Collective intelligence vs artificial intelligence
Beyond the hype, generative AI is reshaping Wikipedia. The data tells the story: 2025 saw an 8% drop in human traffic alongside a 50% surge in bot activity. AI isn’t merely consuming or reading Wikipedia content—it’s replacing the online encyclopedia as an interface to knowledge. How these trends will develop remains unclear, but Wikipedia faces significant changes and uncertainty.
To explore this challenge, Wikimedia CH organized a roundtable in November 2025, collaborating with IMD Business School in Lausanne and Open Future. The event aimed to discuss these questions and develop an informed response to the rise of generative AI within the online encyclopedia. It was built on two central considerations:
- The debate should extend beyond the use of AI tools for Wikipedia editing.
- Experts from outside the Wikimedia Movement should participate: data scientists, AI developers, journalists, and researchers.
Insights from the discussion have been compiled in a new report available on Wikimedia Commons. This marks a further step in a Wikimedia CH initiative planned for 2026 that aims to define a “Wikimedia and AI” mission by developing context and forecasts, creating a roadmap for technological development, and providing information to support informed AI governance decisions.
The report frames the current moment as a “Peak Wikipedia” turning point. While Wikipedia is widely used by both humans and machines, AI services increasingly provide answers users previously got from Wikipedia directly. This risks making Wikipedia an “invisible layer”—heavily used as training data but less visited by people, affecting editors, fundraising, and sustainability.
The publication uses “collective intelligence” (Pierre Lévy)—human-driven capacity from collaboration, diversity, and transparency—to contrast with centralized, opaque AI systems. This creates tension between human-governed knowledge commons and AI systems reusing its outputs at scale.
The report analyzes the “new knowledge loop”: declining human visits weaken the feedback loop for editors and fundraising, while machine traffic pressures infrastructure and reshapes knowledge access. Wikimedia needs an ecosystem approach—working with other knowledge organizations, media, libraries, and open-source AI efforts—to actively shape standards and governance rather than remain a passive data source.
Key insights explored:
- Wikipedia’s knowledge production at a crossroads—as a data commons, Wikipedia offers opportunities for collectively governed data pipelines that include both human and synthetic data approaches.
- Democracy and information integrity in the AI age—Wikimedia’s growing role as a source of verifiable “ground truth.”
- The paradox of openness in knowledge commons—ensuring sustainability while maintaining openness.
- The emerging “knowledge loop”—where AI mediates access to knowledge, widening the gap between public knowledge and the expanding “private truth” held by commercial platforms.
To learn more and stay updated, visit the project’s website. The next phase will explore what actions could follow and how investments can reinforce Wikimedia’s principles in the emerging AI-driven knowledge ecosystem. A forthcoming white paper will focus on this transition—from understanding to action.
