Stochastic Parrots and Beyond: Normative Dimensions of Generative AI

Aims

Generative AI (GenAI) systems are increasingly applied in academic research, yet their environmental and social costs as well as normative dimensions (discriminatory representations, labor exploitation, and governance gaps) are rarely examined holistically. This one-day workshop equips doctoral candidates and postdocs with analytical frameworks and practical tools to assess the risks and responsibilities entailed in using GenAI in their own research contexts.

After completing the course, participants will be able to:

• Analyze GenAI systems through different analytical lenses, considering both material costs such as environmental impacts and labor conditions and normative questions such as justice, accountability, and governance.
• Evaluate risks across the lifecycle of GenAI models, including design choices, data, training, deployment, evaluation, and governance, and identify normative issues at each stage.
• Apply practical tools for responsible AI use, including risk mapping, documentation and disclosure heuristics, evaluation checklists, and governance questions, to their own research contexts.
• Formulate and defend a position on the responsible use of GenAI in their research, grounding their arguments in established theoretical frameworks and governance principles such as the EU AI Act.

Content

The course is structured around four analytical lenses followed by collaborative exercise using the Panarchy framework of thinking in social-ecological systems:

1. The "Stochastic Parrots" Lens: Scale & Environment examines why “bigger” is not automatically “better” in GenAI systems. Participants explore environmental costs such as energy consumption, water use, and rare earth extraction, alongside questions of data provenance, consent, and dataset curation practices.
2. The Representation & "Pygmalion" Lens explores GenAI models with a focus on visual culture and harmful stereotyping in multimodal datasets as well as resulting AI-facilitated harms.
3. The Lifecycle & Labor Lens: Materiality & Extraction traces the material and human infrastructures and costs behind AI systems, mapping the lifecycle from design to deployment: Invisible labor (gig work, RLHF, "fauxtomation"); data politics: consent, copyright, provenance.
4. The Political Economy Lens: Power & Markets situates GenAI within its broader market and governance structures. Topics include tensions between innovation and safety, platformization and commercial incentives and lastly, co-emerging governance frameworks such as the EU AI Act and institutional research policies.

The course concludes with a hands-on workshop in which participants apply these perspectives in a practical, collaborative exercise. Using the Panarchy and Ecocycle frameworks, they map how GenAI is shaped by interacting system levels, from individual research practices and institutional norms to industry dynamics and regulation. By visualizing these interdependencies, participants identify obstacles and opportunities for change towards more responsible GenAI use. This systems-level exercise brings together insights from the four lenses and helps participants translate critical analysis into concrete strategies and first action steps for more responsible AI practices in their own research.

Methods

The workshop is divided into two parts. In the first half of the workshop, the instructors introduce GenAI from a holistic perspective using four analytical lenses. Short input sessions are combined with guided discussion and case-based analysis, and participants collaboratively map how GenAI systems are shaped by interacting system levels using the Panarchy framework. Afterwards, participants reflect on the implications for their own research contexts and responsible GenAI use practices. Working in small groups, they use their established Panarchy charts to identify obstacles and opportunities for more responsible AI use in their own research.

Target Group

All Doctoral Candidates & Postdocs

About the Trainer

Dr. Moritz Mähr (Dr. sc. ETH Zurich) is an associate researcher in digital humanities at the University of Bern and an information and library science specialist with Research Analytics Services at ETH Zurich. Until 2025, he was project manager of Stadt.Geschichte.Basel at the University of Basel and a visiting research fellow at the Luxembourg Centre for Contemporary and Digital History (C²DH), University of Luxembourg. His research bridges digital history, science and technology studies, and open research infrastructure, with focus areas including digital source criticism in the age of AI, the history of digitization in public administration, minimal-computing approaches to public history, and digital sustainability grounded in FAIR and CARE principles. He studied history, computer science, and banking and finance in Zurich and Berlin, and received his doctorate from ETH Zurich for his dissertation on the digitization of Swiss migration authorities in the 1960s.

Laura Wagner is a PhD candidate at the University of Zurich within the project "The Canon of Latent Spaces - How Large AI Models Encode Art and Culture" (funded by the Swiss National Science Foundation, led by Eva Cetinic). Her doctoral research examines how open-source text-to-image and video GenAI models are adopted, personalized, and shared across online communities, with a focus on how systemic issues, including unlicensed or violent training data, resurface in personalized model outputs. She holds a Master's degree in Integrated Design from KISD Köln International School of Design, where she subsequently worked on the KITEGG project developing AI-integrated curricula for design education and co-headed the Living Objects Lab, a Lab for prototyping with embedded electronics at KISD.

Workload

Preparation: 8h
Course attendance: 8h
Follow-up: 14h

Feature

Once registration is open, applications will be collected for 24 hours and course places allocated by lot. All registrations received after the initial 24h period will be put on a waiting list and assigned on a first come, first served basis.

Course places/places on the waiting list will be confirmed by e-mail. Course registrations can only be canceled before the registration period ends (send an e-mail to grace@unibas.ch). Full course attendance is mandatory. Participants who fail to attend a course without prior notification or withdraw after the registration deadline are subject to a fee of CHF 30. In addition, participants who cancel their course registration at a later point in time, are absent without an excuse or do not attend the entire course will, for reasons of fairness, not be considered for course registration in the following semester and will be removed from other courses offered in the same semester. Please find the detailed regulations on the Transferable Skills Homepage.