Lecture 1 — Introduction & IOS (Van Rooij)
Paper: Elliott, K., Price, R., Shaw, P., et al. (2021). Towards an equitable digital society: Artificial Intelligence (AI) and Corporate Digital Responsibility (CDR). Society, 58, 179–188.
Course framing (likely exam-relevant)
Central question: How can we apply AI methods, techniques and lessons to contribute to the formation and flourishing of open societies? And conversely: how do developments in AI shape the possibilities and threats to open societies?
Course philosophy: Not "how does this AI technique work in detail?" — instead "what types of AI exist, how are they used, and which societal problems do they address?" The emphasis is on integration of social/behavioural sciences, humanities, and governance with AI.
Three lecture types (color-coded in the schedule): - Methodological (blue, lectures 2–4): tools/models to study AI's societal impact (cognitive models of decision making, autonomy/HCI, collective-pattern models like bounded confidence) - Thematic (orange, lectures 5–7): examples of AI impacting society + how impacts can be mitigated (linguistic models, trust, medical AI) - Practical (green, post-midterm): workgroups for the grant proposal
Key concept 1 — Open Society
An open society is based on:
- Openness to diversity of knowledge (science, arts, cultural)
- Openness to emancipatory movements & respect for individual rights
- Constitutional democracy & rule of law
- Contestable markets & open borders
Source for the definition: IOS Position Paper by Avelino, Castaldi, Claassen, Gerbrandy (UU).
Why this matters for AI: every societal issue AI might address (or worsen) — distrust in institutions, the power of big tech, climate, migration, geopolitical upheaval, pressures on democracy/rule-of-law/free-speech — can be reframed as a threat to one of these four pillars. Essay answers benefit from explicitly naming which pillar(s) an AI application threatens or supports.
Key concept 2 — Institutions
"Institutions are the building blocks of society. The 'rules of the game' — the written rules and associated organisations, but also unwritten rules & networks."
This is the classic North-ian definition (formal + informal institutions). The course title "AI for an Open Society" is funded by UU's strategic theme "Institutions for Open Societies" (IOS) — both formal institutions (parliament, courts, regulators) and informal ones (norms, expectations, conventions) shape and are shaped by AI.
Formal vs. informal in exam terms: - Formal: laws, regulations, organisations (e.g., GDPR, the EU AI Act, courts, the NWO) - Informal: norms, conventions, social expectations (e.g., what users expect from a chatbot, professional norms in journalism, group consensus around what counts as "misinformation")
When the manual says learning goal 2 is "analyze and evaluate concrete problems related to formal or informal institutions" — that's the formal/informal split they mean.
Key concept 3 — The IOS framework: 3 pillars, 15 platforms
UU's "Institutions for Open Societies" theme combines 700+ researchers across 4 faculties (Social/Behavioural Sciences, Humanities, REBO/Law–Economics–Governance, Geosciences). It is structured as three intersecting pillars, each containing several research platforms (15 total):
Pillar A — Democracy and Good Governance
- Contesting governance
- EU Platform
- Futures of democracy
- Markets and corporations
- Openness challenged: the university at risk?
- Security in open societies
Pillar B — Transitions and Wellbeing
- Behaviour and institutions
- Bottom-up initiatives for societal change
- Fair transitions
- Future of work
- Longtermism and institutional change
- The transactional state as an institution for good
Pillar C — Equity and Diversity
- Gender, diversity and global justice
- (In)equality
- Open cities
Exam tip: if an essay question asks how an AI application is "relevant to IOS" or "to open societies," picking a specific platform from this list and arguing in its terms is a stronger answer than gesturing at "society" generically.
Paper 1 — Elliott et al. (2021): AI and Corporate Digital Responsibility (CDR)
Core question (quoted in the lecture)
"If we permit AI to make life-changing decisions, what are the opportunity costs, data trade-offs, and implications for social, economic, technical, legal, and environmental systems?"
Key construct: the TRUST framework
CDR rests on trust, which the authors decompose into 5 components:
| Letter | Concept | Plain meaning |
|---|---|---|
| T | Transparency | Be clear about what you're doing and why — both the opportunities and the risks |
| R | Responsibility | Be reputable and accountable for what you do |
| U | Understanding | Provide services so customers understand the outcomes they can expect and how it will affect them and society |
| S | Stewardship | Be a good custodian of the data, in line with the kind of society we want to create |
| T | Truth | Validate accuracy of data; ensure insights/inferences are beneficial and not harmful |
Three central CDR tenets (Venn diagram in slides)
- Promoting Economic Transparency
- Promoting Societal Wellbeing
- Reducing Tech Impact on Environment
Intersections labelled in the slides: - Economic ∩ Societal → Fair & Equitable Access for All Society - All three centre → Purpose & TRUST - Societal ∩ Environmental → Promoting a Sustainable Planet to Live - Economic ∩ Environmental → Invest in the New Eco-Economy
Why CDR matters as a course frame
- Shifts the conversation from "AI ethics" as a philosophical/technical problem to a corporate governance problem: companies, not just regulators, bear an obligation.
- Connects directly to the IOS pillars: TRUST/transparency speaks to Democracy & Good Governance; equitable access speaks to Equity & Diversity; sustainability speaks to Transitions & Wellbeing. Expect questions that ask you to map CDR onto IOS or vice versa.
Likely essay-question angles
- "Apply the TRUST framework to [some AI system, e.g. an algorithmic hiring tool / a medical AI / a social-media recommender]. Which components are most at risk?"
- "CDR places responsibility on corporations rather than regulators. Argue for or against this in the context of an open society."
- "How does Elliott et al.'s CDR framework relate to the IOS framing of formal vs. informal institutions?"
Quick self-test
- Name the four bases of an open society.
- Distinguish formal vs. informal institutions; give one AI-relevant example of each.
- What are the three IOS pillars, and which one would "AI-driven misinformation" most naturally sit in?
- Spell out TRUST and give a one-line gloss for each letter.
- What is the corporate in "Corporate Digital Responsibility"? Why is the corporate level the locus rather than the state or the individual?