marloes Geboers, examines six image formations collected from social media, which then were used as prompts to explore six issues: war, memes, art, protest, porn, synthetics.In line with feminist approaches, we attend specifically to the hierarchies of power and (in)visibility perpetuated by GenAI, asking: Which synthetic imaginaries emerge from various issue contexts and what do these imaginaries reveal about the model’s ways of seeing? To which extent can we repurpose generative AI as a storytelling and tagging device? How do different models classify sensitive and ambiguous images (along the trajectories of content, aesthetics, and stance)?
Facilitators will combine situated digital methods with experimental data visualization techniques tapping into the generative capacities of different AI models. The fabrication and collective interpretation of data with particular attention to the transitions between inputs and outputs will guide our exploration throughout. Participants will learn how to:
Conduct “keyword-in-context” analysis of AI-generated stories to identify patterns or “formulas” within issue-specific imaginaries (where, who/what, and how).
Perform network analysis of AI-generated tags, where input keywords are tags for the original images and output keywords are tags for AI-regenerated images.
Design prompts to generate canvases that synthesize vernaculars of different transformer models.
The project builds on our earlier work, developing ethnographic approaches to explore cross-model assemblages of algorithmic processes, training datasets, and latent spaces.
Registration
Please register via the form above until December 15. Your registration will be confirmed by December 20, 2024. Participation is limited to 20 people.
Venue
University of Siegen
Campus Herrengarten
Herrengarten 3
room: AH-A 125
57072 Siegen
Das in der zweiten Förderphase eingerichtete Modul „Integriertes Graduiertenkolleg“ (MGK) hat sich zum zentralen Ort der Promovierendenausbildung entwickelt und trägt maßgeblich zum übergeordneten Ziel des SFB bei, digitale Grundlagenforschung und innovative Methodenentwicklung zu leisten. Durch universitätsfinanzierte Promotionsstellen und Stipendien konnte das MGK das Themenspektrum der Teilprojekte erweitern und maßgeblich zur Entwicklung einer digitalen Praxeologie beitragen.
Auch in der dritten Förderphase bildet das MGK den Ort der international orientierten, interdisziplinären Promovierendenausbildung, die quer zu den Teilprojekten des SFB verläuft. Inhaltlich schließt das MGK an den Anspruch des SFBs an, Kooperation in sensorintensiven Umgebungen zu untersuchen. Es verfolgt drei zentrale Ziele: (1) die Analyse von Medien-, Daten- und sensorischen Praktiken konzeptionell und methodisch zu strukturieren, (2) die Rolle von Sensoren im Kooperationsprozess im Spannungsfeld von Öffentlichkeiten und Infrastrukturen zu untersuchen und (3) Methodeninnovation zu realisieren. Der Einsatz von Sensoren erweitert die Dimensionen, die zur wechselseitigen Vollzugswirklichkeit von Medien beitragen: Neben Infrastrukturen und Öffentlichkeiten rücken Umgebungen und Körper in den Fokus.
Das MGK nutzt die in der zweiten Phase erfolgreich erprobten Kooperations- und Veranstaltungsformate, um diese Debatten quer zu den Teilprojekten und den involvierten Disziplinen zu führen und neue thematische Impulse zu generieren. Promovierende werden so befähigt, relevante Fähigkeiten für ihre Karriere in der interdisziplinären Medienforschung zu erlangen. Das MGK organisiert dazu Veranstaltungen, Trainings, Postdoc-Mentorships und Publikationen, die sowohl den Kooperations- als auch den Datenbegriff im Kontext von Sensordaten weiterentwickeln. Es kooperiert dafür mit den universitätseigenen Institutionen zur Graduiertenausbildung, wie dem House of Young Talents, dem Welcome Center für internationale Forschende und dem Women Career Service. Das Ausbildungsprogramm ist eng mit der Internationalisierungsstrategie des SFB verknüpft, indem internationale Promovierende als Stipendiat*innen eingebunden und internationale Fellows und Gäste systematisch in die Ausbildung integriert werden.
Das MGK ist der zentrale Ort der Methodenvermittlung und -innovation. Durch profilbildende Formate wie das Research Tech Labs, internationale Summer Schools, interdisziplinäre Kleingruppen, die Meisterklasse für Medienethnographie, die Werkstatt Medienpraxistheorie, Data Labs sowie gemeinsame Publikationen werden die Grundlagen praxeologischer, digitaler und sensorbasierter Methoden vermittelt. Das Methodentraining des MGK schafft Raum für die Entwicklung von InventiveMethodologies, die auf die spezifischen Herausforderungen sensorbasierter empirischer Felder und Gegenstände eingehen, welche es in ihren infrastrukturelle Verteiltheit, in ihren öffentlichen Kontroversen aber auch in ihrer kooperativ-körperlichen Sensorik zu untersuchen gilt.
Ein PI-Konsortium mit Expertisen in Medien- und Kommunikationswissenschaften, Digital Sociology and Methods, Angewandte Linguistik und Human Computer Interaction begleitet die interdisziplinären Promotionen.
Kooperation mit externen Promovierenden durch das Stipendienprogramm.
Das Veranstaltungsprogramm vermittelt theoretische und methodische Grundlagen der digitalen Medienpraxisforschung
Kolloquien & Werkstatt Medienpraxistheorie
Interdisziplinäre Kleingruppen & Schreibretreats
Research Tech Labs &
Data Labs
Internationale Sommerschulen
Breites Unterstützungsnetzwerk
Internes Mentoring
Internationales Netzwerk an Mercator-Fellows und Kooperationspartner*innen