Course description
Summary, audience, objectives, and the four lenses
Summary
Generative AI is used increasingly in research, yet its environmental costs, training data, labor conditions, discriminatory representations, commercial incentives, and governance structures are often considered separately. This workshop combines four critical lenses with practical risk-analysis tools. Participants investigate the design, data, training, deployment, evaluation, and governance of text- and image-generating systems; then use two Panarchy blocks to identify feasible changes in their own research environments. Each participant leaves with an individual Action Plan. The aim is a defensible, transparent approach to responsible generative AI use rather than a universal list of permitted and forbidden tools.
Audience
The workshop is for doctoral candidates and postdocs from all disciplines. Each edition sets its own eligibility, capacity, and credits. The edition pages give these facts. No programming experience is required.
Learning outcomes
After the course, participants should be able to:
- Analyze environmental, social, cultural, and institutional issues in generative AI through four distinct analytical lenses.
- Identify normative risks at design → data → training → deployment → evaluation → governance stages, including affected people and potential mitigation.
- Apply the guiding questions of the four lenses, together with peers, to their own Research Use: one actual or planned use of generative AI in their research.
- Identify opportunities and barriers for change across interconnected levels, from their own research project to regulation and society.
- Formulate and defend a context-specific position on responsible research use of generative AI, informed by scholarship and governance frameworks (including the EU AI Act as one example), and plan a next step in an individual Action Plan.
The four lenses
The four lenses give the critical vocabulary. Each lens has its own lesson page, slide deck, documented case, and four guiding questions. Each lens takes 15 minutes in the morning. The participants then apply the guiding questions to their own Research Uses in Lens groups.
| Lens | Essential teaching points |
|---|---|
| 1. Stochastic Parrots: Scale and Environment | Scale is not a sufficient criterion for quality or social benefit. Energy, water, minerals, and infrastructure. Data sourcing, scraping, consent, filtering, and exclusions. |
| 2. Representation and Displacement | Text-to-image datasets. Cultural categories, aesthetics, stereotypes, and representational harm. Humanizing models while dehumanizing women. Non-consensual synthetic imagery. |
| 3. Lifecycle and Labor: Materiality and Extraction | The model lifecycle and the people it affects. Data labeling, content moderation, RLHF, and gig work. “Fauxtomation.” Copyright, consent, provenance, and invisible labor. |
| 4. Political Economy: Power and Markets | Concentrated infrastructure and platform power. Incentives for safety and release decisions. Open versus proprietary claims. Procurement, policy, and the EU AI Act. |
Lens groups and Panarchy
The course day centers on the participants’ own Research Uses. Each participant describes one actual or planned use of generative AI in their research on a Research Use Card. The term Research Use stays the same all day.
In the morning, participants form Lens groups of four. Each group applies the four guiding questions of one lens to the Research Use of each member. A spokesperson reports the conceptual insights to the plenary.
Panarchy is the central method of the afternoon. It is not an optional extra. Each participant maps how changes at fast, local levels and at slow, institutional levels influence their own challenge. The common question is:
What changes across interconnected levels would make generative AI use in academic research more responsible?
In Block 1, groups of four select one challenge and consolidate a group Action Plan. In Block 2, pairs examine the open challenges, so that every challenge receives an in-depth discussion. At the end of the day, each participant writes an individual Action Plan, and a peer reviews it.
Format
The course uses short lectures, sourced cases, collaborative analysis of the participants’ Research Uses in Lens groups, two Panarchy blocks, and an individual Action Plan with peer review. Paper, digital documents, and flipcharts are enough.
Bring these items:
- A laptop.
- Your Research Use: a short description of one actual or planned use of generative AI in your research. The description must not be confidential.
- Do not disclose protected research data. Use a description, not the data.
- Do not create a commercial AI account. No mandatory exercise needs one.
The website is in English. The language of instruction is not stated on this site. Ask the course administration of your edition if you need to know it.
What the course is not
The course does not give a list of permitted and forbidden tools. It does not replace the rules of your university, your faculty, or your funder. It gives you questions, evidence, and methods, so that you can make and defend your own decision.