The Knowledge Commons in the Age of AI: Keynote speech at WikiConference India 2026

Written version of the opening keynote delivered at WikiConference India 2026 (WCI 26) by Tanveer Hasan, Open Knowledge Initiatives, India.

I want to begin by acknowledging who is in this room. We are the people who build, maintain, and govern the most remarkable knowledge infrastructure humanity has ever created. We do this work not for profit or acclaim but because of our belief in the principle of free knowledge for everyone.

Keynotes are either information-heavy or seen as inspirational talks. I am sorry to disappoint you: this one is neither. What this is, is a collaborative discussion that I am presenting on behalf of all the organisers, the programme team and my team members. I am the voice of the talk, but the thinking is truly shared and collaborative. My goal is not to inspire you. You are already inspired, and that is why the scholarship team had such a difficult job in making its decisions. My intention is to share some data, to offer a framework for thinking about our challenges, and to ask some difficult questions that I believe we, as a community, must answer together.

I want to begin with a question that I believe we need to grapple with if we are serious about the future of the knowledge commons.

What happens when the knowledge from hundreds of languages from India and the Global Majority becomes visible and connected? This, I feel, is the question that will define our work for the next decade.

As acknowledged earlier, we have built something remarkable. We have the trust of millions of people and the most extraordinary knowledge commons the world has ever seen.

Unfortunately, we are now forced to rely on platforms that distribute our content but do not sustain our commons. We rely on corporations that extract our labour but do not reciprocate.

Many people, even around us, either do not know about the Wikimedia ecosystem or think that we are “Google”. How do we explain to people that our movement is not simply a collection of websites? It is a community-governed knowledge infrastructure.

When we say infrastructure, I mean something like a power grid. It is the underlying system that enables other things to happen. The power grid is an enabler; it catalyses a lot of things that are essential for our everyday life. What we are building with Wikipedia, Wikidata, Wikimedia Commons, and Wikisource is the knowledge infrastructure for the entire digital age. Most other infrastructure is either state-owned or individually controlled; this is where we stand unique. We are community-governed. Many platforms are built for profit or control.

Ours is built on the principle of shared labour for the greater good. It is maintained not by a corporation but by a global network of volunteers who decide together what the rules are and how the system should work.

Let us begin by asking the invisible questions that WCI 26 wants to make visible.

1. What is the Knowledge Commons?

No, it is not a project on Wikimedia Commons.

The concept of the commons refers to resources that are owned and managed collectively by a community, not by the state or the market. The knowledge commons, specifically, refers to an institutional approach to governing the production, use, management, and preservation of knowledge as a shared resource.¹ This framework draws on the work of Elinor Ostrom, who won a Nobel Prize for her research on how communities successfully manage common resources.² Ostrom showed that commons governance works when communities establish clear boundaries, make collective decisions, monitor use, and apply graduated sanctions for rule-breakers.³

David Bollier, a leading commons scholar, has extended this thinking. He argues that we should see commons not primarily as creatures of property law and standard economics but as relational social systems. Commons arise when groups of people decide to manage resources collectively, with a special regard for equitable access and long-term stewardship.⁴

What does this mean for us? It means the Wikimedia projects are not just websites. They are a global knowledge commons. They represent one of the most ambitious experiments in collective governance the world has ever seen. But, as mentioned earlier, a commons is not self-sustaining. It requires constant attention, active governance, and a shared ethics of care.

2. How do we understand the Knowledge Commons in this fast-evolving landscape?

Researchers say that with generative AI, we are in a third knowledge loop.⁵ The first loop was built around search engines, which brought readers to Wikipedia. The second loop was shaped by social media, which distributed Wikipedia content across platforms. The third loop, driven by AI, is fundamentally and dangerously different. The way knowledge is created, used, and shared is undergoing a fundamental transformation.

In this new loop, AI systems consume the knowledge commons at a massive scale. They use our content, our labour, and our community’s work to train their models. But unlike search engines or social media, they do not bring readers back to our projects. They do not create new contributors. They do not acknowledge or reciprocate. This creates a sustainability challenge.

Our work powers the AI revolution, but we are increasingly invisible within it. There is a risk of Wikimedia projects becoming a “dark commons”: widely used yet lacking agency, an unseen entity feeding proprietary systems.⁵

How do we respond to this? We cannot close or restrict access; that would be a betrayal of our mission. We cannot passively accept extraction. We need a more active position on generating value, by contributing to the discourse and governance around these systems.⁵

3. Is India / South Asia / the Global Majority the new frontier of discovery and investment?

It is important to remember and reflect that the current AI revolution is built on a foundation that is profoundly English-centric. The large language models that power generative AI are trained predominantly on English-language data. An estimated 90 percent or more of the training data used for current generative AI systems comes from English content.⁶ While these systems are designed to handle many languages, they perform best and respond most naturally in English.

Let me put this in context. India, South Asia and the Global Majority (my numbers and figures are specific to India, but this is not an isolated phenomenon) are undergoing a profound digital transformation. According to a report by Google and KPMG, Indian-language internet users are expected to grow to 53.6 crore. Nine out of ten new internet users in India are likely to be Indic-language users. Marathi, Bengali, Tamil and Telugu online users are expected to form almost one-third of the total Indic-language user base.⁷

Data from the Kantar-Google report in 2023 confirms this trajectory. In July 2022, close to 72.9 crore people actively used the internet in Indian languages: 34.5 crore users in urban India and 38.4 crore in rural India. One in every two people who use the internet in Indian languages employs it to consume information and news.⁸ Seventy percent of India’s internet users prefer content in their local language, and by 2026, regional-language internet users are expected to be 2.5 times larger than English users.⁷

This is what some call the vernacular internet revolution (I am not a big fan of the word). It is a shift from an English-first internet to a multilingual internet where voice-first and AI-enabled access are crucial for digital inclusion. The digital future of the Global Majority will not be built in one language. There are four pillars to this revolution: Access, Availability, Adoption, and AI Enablement.

4. What does this mean for our future?

Articles on the 14 top Indian-language Wikipedias were read a combined 14.0 crore times in June 2023.⁹ That is a fraction of the 72.9 crore people accessing the internet in Indian languages.⁹ Hindi, Bangla and Tamil made up two-thirds of the total readership. Some languages are rising fast, like Maithili and Bhojpuri, while some seem to have plateaued, like Kannada and Malayalam.⁹

This represents both a crisis and an enormous opportunity. We have the potential to define how knowledge from our languages, our histories, and our cultures is represented in the digital age. So why is this opportunity not being utilised? Our own policies present a significant barrier to documenting knowledge from oral cultures. We have all shared the frustration that Wikipedia’s citation policies are focused on written language and do not support oral citation to a great extent, which is key to knowledge transmitted orally in the majority of the world’s languages.¹⁰

An idea as lofty as the Commons can only be as strong as its community, and our community has deep structural problems. Consider the representation of women and LGBTQ+ related knowledge on our projects. The numbers have been abysmal for years. According to research data released by the Wikimedia Foundation in 2023, only 20% of new Wikipedia contributors identified as women, and only 7% were identified as gender diverse.¹¹

The history of the internet is largely a history of enclosure. The over-usage of the Commons has enabled shared resources to be treated with suspicion, privatised and converted into commodities. Big tech companies use our content extensively. It powers their search results, their voice assistants and, increasingly, their AI models. They have built businesses on the back of the labour of the Commons, and yet their commitment to sustaining and growing the Commons is minimal.⁵

A Google-Kantar survey of 8,000 Indians found that 60% barely know AI, and only 31% have tried it.¹² Google’s new Gemini campaign is built for the 470 million people across Tier 2 and Tier 3 India, where AI awareness is growing but everyday use isn’t.¹³ The new infrastructure for Indian languages is being built, but it is being built by corporations and governments, not by the Commons and free knowledge movements.

In a world where 80% of consumers are encountering suspicious information,¹⁴ the need for a trusted, verifiable, community-governed knowledge source is paramount. Our emphasis on veracity, accuracy, neutral point of view, and reliable references is not just a feature of our platform.

I want to emphasise that we do not have to do this alone. There is a broader ecosystem of open knowledge and open data movements that we can learn from and collaborate with. There is growing interest in governing digital public infrastructure (DPI) as a commons. A policy brief submitted to the G20 argues that a commons-based approach to governing DPI can help scale and localise DPI exchanges, increase transparency and accountability, and secure community engagement beyond the governments and companies involved. Important public goods as diverse as Wikipedia and Linux, constituted and governed as digital commons, offer valuable lessons in digital governance. The commons-based governance structure proposed for DPI enables cooperation by multiple actors, including the public sector, commercial providers and civil society, even across borders.¹⁵ The future of the commons is not just about grand strategy. It is also about small, impactful initiatives that build capacity, expand access, and strengthen our community.

Consider Kiwix. Kiwix is a free, open-source tool that lets anyone browse parts of the internet offline.¹⁶ Kiwix for Schools is what we need to remember. Access is not just about connectivity. It is about making knowledge available in the places and ways people can actually use it. Kiwix is a model for how the commons can reach communities that are otherwise excluded from the digital revolution.

SchoolWiki is one of the most inspiring examples of what is possible when a community takes ownership of its knowledge commons. Launched in Kerala in 2009, SchoolWiki connects around 15,000 schools for collaborative content development. It is a portal where students and teachers can contribute educational content, share best practices, and document the history and achievements of their schools. Today, SchoolWiki has over 30,000 articles and 13,000 users. It is the largest digital repository in the school education sector among Indian regional languages.^{17 18}

What makes SchoolWiki remarkable is its governance model. There are state-level, district-level, and school-level administrators.¹⁷

We can leverage existing professional networks and expertise. Librarians are natural allies in the fight for veracity and reliability. Campaigns like 1Lib1Ref provide a low-barrier entry point for new contributors.¹⁹

Provocations

I want to offer some provocations that we will keep discussing during WCI, and hopefully build projects based on these discussions, for how we might approach our work together.

• We must address our gaps. The gender gap, the lack of equity, the harassment and discrimination, the governance problems: these are not minor issues. They are existential threats to the movement.
• We need to continue to build the Knowledge Commons. This means contributing content, improving quality, and filling gaps. It means making sure that the knowledge of our region is represented in all its diversity. It means working with communities to document knowledge that would otherwise be lost.
• We must move from being a resource that is extracted to a partner that is respected. This means building impactful collaborations, not just being used and cited. It means demanding reciprocity.
• We need to expand the boundaries of the commons. This means moving beyond written, text-based knowledge. It means addressing the oral knowledge barrier and working with non-dominant, oral and endangered languages.
• We need to deepen our collaboration with other Commons movements. OpenStreetMap, Creative Commons, Open Data and the broader open movement are not competitors.
• We need to nurture curiosity, inquisitiveness and a temperament towards discernment in our society. A society that is curious, that values truth, that seeks to understand, is the society that will sustain the Commons.

One last thing

Can we build a Knowledge Commons that is future-ready, that is better prepared to address our structural problems, and that makes us active contributors in the knowledge ecosystems that will shape our future needs?

The answer depends on us. It depends on whether we are willing to confront our failures, learn from our allies, and build a knowledge commons that is truly equitable and sustainable. The future of the knowledge commons will not be built by passive consumers. It will be built by people who are curious, who care about the truth, who are willing to engage deeply and contribute their understanding. It will be built by small, impactful initiatives like Kiwix, SchoolWiki, and 1Lib1Ref. It will be built by first-time editors in Vidisha and by veteran contributors in Kerala.

The systems we build must nurture that curiosity and make that contribution easy and rewarding. This is not just a technical challenge. It is a social, cultural, and philosophical one. It is the work that will define our movement for the next decade.

The questions I have asked today are not rhetorical questions. These are the questions that will shape our future. The answers are not in this keynote. All of the answers are in this room.

Thank you, on behalf of all of us who have worked to put WCI in front of you.

References

1. Hess, C. and Ostrom, E. (eds.) (2007). Understanding Knowledge as a Commons: From Theory to Practice. MIT Press. https://mitpress.mit.edu/9780262516037/understanding-knowledge-as-a-commons/ ↩︎
2. The Nobel Prize (2009). “The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2009: Elinor Ostrom and Oliver E. Williamson.” https://www.nobelprize.org/prizes/economic-sciences/2009/summary/ ↩︎
3. Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press. (Design principles for long-enduring common-pool resource institutions, chapter 3.) ↩︎
4. Bollier, D. and Euler, J. (2025). “How Patterns of Commoning Reveal Commons as Relational Social Systems.” Panel 10.4, IASC 2025 – XX Biennial Conference of the International Association for the Study of the Commons, 18 June 2025. https://2025.iasc-commons.org/abstract_author/author-108/ . See also Bollier, D. and Helfrich, S. (2019). Free, Fair and Alive: The Insurgent Power of the Commons. New Society Publishers. ↩︎
5. Tarkowski, A. and Valdelli, I. (2026). Wikimedia, the Commons, and the New AI Knowledge Loop: A Mission for 2030. Open Future Foundation and Wikimedia CH, 21 July 2026. https://openfuture.eu/publication/wikimedia-the-commons-and-the-new-ai-knowledge-loop/ ↩︎
6. Brown, T. B. et al. (2020). “Language Models are Few-Shot Learners.” Advances in Neural Information Processing Systems 33. https://arxiv.org/abs/2005.14165 . Dataset language statistics (92.6% of training-corpus words in English): https://github.com/openai/gpt-3/blob/master/dataset_statistics/languages_by_word_count.csv ↩︎
7. KPMG and Google (2017). Indian Languages: Defining India’s Internet. April 2017. https://assets.kpmg.com/content/dam/kpmg/in/pdf/2017/04/Indian-languages-Defining-Indias-Internet.pdf ↩︎
8. Kantar and Google (2023). Indian Languages – Understanding India’s Digital News Consumer. May 2023. https://ruralindiaonline.org/en/library/resource/indian-languages—understanding-indias-digital-news-consumer/ ↩︎
9. Wikimedia Foundation. Wikistats: Total page views by project, monthly data for June 2023 for the Hindi, Bengali, Tamil, Telugu, Marathi, Malayalam, Kannada, Urdu, Gujarati, Punjabi, Odia, Assamese, Maithili and Bhojpuri Wikipedias. https://stats.wikimedia.org/#/all-wikipedia-projects/reading/total-page-views (see also the Meta-Wiki overview: https://meta.wikimedia.org/wiki/List_of_Indian_language_wiki_projects) ↩︎
10. Prabhala, A. (2011). “Research:Oral Citations – People are Knowledge.” Wikimedia Foundation, Meta-Wiki. https://meta.wikimedia.org/wiki/Research:Oral_Citations ↩︎
11. Wikimedia Foundation (2023). Community Insights 2023 Report. Meta-Wiki. https://meta.wikimedia.org/wiki/Community_Insights/Community_Insights_2023_Report ↩︎
12. Google India (2025). “75% of Indians desire a daily growth collaborator: Google-Kantar report.” Google India Blog, 25 April 2025. https://blog.google/intl/en-in/products/75-of-indians-desire-a-daily-growth-collaborator-google-kantar-report/ ; coverage: https://www.storyboard18.com/how-it-works/60-not-familiar-with-ai-only-31-have-tried-a-gen-ai-tool-google-kantar-report-63333.htm ↩︎
13. BestMediaInfo (2026). “Shah Rukh Khan helps Google Gemini take AI beyond English-speaking metros.” 10 August 2026. https://bestmediainfo.com/ad-craft/google-gemini-finds-its-bharat-voice-with-shah-rukh-khan-12247152 ↩︎
14. Adobe (2024). Future of Trust Study. April 2024. https://news.adobe.com/news/news-details/2024/media-alert-adobe-study-reveals-high-concern-over-misinformation-and-potential-to-impact-elections ↩︎
15. Avila, R., Chandrasekhar, R., Dulong de Rosnay, M. and Rens, A. (2024). Governing Digital Public Infrastructure as a Commons. T20 Brasil Policy Brief, Task Force 05 – Inclusive Digital Transformation, July 2024. https://t20brasil.org/media/documentos/arquivos/TF05_ST_02_Governing_Digital_P66ccd2697b15b.pdf (summary: https://www.global-solutions-initiative.org/publication/governing-digital-public-infrastructure-as-a-commons/) ↩︎
16. Kiwix. “Kiwix: Explore offline Wikipedia and educational content.” https://kiwix.org/en/ ↩︎
17. Kerala Infrastructure and Technology for Education (KITE). Schoolwiki. https://schoolwiki.in/ ; IT@School Project, Government of Kerala (2017). “Schoolwiki – A digital repository of 15,000 schools of Kerala.” Ministry of Education repository. https://repository.education.gov.in/wp-content/uploads/2017/07/school-wiki–a-digital-repository-of-15000-schools-of-kerala.pdf ↩︎
18. The News Minute (2022). “Helmed by teachers and students, Schoolwiki in Kerala going strong after 13 years.” 22 August 2022. https://www.thenewsminute.com/kerala/helmed-teachers-and-students-schoolwiki-kerala-going-strong-after-13-years-167065 ; Open Digest (2023). “Kerala SchoolWiki wins applause from UNESCO.” 15 September 2023. https://opendigest.in/kerala-schoolwiki-wins-applause-from-unesco/ ↩︎
19. Wikimedia Foundation. “The Wikipedia Library/1Lib1Ref.” Meta-Wiki. https://meta.wikimedia.org/wiki/The_Wikipedia_Library/1Lib1Ref ↩︎