# Jimmy Wales: ‘AI can generate answers. It still can’t earn trust’ > Source: > Published: 2026-08-16 07:30:00+00:00 *The man behind the world’s largest digital encyclopedia talks to Fortune Greece, republished with permission.* There are few digital platforms that have become as quietly indispensable as Wikipedia. With no ads, no subscriptions, and none of the noise that typically accompanies anything “big” on the internet, it has been operating for nearly a quarter of a century as the most stable knowledge infrastructure of the digital world. Behind it stands Jimmy Wales, a founder who seems almost stubbornly resistant to the logic of Silicon Valley. Today, in an environment where truth is constantly questioned, artificial intelligence is reshaping our relationship with knowledge, and trust in institutions is at historical lows, Jimmy Wales appears almost unconventional. He doesn’t make any reference to disruption, nor does he promise technological “revolutions.” Instead, he insists on something far simpler—and far more difficult: the pursuit of transparency, verification, and group responsibility. The story of Wikipedia is now well known. What matters more today is not how it began, but how it endured—and what it means to remain relevant in an ecosystem that evolves faster than ever. There is something oddly reassuring about talking to someone whose creation has become so deeply embedded in our daily lives that it’s hard to remember what the world looked like before it. It’s not just the scale of the project. It’s his commitment to principles that sound obvious, but in practice feel almost radical. At a time when information is becoming increasingly “easy,” Jimmy Wales insists that understanding remains a deeply human process—one of the most hopeful messages not only for the future of Wikipedia, but also for the future of knowledge itself. **Wikipedia has shaped how billions of people access and understand knowledge. Nearly 25 years on, how has it managed to remain sustainable without ads or paywalls?** The model has worked very well from the beginning, and we see no reason to change it. We have incredibly strong support from the public. The average donation to Wikipedia is about $10, we have millions of donors, and many of them give year after year. This model also brings important advantages. We are not dependent on advertising revenue, nor are we under pressure to chase sensational headlines to drive traffic. That allows us to remain calm and consistent – and that matters. **You’ve argued that tech companies should be paying to train their AI models on Wikipedia’s content. If Wikipedia becomes a core data source for AI, how do you ensure it remains a public good?** One of Wikipedia’s defining features is that all of its content is freely licensed, similar to open-source software. Anyone can use it, modify it, and redistribute it, for commercial or non-commercial purposes, at no cost. We don’t charge AI companies to use our data, and under our licensing model, we couldn’t. What concerns us is how that data is used. When the use of Wikipedia puts a significant burden on our infrastructure, it needs to be done in a more structured and fair way, through systems we can manage. It’s not reasonable for our millions of donors, who contribute an average of around $10, to subsidize large technology companies. They are supporting our mission. That said, things are moving in the right direction. Most major AI companies are beginning to recognize that they need to be fair to Wikipedia. At the same time, our mission is free knowledge for everyone. In that sense, it is a good thing for AI to be trained on Wikipedia data. I wouldn’t want to use an AI trained only on X. It would be a very stupid and angry AI. **Wikipedia was built on a model where users search, read, and evaluate information. Today, AI delivers ready-made answers. Does that concern you, in the sense that it might make people more passive towards information?** I use AI extensively, and in my experience, it actually makes you more active, not more passive. Instead of simply reading a travel article, for example, you can ask questions, go deeper, and explore a topic in more detail. The same applies to programming, which I do as a hobby. I’m not particularly skilled, but AI helps me understand concepts—as long as you use it properly, asking for explanations and following up with more complex questions. Of course, not everyone will use it that way. It’s a complex picture. But overall, I don’t think AI leads to passive consumption of knowledge. **So, the issue isn’t just the technology, but how we use it. What does human collaboration still do better than AI when it comes to reliable knowledge?** One of the core problems with AI today is what we call “hallucinations.” Large language models work by predicting the next word based on what came before, choosing what seems most likely. The results can be impressive. We’ve all seen AI produce responses that sound coherent and reasonable. But that doesn’t mean it actually understands reality or facts. In practice, the more obscure the topic, the higher the error rate. If you ask about someone like Taylor Swift, the answer will probably be accurate. But if you ask about a lesser-known subject, the model may start inventing things—because it wants to provide a complete and confident answer. That’s why we don’t allow AI to be used to write Wikipedia articles. It can support parts of the process, and that will likely increase in the future. But at the level of final authorship, the error rate is still too high. **How does Wikipedia protect itself from AI-generated or plausible but incorrect information?** Wikipedia operates on strict sourcing standards. Much of our time is spent discussing the reliability and quality of sources. If someone adds information without proper sourcing, it is immediately challenged. Other editors ask where it comes from, and if it proves false, the edit is quickly reverted. Repeated violations can lead to bans. Detecting AI use is difficult. What matters is the quality of the work. That’s why we discourage using AI to write articles—it often introduces information that sounds plausible but isn’t true. And that’s the real challenge. These systems don’t produce obviously absurd errors. They produce errors that sound entirely believable. If an AI claimed Taylor Swift was the first person on Mars, you’d immediately dismiss it. But if it gets a relative’s name wrong or invents a plausible album, you might not notice. That’s what makes it difficult. **What happens when a new editor adds incorrect or poorly sourced information?** Typically, the edit is reverted, and other users ask for clarification. There was a case in the German Wikipedia where someone was adding book references using ISBN numbers. At first, the errors seemed like typos, but eventually, it became clear the books didn’t exist. The user explained that they were new and had used an AI tool to generate references, without realizing it could fabricate ISBN numbers or books. That’s the danger: the output looks convincing but is entirely false. In that case, it was a good-faith mistake. The user apologized and wasn’t banned. But if they had continued, they would have been. **What are the most common mistakes Wikipedia editors make?** For newcomers, the biggest mistake is not understanding the need for neutral writing. Wikipedia is not X. The goal is a calm, balanced presentation of information. Sourcing is equally important—especially high-quality sources. As for the size of the community, it’s hard to measure precisely. Someone who makes one edit isn’t necessarily part of it. But among regular editors, there are roughly 60,000 to 80,000 globally. A smaller group of around 5,000 highly active users does most of the work. **Wikipedia operates in a highly polarized geopolitical environment. How difficult is it to defend against coordinated influence?** Some of these challenges have always existed. There will always be people trying to push an agenda, and we deal with that continuously. But the bigger issue today is the decline of local media and journalism. It’s becoming increasingly difficult to document the history of smaller communities. In many parts of the world, it’s easier to write about a city in 1975, when there was a local newspaper, than today, when there may be none. That’s a real problem, and we don’t yet have a solution. The broader decline in public discourse is a societal issue, but it doesn’t affect Wikipedia in the same way. We still rely on quality journalism and credible sources, and we’re careful about how we use them. **In a world where political and social polarization has become the new normal, do you think we can return to a more balanced public discourse?** It’s a good question. This period is certainly unusual, though I’m not sure there has ever been a truly “normal” time. A friend of mine used to say, “I’m always waiting for life to get back to normal so I can do certain things—but it never does.” He was talking about everyday life with kids, but there’s truth in that. It may feel like things were calmer a few years ago. I often point to the era of John McCain and Barack Obama—there were disagreements, but also a level of respect. I think we can return to something like that, because people want it. There’s real dissatisfaction with the current toxic environment. The question is how—and that’s exactly what my book, *The Seven Rules of Trust*, is about. **English-language Wikipedia has enormous influence over how the world understands events, people, and histories. How do you address bias, and how do you ensure that smaller languages, such as Greek, are not left on the margins?** Wikipedia operates in more than 300 languages. Some communities are small, but there are more than 100 that are particularly active and vibrant, including the Greek one. There is constant communication among users across different language editions. English often serves as a common language, but in many parts of the world, local languages remain just as critical. The idea of neutrality is discussed and developed collectively, as a global principle that runs through the entire project. At the Wikimedia Foundation, we study how this principle is applied across different linguistic environments. In some smaller editions, there is not yet a formally written neutrality policy – not because neutrality is rejected, but because it has not yet been codified. That also shows the different nuances the concept of neutrality can carry. One example I often use is the question of who invented the airplane. In the English-speaking world, the answer is almost obvious: the Wright brothers. In other countries, however, the story may be different. And in reality, the history is more complex than we tend to think. This points to something essential: very often, we carry biases without even realizing it. Wikipedia works precisely as a mechanism that brings those different perspectives to the surface, allowing for a fuller account of reality. Even in highly conflicted environments, this approach works. If you compare, for example, the Russian and Ukrainian Wikipedias, you’ll see that, despite the differences, they are surprisingly similar in their effort to explain events. And that is something we can be very proud of. **Can Wikipedia help people understand the perspective of “the other side,” even when they deeply disagree with it?** Absolutely. One of the most valuable things Wikipedia can offer is that, in times of conflict, it allows you to understand the other side’s point of view. You may still disagree, but at least you understand the reasoning behind it. I remember a striking experience in Taiwan. A young volunteer, who was accompanying me to a series of meetings, explained that he had grown up in a strongly nationalist environment, where he had been taught that people from mainland China had been brainwashed and did not understand the facts. Through his involvement with Wikipedia, he came into contact with users from mainland China. And, as he told me, although he still disagreed with them on many issues, he could now understand where their views were coming from. For me, that is a meaningful step toward understanding. You don’t have to agree with someone. It is enough to recognize them as a human being with reasons and arguments. From there, a real conversation can begin – and that is fundamental. **In your book, *** The Seven Rules of Trust***, you write that transparency is hardest precisely when an organization has something to hide. What does that mean for a CEO who needs to protect the company’s reputation while also preserving public trust?** That is exactly where the difficulty lies. Transparency is important, but it becomes truly critical when you have something to hide. When something goes wrong, the temptation to cover it up or avoid responsibility is strong. But that is a choice that rarely works in the long term. People – customers, partners, employees – can forgive a mistake. What they do not forgive easily is denial and dishonesty. The story of [Airbnb](https://fortune.com/company/airbnb/) is a good example. When a crisis of trust broke out after a property was damaged, the company’s initial response was defensive and inadequate. Very quickly, though, they realized that without trust, there is no sustainable business model. They took responsibility, admitted the mistake, and made meaningful changes. That shift proved decisive for their trajectory. It takes courage to say, “We got it wrong.” Once you choose to deny it or cover it up, the problem doesn’t shrink – it multiplies. **Do you see more fear or pride among today’s leaders when it comes time to admitting a mistake?** We definitely see it, and in an almost pathological way. At the same time, we can also see the results, because people no longer trust their leaders. In reality, admitting a mistake would be far more effective. When a leader takes responsibility, people are willing to say, “Okay, move forward and do better.” By contrast, persistent denial and excessive pride undermine credibility and, ultimately, push the public even further away. **At a time when many companies struggle to keep employees engaged and connected to their mission, Wikipedia relies on thousands of people who contribute voluntarily. What keeps this ecosystem alive?** One of the seven rules in the book is to have a clear reason for existing. In our case, that purpose is simple and very specific. Many companies also have a strong core of values. The challenge is being able to express it clearly. When that happens, decision-making becomes easier: everyone knows why the organization exists, and it avoids being pulled in random, opportunistic directions. Even in a seemingly “simple” field – say, making cardboard boxes – there can be real substance. If you define your goal as making the best possible products in an area the world genuinely needs, then you have a clear direction. And when that direction is clear, it aligns everything: decisions, priorities, and ultimately the way an organization operates. **Alternative knowledge models are now emerging alongside Wikipedia, based more on AI and less on human curation. Elon Musk, for example, argues that machines can be more objective than humans. How do you respond to that?** I don’t think that’s true. Elon argues that his model can be more neutral than Wikipedia. In practice, however, it seems to align with some of his own, rather unusual, political views. And to me, that is not neutrality. The core issue is transparency. We don’t know how the content is produced, what data the AI has been trained on, or what instructions it has been given. Without that visibility, trust becomes extremely difficult. You can use it to understand a particular view of the world. But that is not the same as an objective account of reality. **Wikipedia has strict criteria regarding who is considered “notable.” What does “notability” really mean to you?** I’ve always thought the term “notability” is not ideal. It sounds as if we are judging whether someone is important or worthy of attention. My mother is very important to me, but that doesn’t mean she should have a Wikipedia entry. In reality, what matters to us is verifiability. Are there sufficient, reliable, independent sources that allow us to write a responsible biography? If the only source is your personal blog, that is not enough. You can write anything, but is it true? Is it complete? Are key facts missing? Ultimately, that is the core criterion. Not how “important” someone appears to be, but whether the information can be documented in a reliable way. **There are people who may not be widely recognizable, but who have made meaningful contributions to science, public life, or a specific field. How does Wikipedia approach them?** Of course. If we take scientists as an example, some are widely known and there are many sources about their lives and work. Others, equally important, are recognized mainly through their scientific work, and the available sources are different in nature. In any case, if someone has made a meaningful contribution and there is sufficient, reliable information to document it, even in the form of a short biography, then they deserve to be on Wikipedia. Contribution matters, but only when it can be documented in a reliable way. **How do you imagine Wikipedia over the next 25 years, in a world where technology and the way we consume information are constantly changing?** Broadly speaking, the path remains steady. Wikipedia has existed for almost 25 years and, looking ahead to the next 25, I don’t see its core changing: it will continue to function as an encyclopedia, serving the same purpose. Of course, technology will evolve. Artificial intelligence will increasingly be used as a tool to support the work of editors, not as a substitute for human judgment. The way people search for and consume information will also change. Still, Wikipedia’s fundamental mission remains unchanged: the reliable and free dissemination of knowledge. On a personal level, I approach things with the same philosophy. I focus on what I find interesting and meaningful at any given moment, whether that has to do with Wikipedia or broader questions around trust, technology, and society. What motivates me is curiosity and the opportunity to work on ideas that have real impact. And above all, I still enjoy it. *breaks the traditional barrier between audience and newsroom. The show transforms* **Fortune Daily*** Fortune*’s trusted reporting into actionable, conversational, and entertaining insights for an emerging class of business leaders. **Watch here.**