Wikimedia CH/Innovation/CI vs AI
AI Roundtable – “Collective Intelligence vs Artificial Intelligence”
Overview
[edit | edit source]Collective Intelligence vs Artificial Intelligence is a Wikimedia CH initiative addressing how generative AI transforms the Wikimedia ecosystem and the wider field of open knowledge.
The project aims to build an informed response to the rise of AI by:
- Examining its impact on Wikimedia’s mission to provide free, reliable and human-curated knowledge;
- Bringing together Wikimedians and experts in the knowledge commons, digital economy, and information ecosystem;
- Producing a white paper that will help shape the Wikimedia Movement’s shared position on AI.
The Lausanne Roundtable
[edit | edit source]
The first convening of the project took place at the IMD Business School (Lausanne, Switzerland) on 4 November 2025 as a one-day, in-person roundtable. It was organized in cooperation by Wikimedia CH, IMD Business School and Open Future. This meeting formed the foundation for the upcoming white paper and collaborative strategy.
- Practical information
- Date: 4 November 2025
- Location: IMD, Lausanne, Switzerland
- Format: One-day, in-person roundtable
Focus of the Roundtable
[edit | edit source]
The Lausanne convening will be dedicated to understanding the “New Information Loop” — analysing how the rise of AI is reshaping the information ecosystem and how Wikimedia projects are both impacted by and contributing to these transformations.
Topics include:
- AI training on open knowledge platforms: How large-scale use of collaborative content affects the sustainability of the knowledge commons.
- Information ecosystem disruption: How AI changes the discovery, sharing, and validation of knowledge on the web.
- Human vs machine distinction: As a guiding principle for responsible AI policies.
- Shifts in the open web: Changing norms and standards around content sharing and the emergence of new intermediaries (AI platforms, cloud services, etc.).
- Intersection of collective intelligence and AI: Exploring how peer production and machine assistance can jointly shape the future of knowledge.
A detailed agenda will be shared closer to the event. Selected participants will be invited to offer short introductory remarks during thematic sessions.
Why Collective Intelligence?
[edit | edit source]Collective Intelligence is the shared cognitive capacity that emerges when individuals collaborate, pool knowledge or coordinate efforts to solve problems and make decisions. It is human-driven, relies on collaborative decision-making, draws from diverse knowledge sources, operates through decentralized processing, and benefits from error correction through diversity.
The concept represents an alternative approach to the automation of knowledge work. In an era where AI rapidly reshapes how knowledge is produced, validated, and distributed, it is essential to define how collective intelligence can benefit from AI tools while preserving human agency.
Our goal is to ensure that:
- AI supports — rather than replaces — collaborative knowledge creation;
- Wikimedia remains a trusted source of original, human-curated knowledge;
- The movement plays an active role in shaping open, ethical, and sustainable AI ecosystems.
Next steps
[edit | edit source]The Lausanne roundtable will generate the foundation to:
- Explore the evolving balance between AI-driven tools and community-based knowledge production
- Identify opportunities and risks for Wikimedia projects in an AI-mediated information landscape
- Define actionable principles for Collective Intelligence in open knowledge ecosystems
- Launch a White Paper and potential working group for future innovation tracks
Timeline
[edit | edit source]| Phase | Period | Status |
|---|---|---|
| Preparatory phase | August–October 2025 | ✅ confirmed |
| AI Roundtable @ IMD Lausanne | 4 November 2025 | ✅ confirmed |
| White Paper drafting | December 2025-March 2026 | 🕓 ongoing |
| Online consultation & follow-up meeting | Q2 2026 | ❌ planned |
| Publication of final report | Wikimania 2026 | ❌ planned |
Participants
[edit | edit source]The event gathers around 25 invited participants from:
- Academia (IMD, EPFL, ETH Zurich, Swiss AI initiative, etc.)
- Civil society and non-profits (Open Future Foundation, Wikimedia CH, Wikimedia Deutschland, etc.)
- Industry and research initiatives (Mozilla, Thomas Reuters, etc.)
- Wikipedia community
Deliverables
[edit | edit source]- Collective Intelligence vs Artificial Intelligence - asynthesis of key insights from the roundtable
- White paper on AI & Collective Intelligence (first half of 2026)
- Recommendations for Wikimedia communities
Summary of the report
[edit | edit source]This report captures the main insights from the Lausanne roundtable “Collective Intelligence vs Artificial Intelligence” (4 November 2025), co-organised by Wikimedia CH, Open Future and IMD Business School. The roundtable addressed a simple but urgent question: what happens to the Wikimedia Movement when AI stops merely reading Wikipedia and starts replacing it as a key interface to knowledge?
The report frames the current moment as a “Peak Wikipedia” turning point. Wikipedia has never been so widely used—by humans and at massive scale by machines—yet there are early signs of disintermediation: AI services increasingly provide answers that users previously obtained by visiting Wikipedia directly. In this scenario, Wikipedia risks becoming an “invisible layer” of the AI ecosystem: heavily used as training data and a live endpoint, but less directly visited by people, with consequences for editors, fundraising, and long-term sustainability.
To understand what is at stake, the report uses the lens of “collective intelligence” (Pierre Lévy): the human-driven capacity that emerges from collaboration, diversity, deliberation, transparency and accountability. This is contrasted with how today’s generative AI systems are often developed and governed—more centralised and opaque—creating a tension between a human-governed knowledge commons and AI systems that can reuse its outputs at unprecedented scale.
A central section analyses the emerging “new knowledge loop”: decreasing human visits weaken the feedback loop that turns readers into editors and supports fundraising, while machine traffic and AI intermediaries increase pressure on Wikimedia’s infrastructure and reshape how knowledge is accessed. The report argues that Wikimedia needs an ecosystem perspective—working with other knowledge organisations, media, libraries and open-source/public AI efforts—to decide whether it remains a passive data source or becomes an active shaper of standards, interfaces and governance norms in the new architecture.
The report then explores four key challenge areas in more detail:
- the “paradoxes of openness” and how to ensure sustainability without abandoning openness;
- Wikipedia as a data commons and opportunities for collectively governed data pipelines (including human and synthetic data approaches);
- the growing divergence between public knowledge and expanding “private truth” held by commercial platforms; and
- implications for democracy and information integrity, where Wikimedia’s role as a source of verifiable “ground truth” becomes even more important.
To be clear: open, non-discriminatory access for people to Wikimedia content is not in question. The report focuses on how large-scale machine use can be governed so that the commons remains sustainable and accountable, and so that the collaborative vision of Wikimedia is represented across the AI lifecycle. It concludes with a call for urgency and coordinated action, and outlines next steps: Wikimedia CH will publish a white paper in 2026 proposing a “Wikipedia in the era of AI” mission and strategy, including a shared policy position on AI governance, a funding proposal, and a vision for technical elements and tools (a “Wiki AI stack”).
Materials
[edit | edit source]Contacts
[edit | edit source]- Ilario Valdelli, Innovation Programme Lead – Wikimedia CH
- Alek Tarkowski, Director – Open Future Foundation
- IMD Lausanne, institutional partner